JP7206617B2 - Color map optimization device and volume rendering device for tomographic image display - Google Patents

Description

The present invention relates to a volume rendering apparatus that generates a volume rendering image based on a group of tomographic images captured by a tomographic imaging apparatus, and a color map optimization apparatus that performs optimization processing on a color map used for volume rendering. .

Image diagnostic systems using technologies such as CT (Computer Tomography), MRI (Magnetic Resonance Imaging), and PET (Positron Emission Tomography) are widely used in the medical field and the like. In such a diagnostic imaging system, a three-dimensional image model is constructed based on a plurality of tomographic images taken by a tomographic imaging device, and is projected onto a two-dimensional plane to generate a volume rendering image, which is displayed on the display screen. The processing shown above is performed.

Generally, a tomographic image obtained by imaging using a tomographic imaging apparatus is a collection of pixels to which predetermined signal values obtained by measurement are assigned, and the image is a monochrome gradation image. Therefore, in order to generate a volume rendering image used for screen display on a display, it is necessary to use a tomographic image display color map to replace individual signal values with color values and opacity. A general color map has a color palette that associates signal values with color values, and an opacity curve that associates signal values with opacity. , the signal value given to each pixel constituting the tomographic image is replaced with the color value and the opacity.

Each medical imaging system prepares its own color map to obtain a volume rendering image suitable for viewing a particular human tissue. For example, in order to obtain a volume rendering image for observing bones, rendering processing using a color map for bone observation is performed to visualize the bones, and a volume rendering image for observing a specific organ is displayed. In order to obtain , a rendering process using a color map for observing the organ is performed to display the organ in a visualized manner.

For example, Patent Literature 1 below proposes a method in which different opacity curves are defined for each observation target, and different colors are set for portions where the signal values overlap to display them in a mutually distinguishable manner. . Further, Patent Document 2 discloses a region growing method as a method for automatically extracting an organ portion to be observed. On the other hand, Patent Document 3 proposes a method of automatically generating a preset map based on a histogram of signal values. Techniques have been proposed to provide a plurality of parameters such as .

Japanese Patent No. 4922734 Japanese Patent No. 4087517 Japanese Patent No. 5576117 Japanese Patent Publication No. 2008-518347

However, the conventional volume rendering apparatus described above has a problem that it is difficult to selectively and clearly display a desired observation target. Specifically, in a conventional volume rendering device, when a volume rendering image is displayed on a display screen, the resolution of the observation target decreases and blurring occurs, or the observation target is buried in other objects. There is a problem that it becomes difficult to see.

Accordingly, the present invention provides a volume rendering apparatus capable of generating a volume rendering image that can selectively and clearly display a desired observation target based on a group of tomographic images captured by a tomographic imaging apparatus. It is also an object of the present invention to provide a color map optimization apparatus that can be used in such a volume rendering apparatus.

(1) A first aspect of the present invention provides an apparatus for optimizing a color map used to generate a volume rendering image based on a group of tomographic images captured by a tomographic imaging apparatus,
a tomographic image input unit for inputting a plurality of tomographic images obtained by tomography of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value;
A color map that inputs a color map that includes an opacity curve that associates opacity with each signal value within a predetermined range and a color palette that associates each predetermined color value with each other as a target for optimization processing. an input unit;
a signal value histogram creation unit that creates a signal value histogram indicating the frequency of appearance of each signal value for each pixel that constitutes a plurality of tomographic images;
a field of view histogram creation unit that creates a field of view histogram by weighting the signal value histogram with an opacity curve;
an opacity curve corrector that corrects the opacity curve to create a corrected opacity curve and outputs a corrected color map containing the corrected opacity curve instead of the opacity curve;
provided,
The opacity curve corrector has at least a view histogram consideration corrector that performs correction using a view field histogram on the opacity curve.

( 2 ) A second aspect of the present invention is the tomographic image display color map optimization device according to the first aspect described above, comprising:
The field histogram consideration correcting unit weights the signal value histogram using the corrected opacity curve compared to the peak position of the field of view histogram obtained by weighting the signal value histogram using the opacity curve. The correction to the opacity curve is performed so that the peak position of the corrected field of view histogram obtained is closer to the peak position of the opacity curve.

( 3 ) A third aspect of the present invention is the apparatus for optimizing a color map for tomographic image display according to the second aspect described above, comprising:
The field-of-view-histogram-considered corrector corrects the opacity curve so that the peak position of the corrected field-of-view histogram coincides with the peak position of the opacity curve.

( 4 ) A fourth aspect of the present invention is the tomographic image display color map optimization device according to the third aspect described above, comprising:
A visual field histogram consideration correcting unit calculates the maximum value of the visual field histogram within a predetermined signal value range as VHmax
, the signal value giving the maximum value VHmax is Svh, the maximum value of the opacity curve within the predetermined signal value range is OCmax, the signal value giving the maximum value OCmax is Soc, the opacity of the opacity curve at the signal value Svh is OC(Svh), the value of the corrected field of view histogram at the signal value Soc is VHmax, and the value of the corrected field of view histogram at the signal value Svh is VHmax OC(Svh)/OCmax for the opacity curve This is for correction.

( 5 ) A fifth aspect of the present invention is the tomographic image display color map optimization apparatus according to the fourth aspect described above, comprising:
The visibility histogram consideration correcting unit assumes that the opacity of the opacity curve at the signal value S is OC(S), the value of the visibility histogram at the signal value S is VH(S), and the correction coefficient γ(S) at the signal value S is
γ(S)=(VHmax/OCmax)·(OC(S)/VH(S))
The opacity OCC (S) of the corrected opacity curve at the signal value S is defined by the following formula:
OCC(S)=γ(S)・OC(S)
It is calculated by the following formula.

(6) A sixth aspect of the present invention is the tomographic image display color map optimization device according to the first aspect described above,
The field-of-view-histogram-considering correcting unit performs weighting on the signal value histogram with the opacity curve, compared to the shape near the peak of the field of view histogram obtained by weighting the signal value histogram with the corrected opacity curve. The corrected field of view at the peak position of the opacity curve is corrected so that the shape near the peak of the corrected field of view histogram obtained by weighting is flattened. The value of the histogram matches the value of the corrected field of view histogram at the peak position of the field of view histogram , and the value between the peak position of the opacity curve and the peak position of the field of view histogram is the peak of the field of view histogram. A correction is made to the opacity curve to match the values of the corrected field histogram at the position .

( 7 ) A seventh aspect of the present invention is the apparatus for optimizing a color map for tomographic image display according to the sixth aspect described above, comprising:
A visual field histogram consideration correcting unit calculates the maximum value of the visual field histogram within a predetermined signal value range as VHmax
, the signal value giving the maximum value VHmax is Svh, the maximum value of the opacity curve within the predetermined signal value range is OCmax, and the signal value giving the maximum value OCmax is Soc, the corrected visual field at the signal value Soc The opacity curve is corrected so that both the histogram value and the corrected visual field histogram value at the signal value Svh are VHmax.

( 8 ) An eighth aspect of the present invention is the apparatus for optimizing a color map for tomographic image display according to the seventh aspect described above, comprising:
The field histogram consideration correction unit sets the opacity of the opacity curve at the signal value S to OC(S), the field histogram value at the signal value S to VH(S), and the correction coefficient γ(S) at the signal value S as
γ(S)=(VHmax/VH(S))
The opacity OCC (S) of the corrected opacity curve at the signal value S is defined by the following formula:
OCC(S)=γ(S)・OC(S)
It is calculated by the following formula.

( 9 ) A ninth aspect of the present invention is the tomographic image display color map optimizing apparatus according to the first to eighth aspects described above,
After the tomographic image input unit inputs a plurality of tomographic images as monochrome images each having a 16-bit signal value, it performs a gradation conversion process for converting these to monochrome images having an 8-bit signal value. has a function of giving a tomographic image to the signal value histogram creation unit,
In the gradation conversion process, the minimum value Dmin and the maximum value Dmax of the signal values for some or all of the input tomographic images are obtained, and a predetermined contrast adjustment coefficient β (where 0<β<0.3) is obtained. , the lower limit Lmin and the upper limit Lmax are obtained by the formulas Lmin = Dmin + (Dmax - Dmin) · β, Lmax = Dmin + (Dmax - Dmin) · (1 - β), and the lower limit Lmin
~ upper limit value Lmax is linearly transformed to 0~255, lower limit value Lmin
Signal values less than Lmax are converted to 0, and signal values exceeding the upper limit Lmax are converted to 255, thereby generating a monochrome image having 8-bit signal values.

( 10 ) A tenth aspect of the present invention is the tomographic image display color map optimizing apparatus according to the fourth , fifth , seventh or eighth aspect described above, comprising:
The visual field histogram consideration correction unit determines the maximum value VHmax within a predetermined signal value range of the visual field histogram.
In order to obtain It is designed to be a value range.

( 11 ) An eleventh aspect of the present invention is the tomographic image display color map optimizing apparatus according to the first to tenth aspects described above, comprising:
The opacity curve corrector corrects the opacity of a part or all of the opacity curve corrected by the field histogram consideration corrector to the power of a correction coefficient γ (where 0<γ<1). It is designed to further have a part

( 12 ) A twelfth aspect of the present invention is the tomographic image display color map optimizing apparatus according to the first to tenth aspects described above, comprising:
The opacity curve correction unit determines the opacity of a part or all of the opacity curve corrected by the field histogram consideration correction unit. 0<γ<1), and when the opacity is less than the threshold value T, a power correcting unit is further provided to perform correction by exponentiating the reciprocal of the correction coefficient γ.

( 13 ) According to a thirteenth aspect of the present invention, the color map optimizing apparatus for tomographic image display according to the first to twelfth aspects described above is configured by incorporating a program into a computer.

( 14 ) A fourteenth aspect of the present invention is to configure a tomographic image display volume rendering device by incorporating the tomographic image display color map optimizing device according to the first to twelfth aspects described above,
In addition to the tomographic image input section, color map input section, signal value histogram creation section, view histogram creation section, and opacity curve correction section, which are the components of the tomographic image display color map optimization device,
Signal values assigned to each pixel of multiple tomographic images input by the tomographic image input unit are converted into voxel values indicating predetermined color values and opacity using the correction color map output from the opacity curve correction unit. a voxel image creation unit that creates a voxel image consisting of a collection of voxels arranged in a three-dimensional space by assigning a voxel value to each voxel defined at each pixel position of the tomographic image;
a rendering processing unit that generates a volume rendering image showing, as a two-dimensional image, a state in which the voxel image is observed from a predetermined line-of-sight direction;
is further provided.

( 15 ) A fifteenth aspect of the present invention is the tomographic image display volume rendering device according to the fourteenth aspect described above, comprising:
further providing a color map storage unit storing a plurality of pre-created standard color maps;
A color map input unit inputs a predetermined standard color map from a color map storage unit according to an operator's selection instruction.

( 16 ) A sixteenth aspect of the present invention is the tomographic image display volume rendering device according to the fourteenth or fifteenth aspect described above, comprising:
further providing a color map creation unit for creating a new color map based on an operator's creation instruction,
The color map input unit inputs the color map created by the color map creation unit.

( 17 ) A seventeenth aspect of the present invention is the tomographic image display volume rendering device according to the sixteenth aspect described above, comprising:
A color map creation unit inputs a plurality of representative signal values, an opacity corresponding to each of these representative signal values, and a color value associated with each of a plurality of signal value intervals as an operator creation instruction, Based on this creation instruction, an opacity curve is created by interpolating the opacity for continuous signal values within a predetermined range, and a color palette is created in which each signal value interval is associated with each color value. It is the one that was made.

( 18 ) An eighteenth aspect of the present invention is the tomographic image display volume rendering device according to the seventeenth aspect described above, comprising:
A color map creation unit creates a first representative signal value, a second representative signal value, and a third representative signal value, which are discretely arranged in ascending order, and opacities respectively corresponding to these representative signal values. Input as an operator's creation instruction, regarding the signal value interval between the first representative signal value and the second representative signal value and the signal value interval between the second representative signal value and the third representative signal value linearly interpolates the opacity, performs interpolation to give an opacity corresponding to the first representative signal value for signal value intervals below the first representative signal value, and performs interpolation to give opacity corresponding to the first representative signal value for signal value intervals exceeding the third representative signal value. is to create an opacity curve by performing interpolation that gives an opacity corresponding to the third representative signal value.

( 19 ) A nineteenth aspect of the present invention is the tomographic image display volume rendering device according to the seventeenth aspect described above, comprising:
The color map creation unit inputs the first representative signal value and the second representative signal value discretely arranged in ascending order and the opacity corresponding to each of these representative signal values as an operator creation instruction. , the opacity is linearly interpolated for the signal value interval between the first representative signal value and the second representative signal value, and the opacity is linearly interpolated for the signal value interval below the first representative signal value to the first representative signal value. An opacity curve is created by performing interpolation to give corresponding opacity, and performing interpolation to give opacity corresponding to the second representative signal value for a signal value section exceeding the second representative signal value. is.

( 20 ) A twentieth aspect of the present invention is the tomographic image display volume rendering apparatus according to the fourteenth to nineteenth aspects described above,
A rendering processing unit performs smoothing processing to replace the opacity of each voxel that constitutes a voxel image with the average value of the opacity of the voxel and adjacent voxels adjacent to the voxel, and the voxel image after undergoing this smoothing processing. It is designed to generate a volume rendering image based on.

( 21 ) A twenty -first aspect of the present invention is the tomographic image display volume rendering device according to the fourteenth to twentieth aspects described above,
A rendering processing unit obtains a gradient vector indicating a three-dimensional gradient of opacity for each voxel that constitutes a voxel image, and calculates a luminance value in consideration of the direction of the light source vector pointing in a predetermined direction and the direction of the gradient vector. , a shading process is performed to correct the color value of each voxel by multiplying the luminance value, and a volume rendering image is generated based on the voxel image after the shading process.

( 22 ) A twenty -second aspect of the present invention is the tomographic image display volume rendering apparatus according to the twenty -first aspect described above, comprising:
The voxel image generators are arranged side by side in the X-axis direction at a predetermined X-axis direction pitch, in the Y-axis direction at a predetermined Y-axis direction pitch, and in the Z-axis direction at a predetermined Z-axis direction pitch. Create a voxel image consisting of a grid-like voxel aggregate,
A rendering processing unit generates an X-axis difference value indicating a difference in opacity between a pair of adjacent voxels arranged on both sides of a voxel of interest along the X-axis direction, and A Y-axis difference value indicating the difference in opacity between a pair of arranged adjacent voxels, and a Z-axis difference value indicating a difference in opacity between a pair of adjacent voxels arranged on both sides of the voxel of interest along the Z-axis direction. , and the product of the X-axis difference value multiplied by the coefficient corresponding to the X-axis direction pitch is defined as the X-axis component, and the product of the Y-axis difference value multiplied by the coefficient corresponding to the Y-axis direction pitch is defined as the Y-axis direction component. A vector whose Z-axis direction component is the product of the Z-axis difference value multiplied by a coefficient corresponding to the Z-axis direction pitch is used as the gradient vector for the voxel of interest.

( 23 ) A twenty -third aspect of the present invention is the tomographic image display volume rendering device according to the fourteenth to twenty -second aspects described above,
A voxel image creation unit creates a voxel image consisting of a voxel aggregate arranged on an XYZ three-dimensional orthogonal coordinate system,
When the rendering processing unit generates a volume rendering image on a predetermined projection plane, after performing coordinate conversion to rotate the XYZ three-dimensional orthogonal coordinate system so that the XY plane is parallel to the projection plane, projection is performed. A color value calculated based on the color value and opacity of each voxel on the virtual ray, which is projected from a projection point on the plane and irradiated with a virtual ray in a direction parallel to the Z-axis to the voxel image after coordinate transformation. is given as the pixel value of the pixel located at the projection point, a volume rendering image is generated.

( 24 ) A twenty -fourth aspect of the present invention is the tomographic image display volume rendering device according to the fourteenth to twenty -second aspects described above,
A voxel image creation unit creates a three-dimensional texture image consisting of a voxel aggregate arranged on a first XYZ three-dimensional orthogonal coordinate system,
When the rendering processing unit generates a volume rendering image on a predetermined projection plane, it defines a second XYZ three-dimensional orthogonal coordinate system in which the XY plane is parallel to the projection plane. A plurality of mapping planes parallel to the XY plane are defined on the XYZ three-dimensional orthogonal coordinate system at predetermined intervals in the Z-axis direction, and from the first XYZ three-dimensional orthogonal coordinate system to the second XYZ three-dimensional orthogonal coordinate system. 3D texture image is mapped on a plurality of mapping planes, and the color value of the projection point having XY coordinate values (x, y) on the projection plane is mapped to the farthest position from the projection point From the color value of the pixel mapped to the position of the XY coordinate value (x, y) on a certain mapping plane, the color value of the pixel mapped to the position of the XY coordinate value (x, y) on the nearest mapping plane A volume rendering image is generated by sequentially blending up to the color value in consideration of the opacity of each pixel.

( 25 ) A twenty -fifth aspect of the present invention is the tomographic image display volume rendering device according to the fourteenth to twenty -fourth aspects described above,
further providing an opacity correction magnification setting unit for setting a predetermined opacity correction magnification according to the position in the three-dimensional space,
When the voxel image creation unit assigns a voxel value to each voxel, the opacity indicated by the correction color map is multiplied by the opacity correction magnification set according to the position of the voxel. A voxel value including subsequent opacity is given to the voxel.

( 26 ) A twenty -sixth aspect of the present invention is the tomographic image display volume rendering device according to the twenty -fifth aspect described above, comprising:
The voxel image creation unit arranges the voxel planes corresponding to each tomographic image so as to be parallel to the XY plane, arranges the voxels in each voxel plane two-dimensionally, and arranges them three-dimensionally on the XYZ three-dimensional orthogonal coordinate system. create a voxel image consisting of a set of voxels,
An opacity correction magnification setting unit defines a reference penetration line that penetrates a plurality of voxel planes and is parallel to the Z-axis at a predetermined position, and for each voxel, determines the distance between the voxel and the reference penetration line in the X-axis direction and The opacity correction magnification is set according to the distance in the Y-axis direction.

( 27 ) A twenty- seventh aspect of the present invention is the tomographic image display volume rendering apparatus according to the twenty -sixth aspect described above, comprising:
An opacity correction magnification setting unit defines on each voxel plane a plurality of concentric ellipses centered at the point of intersection with the reference penetration line and having a major axis parallel to the X axis and a minor axis parallel to the Y axis. The same opacity correction magnification is set to voxels positioned on the same ellipse, and a larger opacity correction magnification is set to an ellipse closer to the center point.

( 28 ) A twenty- eighth aspect of the present invention is the tomographic image display volume rendering device according to the twenty -seventh aspect described above, comprising:
The opacity correction magnification setting unit sets an opacity correction magnification that attenuates in inverse proportion to the square of the length of the long axis or the short axis of the ellipse.

( 29 ) A twenty -ninth aspect of the present invention is the tomographic image display volume rendering apparatus according to the fourteenth to twenty-eighth aspects described above, configured by incorporating a program into a computer.

( 30 ) A thirtieth aspect of the present invention is a tomographic image display color that optimizes a color map used to generate a volume rendering image based on a group of tomographic images captured by a tomographic imaging apparatus. In the map optimization method,
A tomographic image input step in which a computer inputs a plurality of tomographic images obtained by tomography of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value;
A computer inputs a color map including an opacity curve in which opacity is associated with each signal value within a predetermined range and a color palette in which predetermined color values are associated with each other as a target for optimization processing. a color map input stage for
a signal value histogram creation step in which a computer creates a signal value histogram indicating the frequency of appearance of each signal value for each pixel constituting a plurality of tomographic images;
A visibility histogram creation step in which a computer creates a visibility histogram by weighting the signal value histogram with an opacity curve;
a correction color map creation stage in which the computer corrects the opacity curve to create a correction opacity curve and creates a correction color map containing the correction opacity curve instead of the opacity curve;
and
At the correction color map creation stage, at least the opacity curve is corrected using the field of view histogram.

( 31 ) A thirty -first aspect of the present invention is a computer-readable color map in which tomographic image display color map data is recorded using the method for optimizing a tomographic image display color map according to the above-described thirtieth aspect. In a method for manufacturing an information recording medium on which color map data for tomographic image display is recorded,
In addition to the steps constituting the above optimization method, the computer further performs a correction color map recording step of recording the correction color map data created in the correction color map creation step on an information recording medium. is.

( 32 ) A thirty -second aspect of the present invention is a volume rendering method for tomographic image display, comprising each step constituting the method for optimizing a color map for tomographic image display according to the above-described thirtieth aspect,
The computer uses the correction color map created in the correction color map creation step in the optimization method to convert the signal values assigned to each pixel of the plurality of tomographic images input in the tomographic image input step in the optimization method. , to a voxel value indicating a predetermined color value and opacity, and assigning the above voxel value to each voxel defined at each pixel position of the tomographic image, thereby creating a collection of voxels arranged in a three-dimensional space. a voxel image creation step of creating a voxel image consisting of
a rendering processing step in which a computer generates a volume rendering image showing a state in which the voxel image is observed from a predetermined viewing direction as a two-dimensional image;
is further performed.

( 33 ) A thirty -third aspect of the present invention provides a computer-readable information recording medium on which volume rendering image data is recorded, using the volume rendering method for tomographic image display according to the thirty -second aspect. In a method for manufacturing an information recording medium on which data of a volume rendering image for displaying a tomographic image are recorded,
In addition to the steps constituting the volume rendering method, the computer further performs a volume rendering image recording step of recording the data of the volume rendering image generated in the rendering processing step on an information recording medium. .

According to the volume rendering apparatus and the color map optimizing apparatus according to the present invention, volume rendering capable of selectively and clearly displaying a desired observation target based on a group of tomographic images captured by a tomographic imaging apparatus. images can be generated.

More specifically, in the color map optimization device according to the basic embodiment of the present application, a signal value histogram is created based on a group of specific tomographic images actually taken, and for this signal value histogram, A visual field histogram is created by weighting with an opacity curve according to the observation target. Then, the field histogram is used to correct the opacity curve. Therefore, the color map containing the corrected opacity curve is optimized for generating a volume rendering image based on the specific tomographic image group, and the resolution of the observation target in the obtained volume rendering image is improved. It is possible to selectively and clearly display the observation target.

Further, the volume rendering device according to the basic embodiment of the present application includes the color map optimization device described above, and can perform rendering processing using the optimized color map. Therefore, the resolution of the observation target in the generated volume rendering image can be improved, and the observation target can be selectively and clearly displayed.

Furthermore, in the volume rendering device according to the additional embodiment of the present application, a predetermined opacity correction magnification is set according to the position in the three-dimensional space, and the opacity indicated by the color map is Correction is performed by multiplying the correction magnification. Therefore, the observation target in the generated volume rendering image is displayed in a manner that is easier to observe. Thus, the observation target can be selectively and clearly displayed.

FIG. 2 is a diagram showing a specific group of tomographic images taken of a human body and a volume rendering image generated based thereon; FIG. 2 is a conceptual diagram showing a pixel configuration of a general tomographic image group; FIG. 10 is a diagram showing a specific example of a volume rendering image suitable for observing a specific human tissue; 3A and 3B are diagrams showing specific examples of a signal value histogram H and a color map C; FIG. 4 is a graph showing the relationship between an opacity curve OC and a field of view histogram VH used in the present invention; 1 is a block diagram showing the configuration of a volume rendering device 100 (including a color map optimization device 101) according to a basic embodiment of the present invention; FIG. 4 is a table showing a specific data structure of color map C; FIG. 7 is a diagram showing an example of a display screen after inputting an instruction to create a color map C in the color map creating unit 162 shown in FIG. 6; FIG. 7 is a diagram showing another example of a display screen after inputting an instruction to create a color map C in the color map creating unit 162 shown in FIG. 6; FIG. 7 is a graph showing the principle of correction processing by a fitting method performed by a field-of-view histogram-considered correction unit 141 shown in FIG. 6. FIG. 7A and 7B are diagrams for explaining mathematical formulas used in correction processing by a fitting method performed by a field-of-view-histogram-considered correction unit 141 shown in FIG. 6; FIG. FIG. 7 is a graph showing the principle of correction processing by a flattening method performed by a field-of-view histogram-considered correction unit 141 shown in FIG. 6. FIG. 7A and 7B are diagrams for explaining mathematical expressions used for correction processing by a flattening method performed by a field-of-view-histogram-considered correction unit 141 shown in FIG. 6; FIG. 7 is a graph showing the principle of uniform power correction processing performed by a power correction unit 142 shown in FIG. 6; 7A and 7B are diagrams for explaining mathematical expressions used in uniform power correction processing performed by a power correction unit 142 shown in FIG. 6; FIG. 7 is a graph showing the principle of S-shaped power correction processing performed by a power correction unit 142 shown in FIG. 6; 7A and 7B are diagrams for explaining mathematical expressions used in S-shaped power correction processing performed by a power correction unit 142 shown in FIG. 6; FIG. 7 is a conceptual diagram showing a voxel configuration of a voxel image 20 created by a voxel image creation unit 170 shown in FIG. 6; FIG. 7 is a flowchart showing a procedure of rendering processing performed by a rendering processing unit 180 shown in FIG. 6; FIG. FIG. 20 is a diagram for explaining specific contents of the smoothing process performed in the smoothing process step S2 shown in FIG. 19; FIG. 20 is a diagram for explaining specific contents of shadow calculation performed in the shadow calculation step S3 shown in FIG. 19; FIG. 20 is a diagram illustrating the basic concept of rendering processing performed in the volume rendering stage S6 (ray casting method) shown in FIG. 19; FIG. 20 is a diagram illustrating the basic concept of rendering processing performed in the volume rendering stage S8 (3D texture mapping method) shown in FIG. 19; FIG. 4 is a block diagram showing the configuration of a volume rendering device 100' (including a color map optimization device 101) according to an additional embodiment of the present invention; 25 is a diagram showing a specific example of an opacity correction magnification ξ(x, y) set by an opacity correction magnification setting unit 190 shown in FIG. 24; FIG. FIG. 25 is a diagram showing a basic procedure of gradation conversion processing performed by the gradation image input unit 110 shown in FIGS. 6 and 24; FIG. 7 is a diagram showing a specific example of a volume rendering image generated by the volume rendering device 100 shown in FIG. 6; FIG. 7 is a diagram showing another specific example of a volume rendering image generated by the volume rendering device 100 shown in FIG. 6; FIG. 25 is a diagram showing a specific example of a volume rendering image generated by the volume rendering device 100' shown in FIG. 24; FIG. FIG. 25 is a diagram showing still another specific example of a volume rendering image generated by the volume rendering devices 100 and 100' shown in FIGS. 6 and 24; FIG.

Hereinafter, the present invention will be described based on illustrated embodiments.

<Contents of embodiment>
§1. General Volume Rendering Procedures and Their Problems 1.1 Volume Rendering Procedures 1.2 Problems of Conventional Techniques and Their Factors §2. 2. Basic embodiment of the present invention 2.1 Tomographic image input unit 110
2.2 Color map input unit 150, storage unit 161, creation unit 162
2.2.1 Color map input unit 150
2.2.2 Color map storage unit 161
2.2.3 Color map creation unit 162
2.3 Signal value histogram creation unit 120
2.4 Visibility Histogram Creation Unit 130
2.5 Opacity curve correction unit 140
2.6 Voxel Image Creation Unit 170 and Rendering Processing Unit 180
§3. 3. Optimization of color map 3.1 Visibility histogram consideration correction unit 141
3.1.1 Fitting method
3.1.2 Flattening method 3.2 Power correction unit 142
3.2.1 Uniform exponentiation correction
3.2.2 S-curve exponentiation correction §4. 4. Creation of Volume Rendering Image 4.1 Voxel Image Creation Unit 170
4.2 Rendering Processing Unit 180
4.2.1 Smoothing stage
4.2.2 Shadow calculation stage
4.2.3 Volume Rendering Stage §5. ADDITIONAL EMBODIMENTS OF THE PRESENT INVENTION 5.1 Additions to the Basic Embodiment 5.2 Independent Embodiments §6. Gradation conversion processing §7. Understanding as a method invention §8. Display example of volume rendering image

<<< §1. General volume rendering procedure and its problems ＞＞＞
First, we will briefly explain the procedure of conventional general volume rendering and the problems to be solved.

<1.1 Volume rendering procedure>
FIG. 1 is a diagram showing a specific tomographic image group photographed for the human body and a volume rendering image generated based thereon. FIG. 1(a) shows a group of tomographic images 10 of chest CT that were actually taken side by side, and is composed of a total of 370 tomographic images. FIG. 1(b) shows one tomographic image 11 (the image indicated by the thick frame in FIG. 1(a)) extracted from these 370 images.

A tomographic image captured by a tomographic imaging apparatus such as a CT is a collection of pixels to which a predetermined signal value (for example, X-ray transmittance) obtained by measurement is assigned. It becomes a gradation image. In this example, the tomographic image 11 is composed of a set of pixels arranged in a matrix of 512 rows and 512 columns, and each pixel is given a 16-bit signal value. The volume rendering apparatus plays a role of generating a volume rendering image 30 as shown in FIG. 1(c) based on such tomographic image group 10. FIG. Although the volume rendering image 30 shown in FIG. 1(c) is shown as a monochrome image due to the technical specifications of the application drawing, it is normally displayed as a color image colored with various colors.

FIG. 2 is a conceptual diagram showing the pixel configuration of a general tomographic image group 10. As shown in FIG. In the present application, for the sake of convenience, an XYZ three-dimensional orthogonal coordinate system as shown is defined, and the pixel configuration of the tomographic image group 10 is explained. The illustrated tomographic image group 10 is composed of a large number of tomographic images parallel to the XY plane. 2 shows only some of the tomographic images 11, . . . , 16, 17, 18, but in the case of the example shown in FIG. They are stacked in the axial direction at predetermined intervals (intervals between images taken by the tomographic imaging apparatus).

Here, let Sx be the size (number of pixels) of the tomographic image group 10 in the X-axis direction, Sy be the size (number of pixels) of the Y-axis direction, and Sz be the size (number of layers of the tomographic images) of the Z-axis direction. In the example shown in FIG. 2, Sx=6, Sy=6, and Sz=8 for convenience of explanation, but in the case of the example shown in FIG. 512, Sz=370. Hereinafter, a specific pixel in each of the tomographic images 11 to 18 is represented as pixel S (x, y, z) using coordinate values (x, y, z) indicating the position (eg, center point) of the pixel. to decide.

As described above, each pixel S (x, y, z) is given a predetermined signal value S (16-bit data in the example shown here). The three-dimensional position of each pixel S (x, y, z) corresponds to the three-dimensional position of the human body to be imaged, and the signal value S given to the pixel S (x, y, z) is Measured values (e.g. X-ray transmittance) are shown for minute parts placed at corresponding positions on the human body. Therefore, by replacing the signal value S of each pixel with a predetermined color value and opacity, a three-dimensional An image model (voxel image) can be constructed. A volume rendering image 30 as shown in FIG. 1(c) corresponds to a projection image obtained by observing this three-dimensional image model from a predetermined viewing direction.

However, in a medical diagnostic imaging system, a volume rendering image suitable for observation of a specific human tissue is created according to the diagnosis target. FIG. 3 is a diagram showing a specific example of such a volume rendering image. FIG. 3(a) shows an example of a volume rendering image 31 suitable for lung observation. The hatched portion in the drawing is the lung portion to be observed, and the blank portion is the surrounding portion. Similarly, FIG. 3(b) shows an example of a volume rendered image 32 suitable for viewing the heart. The hatched portion in the drawing is the heart portion to be observed, and the blank portion is the surrounding portion.

Of course, FIG. 3 shows a deformed image for convenience of explanation. ) are not clearly distinguished from each other, but in the volume rendering image 31 shown in FIG. In the image 32, the portion of the heart is displayed more emphasized.

Thus, a unique color map is used for each individual viewing object to create a volume rendering image suitable for viewing that particular object. Even volume rendering images created based on exactly the same tomographic image group 10 will be different images if different color maps are used. A volume rendering image 31 shown in FIG. 3(a) is an image created using a color map suitable for observing the lungs, and a volume rendering image 32 shown in FIG. 3(b) is an image suitable for observing the heart. It is an image created using a color map.

Next, the configuration of a specific color map and its role will be explained. 4A and 4B are diagrams showing specific examples of the signal value histogram H and the color map C. FIG. The signal value histogram H, as shown in FIG. 4(a), is a histogram showing the appearance frequency h of the signal value of each pixel constituting the tomographic image group. For example, the hierarchical image group 10 shown in FIG. 2 has multiple hierarchical images, and multiple pixels are arranged in each hierarchical image. A predetermined signal value S is assigned to each of these large numbers of pixels. The signal value histogram H indicates the appearance frequency of each signal value for these many pixels. As a matter of course, the signal value histogram H differs for each subject (patient) to be imaged, and even for the same subject, it varies depending on different imaging conditions.

In the example shown here, each pixel is given a 16-bit signal value (signed value within the dynamic range of -32768 to +32767), so the signal value histogram H shown in FIG. The minimum value Smin of the horizontal axis (signal value S) of is -32768, and the maximum value Smax is +32767. The vertical axis is the appearance frequency h, which indicates the total number of pixels to which each signal value S is assigned. In general, the signal values S of pixels included in each tomographic image are distributed over a wide range. , the distribution range of the signal value S is limited.

Therefore, in order to create a volume rendering image suitable for observation of a desired object, a higher opacity α is set for pixels having signal values corresponding to the distribution range of the object, and other pixels are A low opacity α may be set for pixels having signal values corresponding to the range, and volume rendering may be performed in consideration of this opacity α. For pixels having signal values corresponding to the distribution range of the object, if a predetermined color value is set and volume rendering is performed, the object can be displayed in a color corresponding to the color value. can. The colormap C serves to set this opacity α and color values.

FIG. 4(b) is a diagram showing an example of such a color map C. As shown in FIG. As shown in the figure, this color map C has an opacity curve OC (symbol OC stands for Opacity Curve) that sets the opacity α for each signal value S, and color values (R, G, B) for each signal value S and a color palette CP (the symbol CP is an abbreviation for Color Pallet).

Since the opacity curve OC can be defined as a graph with the horizontal axis representing the signal value S and the vertical axis representing the opacity α, the opacity curve OC is shown as such a graph in the present application. do. In the illustrated example, the opacity curve OC is defined as a graph having a triangular peak near the center. When handling a tomographic image in which a 16-bit signal value S is assigned to each pixel, the minimum value Smin of the signal value S on the horizontal axis of the opacity curve OC is -32768, and the maximum value Smax is +32767. The opacity α on the vertical axis is set to a value within the range of 0≦α≦1.

On the other hand, the color palette CP divides the signal value S shown as the horizontal axis in the figure into a plurality of intervals, and associates predetermined color values (R, G, B) with each signal value interval. be. Here, the color value R indicates a red component, the color value G indicates a green component, and the color value B indicates a blue component, and each color component is indicated by, for example, 8-bit (0 to 255) data. In the illustrated example, the horizontal axis is divided into three signal value intervals I, II, and III. The portion of Sb is set to the signal value section II, and the portion of Sb<S≤Smax is set to the signal value section III. Color values (R1, B1, G1) are set for the signal value interval I, color values (R2, B2, G2) are set for the signal value interval II, and color values (R3 , B3, G3) are set.

If such a color map C is used, the signal value S given to each pixel S (x, y, z) constituting the tomographic image group 10 shown in FIG. , G, B, α). For example, when the color map C shown in FIG. 4(b) is used, the opacity α1 is associated with the predetermined signal value S1 by the opacity curve OC, and the color values (R2, G2, B2) are associated. Therefore, a combination of color value and opacity (R2, G2, B2, α1) can be assigned to a pixel to which signal value S1 is assigned. Similarly, a combination of color value and opacity (R1, G1, B1, α2) can be assigned to a pixel assigned signal value S2.

In the present application, a combination of color value and opacity (R, G, B, α) corresponding to the signal value S assigned to each pixel on the tomographic image to which the signal value S is assigned is called a voxel. , a three-dimensional image obtained by arranging such voxels in a three-dimensional space is called a voxel image. If such a voxel image is obtained, a two-dimensional projection image when the voxel image is observed from the above-mentioned line-of-sight direction can be calculated by performing rendering processing by setting a predetermined line-of-sight direction. The two-dimensional projection image thus obtained is a volume rendering image 30 as illustrated in FIG. 1(c).

In the case of the color map C shown in FIG. 4(b), the opacity curve OC is set so that the opacity α in the signal value interval II is high, so that the signal value S within the range of Sa≦S≦Sb is A voxel corresponding to a given pixel (a portion of human tissue or the like exhibiting such a signal characteristic) will be given a high opacity α, and the pixel on the volume rendering image 30 will have a color value of It is prominently displayed as the region with (R2, G2, B2). On the other hand, voxels corresponding to pixels to which a signal value S belonging to signal value intervals I and III is assigned (portions of human tissue exhibiting such signal characteristics) have a low opacity α (in other words, , high transparency). The pixel is displayed as an area with color values (R1, G1, B1) or (R3, G3, B3) on the volume rendering image 30, but it is not so noticeable due to its high transparency. .

As a result, if volume rendering is performed using the color map C shown in FIG. is obtained, and an image display suitable for observing the specific tissue portion can be performed. Also, the particular tissue portion can be represented by a particular color value (R2, G2, B2). As described above, the characteristics of the volume rendering image 30 displayed on the display screen greatly differ depending on what kind of color map C is used when volume rendering is performed. Selecting a suitable colormap C for volume rendering is very important.

The form of the color map C shown in FIG. 4(b) is a typical example, and the form of the color map C used in the present invention is not limited to the illustrated example. Also, terms such as "color map" and "color palette" are not necessarily technical terms used in a unique sense. In some cases, the "color palette CP" of is called a "color map". However, in the present application, the correspondence between the signal value S assigned to each pixel of the tomographic image and the opacity α is called an “opacity curve OC”, and the signal value S is displayed on the display screen. A mapping of actual color values is called a "color palette CP", and a complex of the "opacity curve OC" and the "color palette CP" is called a "color map C".

<1.2 Problems and Factors of Conventional Technology>
As already mentioned, it is difficult to selectively and clearly display a desired observation target in a conventional general volume rendering apparatus that creates a volume rendering image based on the basic procedure described above. There's a problem. Specifically, when a volume rendering image created by a conventional volume rendering device is actually output to a display screen or a print screen, the resolution of the observed object decreases and blurring occurs, or the observed object becomes blurred. A problem arises in that the object is buried in the object and becomes difficult to see. The inventor of the present application believes that the following factors are related to the reason why such problems occur.

<Factor 1: Deviation of Visibility Histogram VH with respect to Opacity Curve OC>
As described above, a color map C as shown in FIG. 4B is used for general volume rendering processing, and based on the opacity curve OC included in this color map C, each voxel is given a predetermined Opacity α is given. Therefore, in the opacity curve OC, pixels with signal values S for which high opacity α is set are more conspicuous on the volume rendering image than pixels with signal values S for which low opacity α is set. Objects having signal values S within a predetermined range are selectively displayed.

Suppose now that the signal value histogram H indicating the appearance frequency h of the signal value S for each pixel constituting a certain tomographic image group 10 is, for example, as shown in FIG. 4(a). Consider the product of the signal value histogram H shown in FIG. 4(a) and the opacity curve OC shown in FIG. 4(b). In both the signal value histogram H and the opacity curve OC, the horizontal axis is the signal value S in the range from the minimum value Smin to the maximum value Smax. If α is obtained, the product H×OC of the signal value histogram H and the opacity curve OC can be defined.

Since the opacity α is, so to speak, a parameter indicating the degree of display of each voxel on the volume rendering image (for example, a voxel to which α=0 is given becomes a completely transparent voxel, so it does not appear on the volume rendering image at all. not displayed), the product H×OC can be said to be a histogram obtained by weighting the signal value histogram H according to the degree of display. In other words, the product H×OC is an index indicating the degree to which the pixels having the respective signal values S contribute to the image within the field of view of the observer when the volume rendering image is observed. Therefore, in the present application, the product H×OC of the signal value histogram H and the opacity curve OC is called a visibility histogram VH (the symbol VH stands for Visibility Histogram).

The shape of the field of view histogram VH depends on both the shape of the signal value histogram H and the shape of the opacity curve OP. Therefore, generally, the position and shape of the field of view histogram VH do not match the position and shape of the opacity curve OP. FIG. 5 is a graph showing the relationship between the opacity curve OC and the field of view histogram VH. Here, the opacity curve OC shown by the solid line in FIG. 5(a) is the same as the opacity curve OC shown in FIG. 4(a) shows the product H*OC of the signal value histogram H shown in FIG. 4(b) and the opacity curve OC shown in FIG. 4(b).

In FIG. 5, the vertical axis of the opacity curve OC has a range of 0≦α≦1 as shown on the left side of the graph, and the vertical axis of the field of view histogram VH has an arbitrary scale (total number of pixels) as shown on the right side of the graph. ). The shape of the field of view histogram VH shown in FIG. 5(a) is the product of the signal value histogram H shown in FIG. 4(a) and the opacity curve OC shown in FIG. 4(b). It is not exactly shown and is distorted for convenience of explanation.

Normally, the opacity curve OC prepared in the color map C has a peak at a predetermined position and forms a mountain-shaped protuberance over a predetermined signal value range. Therefore, the field of view histogram VH also forms a mountain-shaped protuberance with a peak at a predetermined position. In general, the peak position of the opacity curve OC (Soc in the illustrated example) and the peak position of the field of view histogram VH (Svh in the illustrated example) do not match, and there is a slight difference between the two. "deviation" occurs. This is because the opacity curve OC shown in FIG. It is natural to consider that the histogram VH is obtained by taking into consideration the signal value histogram H of the tomographic image group of a specific individual patient.

Thus, even if the opacity curve OC showing the standard opacity α distribution is commonly used for a group of tomographic images of many patients, the signal value histogram H differs for each individual patient. Histogram VH is also different for each individual patient. The degree of "deviation" of the position of the field of view histogram VH with respect to the opacity curve OC also differs for each individual patient. The inventor of the present application believes that such a positional "deviation" between the opacity curve OC and the field of view histogram VH causes a decrease in the resolution of the observation target, and is one of the causes of blurring.

For example, in order to display the desired observation object, as shown in FIG. If a voxel corresponding to a mountain whose peak position is the signal value Svh is displayed, voxels outside the observation target are mixed and displayed, and as a result, the resolution of the observation target is reduced. It is inferred that this will decrease and blur will occur. For example, Shengzhou Luo's Ph.D. Based on Visibility and Saliency", Dissertation for Degree of PhD., University of Dublin, Trinity College, Sep. 2016.

According to such a theory, for example, as in the example shown in FIG. If the graph can be obtained, the positional "deviation" between the horizontal axes of the two disappears, preventing deterioration of the resolution of the observed object and eliminating blurring. In fact, by correcting the peak position Svh of the field of view histogram VH closer to the peak position Soc of the opacity curve OC by a method described later, the reduction in resolution of the observation target on the volume rendering image is suppressed, and the observation target is made clearer. It is now possible to display on

A specific method for eliminating the "bokeh caused by deviation of the field of view histogram VH from the opacity curve OC" will be described in detail in §3.1 "field of view histogram consideration correction unit 141".

<Factor 2: Weber-Fechner Law>
Generally, it is known as the Weber-Fechner law that the sensitivity of human sensory organs has a logarithmic characteristic. For example, in the human visual system, the amount of sensation that a person perceives is proportional to the logarithm of the stimulus intensity, which is determined based on a physical quantity such as opacity. Therefore, the inventors of the present application adopted a curve having a linear characteristic (triangular mountain shape) as shown in FIG. , I think that it may be one of the causes of blurring in the outline of the observation target.

According to such a theory, in order to remove blurring factors based on the Weber-Fechner law, it is necessary to adopt a curve that considers the logarithmic characteristics of the sensitivity of the human sensory organs as the opacity curve OC. In fact, by correcting the shape of the opacity curve OC to a shape that considers the logarithmic characteristics by a method described later, it is possible to suppress the deterioration of the resolution of the observation target on the volume rendering image and display the observation target more clearly. became possible.

A specific method for eliminating the "bokeh based on the Weber-Fechner's law" will be described in detail in §3.2 "Power Correction Unit 142".

<Factor 3: Mixture of Different Objects with Proximity Signal Values>
As described above, the distribution range of the signal values of pixels included in a tomographic image of the human body is limited for each tissue such as internal organs, bones, skin, muscle, and fat. Therefore, by preparing an opacity curve OC having a raised portion in its own signal value distribution range for a specific tissue to be a desired observation target, a volume rendering image suitable for observation of the desired observation target can be created. be able to.

However, in reality, if there is another body with a close distribution range of signal values in another tissue different from the observation target, or in body substances such as air and water, on the volume rendering image, Together with the tissue to be observed, the separate body is displayed in a mixed state. In addition, when performing imaging with a tomographic imaging system, a bed on which the patient lies and a belt to fix the patient are usually used. These materials will also be displayed in a mixed form in the observation object if they are included.

In this way, when there is a separate object exhibiting a signal value similar to the signal value exhibited by the observation object, the object is displayed in a mixed state with the observation object, resulting in an obscure observation object. put away. Concrete techniques for solving the blurring of the observation target due to the mixture of different objects having proximity signal values will be described in detail in Section 5 "Additional Embodiments of the Present Invention".

<<< §2. Basic embodiment of the present invention >>>
Next, a basic embodiment of the invention will be described. FIG. 6 is a block diagram showing the configuration of the volume rendering device 100 according to the basic embodiment of the invention. This volume rendering apparatus 100 is a volume rendering image 30 (for example, FIG. 1 ( c) It is a device that has the function of creating A volume rendering image 30 created by the volume rendering device 100 is provided to the display device 60 and displayed on the display screen. Of course, it is also possible to output the volume rendering image 30 with a printer, if necessary.

A volume rendering apparatus 100 according to this basic embodiment is composed of an assembly of blocks surrounded by a solid line frame in the drawing. A portion of the color map optimizer 101 is shown circled. The color map optimization device 101 is a device that performs optimization processing on the color map C used for generating the volume rendering image 30 based on the tomographic image group 10 captured by the tomographic imaging device 50. It has the function of inputting the color map C to be optimized and outputting the corrected color map CC subjected to the optimization processing. Here, “optimization” means correcting the target color map C to be suitable for the group of tomographic images 10 given from the tomographic imaging apparatus 50 .

As shown in the figure, the color map optimization device 101 includes a hierarchical image input unit 110 , a signal value histogram creation unit 120 , a field of view histogram creation unit 130 , an opacity curve correction unit 140 and a color map input unit 150 .

On the other hand, the volume rendering device 100 is a device that includes the color map optimization device 101 described above. Specifically, the volume rendering device 100 shown in the drawing further includes a color map storage unit 161, a color map creation unit 162, a voxel image creation unit 170, and a rendering processing unit 180 in addition to the color map optimization device 101. Consists of The color map optimization device 101 takes in the color map C stored in the color map storage unit 161 or the color map C created by the color map creation unit 162, and optimizes the color map C that has been taken in. Correction is performed, and a corrected color map CC obtained after correction is output.

Note that the color map optimization device 101 and the volume rendering device 100 including this shown in FIG. 6 are actually configured by installing a predetermined program in a computer, and are shown as individual blocks in the figure. The functions of each component in the system are realized by a combination of computer hardware and software. Specific functions of each component will be described in order below.

<2.1 Tomographic Image Input Unit 110>
The tomographic image input unit 110 is a component that plays a role of inputting a plurality of tomographic images obtained by tomography of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value. As already explained, the group of tomographic images 10 captured by the tomographic imaging apparatus 50 includes, for example, a large number of pixels S (x, y, z) defined on a three-dimensional coordinate system, as shown in FIG. A signal value S is assigned to each pixel.

The tomographic image group 10 input by the tomographic image input unit 110 is given to the signal value histogram creating unit 120 and the voxel image creating unit 170 . The tomographic image group 10 given to the signal value histogram creating unit 120 is used as a material for performing the optimization processing of the color map C, and the tomographic image group 10 given to the voxel image creating unit 170 is used as the voxel image 20 used to create

A general tomographic imaging apparatus 50 currently in use outputs a tomographic image composed of a set of pixels having a signal value S of 16 bits. In this case, the tomographic image input unit 110 inputs a plurality of tomographic images as monochrome images each having a 16-bit signal value S. Thereafter, each process in the volume rendering apparatus 100 is performed using this 16-bit signal value S. It will be performed on the basis of the tomographic image with signal value S.

Note that the tomographic image input unit 110 may be provided with a gradation conversion processing function for reducing the gradation of the input tomographic image, if necessary. For example, if a function for converting a 16-bit signal value to an 8-bit signal value is provided, subsequent processing can be executed using pixels having 8-bit signal values, reducing the processing load. greatly reduced. Such gradation conversion processing will be described in Section 6 “Gradation Conversion Processing”.

<2.2 Color Map Input Unit 150, Storage Unit 161, Creation Unit 162>
Next, the color map input unit 150, the color map storage unit 161, and the color map creation unit 162 will be explained.

<2.2.1 Color map input unit 150>
The color map input unit 150 is a component that inputs the color map C stored in the color map storage unit 161 or the color map C created by the color map creation unit 162 as an optimization target.

The volume rendering apparatus 100 shown here is provided with both a color map storage unit 161 and a color map creation unit 162, and the color map input unit 220 can input the color map C from either of them. Of course, the color map storage unit 161 and the color map creation unit 162 do not necessarily have to be provided with both, and only one of them may be provided. Further, if the color map input unit 150 is provided with a function of inputting the color map C from an external device, it is not always necessary to provide the color map storage unit 161 and the color map creation unit 162 inside the volume rendering apparatus 100. .

As described in Section 1, the color map C referred to in the present application includes an opacity curve OC in which opacity α is associated with each signal value S within a predetermined range, and predetermined color values (R, G, B ) are associated with the color palette CP. A specific example of the color map C is as illustrated in FIG. 4(b).

In the example shown in FIG. 4(b), the opacity α and the color values (R, G, B) are associated with each other over the entire range of signal values Smin to Smax. There is no need to associate opacity α with color values (R, G, B) for the entire range. For example, regarding the low-region signal value near the minimum value Smin and the high-region signal value near the maximum value Smax, if an operation is adopted in which these signal values are not used in actual rendering processing, even these signal values have an opacity α and a color value (R , G, B) need not be associated. In the case of the example shown here, since the 16-bit signal value S is used as it is in the optimization processing, Smin =-32768 and Smax =+32767. When thinning to 8 bits for use, it is sufficient to set Smin=0 and Smax=255.

In FIG. 4(b), the color map C is shown in the form of a graph for convenience of explanation. Any digital data that associates the values of (R, G, B, α) may be used. The example shown in FIG. 7 associates the color values R, G, and B with the opacity α for each 16-bit signal value S (−32768 to +32767). , G, and B are represented by 8-bit data from 0 to 255, and the opacity α is represented as a numerical value within the range of 0≦α≦1 (actually, α is also data from 0 to 255). Here, the portion to which the color values R, G, and B are associated corresponds to the color palette CP, and the portion to which the opacity α is associated corresponds to the opacity curve OP.

Thus, the color map C input by the color map input unit 150 is actually digital data as illustrated in FIG. , in the form of a graph with the signal value S on the horizontal axis. That is, the opacity curve OC is indicated by lines forming a graph with the opacity α on the vertical axis, and the color palette CP is the color values (R, G, B).

In FIG. 6, the symbol "C (OC & CP)" attached to the arrow pointing downward from the block of the color map input unit 150 indicates that the color map C including the opacity curve OC and the color palette CP is the color map input unit 150. indicates that the output is from As shown, the opacity curve OC is provided to the field of view histogram creation unit 130, and the color map C including the opacity curve OC and the color palette CP is provided to the opacity curve correction unit 140.

<2.2.2 Color map storage unit 161>
The color map storage unit 161 is a component that stores a plurality of standard color maps C created in advance. As already explained, the substance of a set of color maps C is digital data as illustrated in FIG. A plurality of sets of such digital data are prepared in the color map storage unit 161 .

As described in .sctn.1, the color map C is used to create a volume rendering image 30 suitable for observation of a specific object based on the group of tomographic images 10. FIG. By using a specific color map C, it is possible to selectively display the tissue showing the signal value S in the mountain-like section of the opacity curve OC with the color specified by the color palette CP. The color map storage unit 161 stores a plurality of standard colors such as a standard color map C1 for lung observation, a standard color map C2 for heart observation, a standard color map C3 for bone observation, and so on. map is stored. These multiple standard color maps C1, C2, C3, .

The color map input unit 150 executes a process of inputting a predetermined standard color map C from the color map storage unit 161 according to the operator's selection instruction. For example, when the operator wants to create a volume rendering image 30 suitable for lung observation and display it on the display device 60, the operator inputs the standard color for lung observation to the color map input unit 150. A selection instruction to select the map C1 may be given.

However, since the various standard color maps C prepared in this manner contain an opacity curve OC indicating a standard distribution of opacity α suitable for observing a specific observation target (for example, lungs), they are not necessarily , based on the tomographic image group 10 of a specific patient, the color map is not optimal for displaying a specific observation target (for example, lungs) of the patient on the display device 60 . The color map optimization device 101 plays a role of optimizing such a standard color map C and outputting it as a corrected color map CC.

<2.2.3 Color map creation unit 162>
On the other hand, the color map creation unit 162 is a component having a function of creating a new color map C based on the operator's creation instruction. The color map input unit 150 also has a function of inputting the color map C newly created by the color map creation unit 162 in this way.

The format of the creation instruction to the color map creation unit 162 and the algorithm for creating the color map C based on this creation instruction may be arbitrary. An embodiment employing a form of instruction and a specific algorithm for creating the color map C based on the creation instruction given in the form will be exemplified.

In the embodiment shown here, the color map generation unit 162 generates a plurality of representative signal values, opacities respectively corresponding to these representative signal values, and color values respectively corresponding to a plurality of signal value intervals. An opacity curve OC is created by interpolating the opacity of continuous signal values within a predetermined range based on the operator's creation instruction, and the color value is associated with the signal value interval. The basic concept of this method is to create a color palette CP attached to each color. Two examples based on this basic concept are shown below.

In the first embodiment, the operator instructs the generation of the opacity curve OC by specifying three points. FIG. 8 is a diagram showing an example of a display screen after such creation instruction input work. The graph shown in FIG. 8 is based on the graph shown in FIG. 4(b), and the opacity curve OC is displayed as a line graph with the signal value S on the horizontal axis and the opacity α on the vertical axis. . On the other hand, the color palette CP is displayed by dividing the horizontal axis indicating the signal value S into a plurality of sections and associating predetermined color values (R, G, B) with each signal value section.

The drawing shows a state in which the color map C has already been created according to the operator's creation instruction. Before the operator inputs a creation instruction, an empty two-dimensional coordinate system in which only the horizontal axis (signal value S) and vertical axis (opacity α) are set is displayed on the display screen. An instruction to create the opacity curve OC may be given by specifying the positions of three specified points on this empty two-dimensional coordinate system. Here, it is assumed that the positions of three specified points Q1, Q2, and Q3 are input as shown in the figure. The position input of each point can be performed, for example, by a click operation using a pointing device such as a mouse on an empty two-dimensional coordinate system displayed on the display screen.

As described above, the horizontal axis of the empty two-dimensional coordinate system indicates the signal value S, and the vertical axis indicates the opacity α. is an input that specifies the horizontal axis coordinate s and the vertical axis coordinate α of the specified point Q. Therefore, in the illustrated example, the input of the designated point Q1 determines the coordinate values (s1, α1), the input of the designated point Q2 determines the coordinate values (s2, α2), and the input of the designated point Q3 determines the coordinate values (s3 , α3) are determined. The illustrated opacity curve OC is automatically created based on these coordinate values.

Here, if the coordinate values s1, s2 and s3 are respectively called a first representative signal value s1, a second representative signal value s2 and a third representative signal value s3, the first embodiment The color map creation unit 162 according to , the first representative signal value s1, the second representative signal value s2, and the third representative signal value s3 discretely arranged in ascending order, and corresponding to each representative signal value It has a function of inputting the opacity α1, α2, α3 to be used as an operator's creation instruction.

Then, the color map creating section 162 creates a signal value interval between the first representative signal value s1 and the second representative signal value s2 and between the second representative signal value s2 and the third representative signal value s3. For the signal value interval between, the opacity α is linearly interpolated. In FIG. 8, the line segment between designated points Q1 and Q2 corresponds to a graph created by such linear interpolation. Further, the color map creating unit 162 performs interpolation to give opacity α1 corresponding to the first representative signal value s1 for the signal value interval I less than the first representative signal value s1. As a result, for the signal value interval I, a horizontal graph showing opacity α1 is drawn. Similarly, for the signal value section III exceeding the third representative signal value s3, interpolation is performed to give opacity α3 corresponding to the third representative signal value s3. As a result, a horizontal graph showing opacity α3 is drawn for the signal value interval III.

Thus, an opacity curve OC having triangular ridges as shown is created. Here, the specified points Q1 and Q3 indicate the boundaries of the three signal value intervals I, II and III, and the specified point Q2 indicates the peak position of the opacity curve OC. Therefore, three signal value intervals I, II, and III are demarcated by performing the above inputs.

Subsequently, the color map generator 162 requests the operator to input an instruction to determine a predetermined color value (a combination of R, G, and B) for each of these signal value intervals I, II, and III. In response to this request, the operator designates color values suitable for each signal value interval. Specifically, in this case, pixels belonging to the signal value interval II are to be observed. For signal value section III, color values (R2, G2, B2) suitable for the observation object should be designated, and for signal value section III, color values (R3, G3, B3) suitable for the background may be designated.

Since such a specific manner of inputting color values is a well-known technique, detailed description thereof will be omitted here. The information indicating the color values of the respective signal value intervals I, II, and III thus inputted is the color palette CP, and the color map C is constructed by the opacity curve OC and the color palette CP.

On the other hand, in the second embodiment, the operator instructs to create the opacity curve OC by specifying two points. 9(a) and 9(b) are both diagrams showing an example of the display screen after such creation instruction input work, and the state in which the color map C has already been created according to the operator's creation instruction. is shown. The second embodiment is the same as the first embodiment in that the operator gives a creation instruction on an empty two-dimensional coordinate system. s2, α2) are specified. That is, if the coordinate values s1 and s2 are called a first representative signal value s1 and a second representative signal value s2, respectively, the color map creation unit 162 according to the second embodiment can It has a function of inputting the first representative signal value s1 and the second representative signal value s2 arranged in a linear fashion and the opacities α1 and α2 respectively corresponding to these representative signal values as an operator's creation instruction.

Then, this color map creation unit 162 linearly interpolates the opacity for the signal value interval between the first representative signal value s1 and the second representative signal value s2, and For the signal value interval, interpolation is performed to give the opacity α1 corresponding to the first representative signal value s1, and for the signal value interval exceeding the second representative signal value s2, the opacity corresponding to the second representative signal value s2 is performed. Create an opacity curve OC by performing interpolation that gives α2.

In the case of the second embodiment, the form of the graph of the opacity curve OC differs depending on the magnitude relation of the opacities of the specified points Q1 and Q2. Specifically, as shown in FIG. 9(a), when α1<α2, the section below the first representative signal value s1 is defined as the signal value section I, and the section above the first representative signal value s1 is defined as the signal value section II. , the signal value interval I becomes a graph corresponding to the flat land portion, and the signal value interval II becomes a graph corresponding to the slope portion and the highland portion. In this case, pixels belonging to the signal value interval II are to be observed. Therefore, for the signal value interval I, color values (R1, G1, B1) suitable for the background are specified, and for the signal value interval II, the observation target is A color value (R2, G2, B2) suitable for .

On the other hand, if α1>α2, the signal value interval I is the interval below the second representative signal value s2, and the signal value interval II is the interval exceeding the second representative signal value s2. , and the signal value section II becomes a graph corresponding to the flat part. In this case, pixels belonging to the signal value interval I are to be observed. Therefore, for the signal value interval I, color values (R1, G1, B1) suitable for the observation object are specified, and for the signal value interval II, the background A color value (R2, G2, B2) suitable for .

The information indicating the color values of the respective signal value intervals I and II input in this manner is the color palette CP, and the color map C is configured by the opacity curve OC and the color palette CP.

While the color map C stored in the color map storage unit 161 is a standard color map created in advance, the color map C created by the color map creation unit 162 is based on the operator's instructions. It becomes a custom-made color map created by Moreover, by adopting the above-described first and second embodiments, it is possible to easily create a color map with a very high degree of freedom through a simple input operation by the operator.

In this way, the color map C newly created by the color map creating unit 162 has a very high degree of freedom, but it is created by reflecting the information contained in the tomographic image group 10 of a specific individual patient. Therefore, it is not the optimal color map for displaying the observation target of the particular patient on the display device 60 either. The color map optimization device 101 also plays a role of optimizing the newly created color map C and outputting it as a corrected color map CC.

<2.3 Signal Value Histogram Creation Unit 120>
The signal value histogram creating unit 120 shown in FIG. A process of creating a signal value histogram H indicating the appearance frequency h is performed. , 16, 17, 18 parallel to the XY plane. A predetermined signal value S is assigned to each arranged pixel S(x, y, z). In the above example, the signal value S is represented by 16-bit data and takes values within the range from the minimum value Smin (-32768) to the maximum value Smax (+32767).

The signal value histogram creating unit 120 obtains the frequency h of appearance in the tomographic image group 10 for each of these signal values S, and performs processing to create a signal value histogram H as shown in FIG. 4(a). In other words, the process performed by the signal value histogram generator 120 is a process of counting the number of pixels having a specific signal value out of the large number of pixels forming the tomographic image group 10 shown in FIG. can be done. Therefore, the appearance frequency h shown on the vertical axis of the graph of the signal value histogram H shown in FIG.

<2.4 Visibility Histogram Creation Unit 130>
The field of view histogram creation unit 130, as shown in FIG. Then, a process of calculating the product "H×OC" of both is performed. This product "H×OC" corresponds to a histogram obtained by weighting the signal value histogram H with the opacity curve OC. As described in §1.2, in the present application, this product "H×OC" is called "visibility histogram VH (symbol VH stands for Visibility Histogram)".

As shown in FIG. 4, in both the signal value histogram H and the opacity curve OC, the horizontal axis is the signal value S in the range from the minimum value Smin to the maximum value Smax. Therefore, for each signal value S, the field of view histogram creating unit 130 calculates the product h×α of the appearance frequency h and the opacity α, for example, as indicated by the dashed line in FIG. 5(a). A process for creating a field of view histogram VH is performed. As described above, the field-of-view histogram VH serves as an index indicating the extent to which pixels having the respective signal values S contribute to the image within the field of view of the observer when observing the volume rendering image.

<2.5 Opacity curve correction unit 140>
As described above, the signal value histogram H is a histogram directly representing the imaging result of the tomographic imaging apparatus 50, whereas the field of view histogram VH is weighted by the opacity curve OC suitable for displaying a specific observation target. It becomes the histogram that went. A feature of the color map optimization apparatus 101 according to the present invention is that the opacity curve OC included in the color map C is corrected based on such a field histogram VH, and a specific tomographic image actually obtained by imaging is obtained. The point is to create an optimum correction color map for use with the group 10. FIG.

As shown in FIG. 6 , the opacity curve correction unit 140 is supplied with a color map C (opacity curve OC and color palette CP) to be corrected from the color map input unit 150 , and the visibility histogram VH from the visibility histogram creation unit 130 . is given. Then, the opacity curve correction unit 140 corrects the opacity curve OC included in the given color map C to create a corrected opacity curve OCC (C at the end means corrected), and replaces the opacity curve OC with output a corrected color map CC (C at the end means corrected) including the corrected opacity curve OCC. Moreover, the opacity curve correction unit 140 performs correction using at least the field of view histogram VH on the opacity curve OC.

In the case of the color map optimization device 101 according to the basic embodiment shown in FIG. Here, the field of view histogram consideration correction unit 141 is a component that performs correction using the field of view histogram VH on the opacity curve OC, and the exponentiation correction unit 142 corrects the opacity curve corrected by the field of view histogram consideration correction unit 141. It is a component that further performs exponentiation correction. The correction performed by the field histogram-considered correction unit 141 will be described in detail in §3.1 “Field-of-view histogram-considered correction unit 141”, and the correction performed by the power correction unit 142 will be described in §3.2 “Power correction unit 142”.

6, the corrected opacity curve OCC included in the corrected color map CC finally output from the opacity curve correction unit 140 is included in the color map C to be corrected. The resulting opacity curve OC is corrected by the field-of-view histogram consideration correction unit 141 and further corrected by the exponentiation correction unit 142 . Of course, in the color map optimization apparatus according to the present invention, the correction by the view histogram consideration correction unit 141 is essential, but the correction by the exponentiation correction unit 142 is optional. Therefore, the opacity curve correction unit 140 needs to have at least the field histogram consideration correction unit 141, and the effects of the present invention can be obtained even if the exponentiation correction unit 142 is omitted.

In the case of the embodiment shown here, the opacity curve correction unit 140 performs a process of correcting the opacity curve OC included in the color map C to be corrected to the corrected opacity curve OCC. No correction. Therefore, the color palette CP included in the corrected color map CC output from the opacity curve correction section 140 is the same as the color palette CP provided from the color map input section 150. FIG.

<2.6 Voxel Image Creation Unit 170 and Rendering Processing Unit 180>
Finally, the voxel image creating unit 170 and rendering processing unit 180 shown in FIG. 6 will be described.

The voxel image generating unit 170 converts the signal values S assigned to the respective pixels S (x, y, z) of the plurality of tomographic images input by the tomographic image input unit 110 into the correction colors output from the opacity curve correcting unit 140 . Each voxel V(x, y, z) defined at each pixel position of the tomographic image is transformed into voxel values (R, G, B, α) indicating a predetermined color value and opacity using the map CC. , the voxel image 20 composed of a collection of voxels arranged in a three-dimensional space is created. Specific processing executed by the voxel image creating unit 170 will be described in detail in §4.1 “Voxel image creating unit 170”.

On the other hand, the rendering processing unit 180 performs a process of generating a volume rendering image 30 showing a state in which the voxel image 20 created by the voxel image creating unit 170 is observed from a predetermined viewing direction as a two-dimensional image. Color values (R, G, B) and opacity α are assigned to the individual voxels that make up the voxel image 20 based on the corrected color map CC, so rendering is performed based on these color values and opacity. processing takes place. The volume rendering image 30 generated by this rendering process is a two-dimensional image suitable for observation of a predetermined observation target and expressing the observation target in a predetermined color. Specific processing executed by the rendering processing unit 180 will be described in detail in §4.2 “Rendering processing unit 180”.

<<< §3. Optimize Color Map ＞＞＞
In this §3, specific correction processing for color map optimization by the opacity curve correction unit 140 included in the color map optimization device 101 shown in FIG. 6 will be described. As illustrated, the opacity curve correction unit 140 includes a view histogram consideration correction unit 141 and a power correction unit 142 . These will be described in order below.

<3.1 Visibility Histogram Consideration Correction Unit 141>
As described above, the field of view histogram consideration correction unit 141 corrects the opacity curve OC included in the given color map C to create the corrected opacity curve OCC, and replaces the opacity curve OC with the corrected opacity curve OCC. function to output a corrected color map CC containing The important point here is that the correction for the opacity curve OC is performed using the field of view histogram VH.

The reason why the field of view histogram VH is used for correction in this way is that the peak position of the field of view histogram VH and the peak position of the opacity curve OC are explained as factor 1 of the problem of the prior art in §1.2. This is to eliminate the discrepancy between For example, in the example shown in FIG. 5(a), there is a slight deviation between the peak position Soc of the opacity curve OC and the peak position Svh of the field of view histogram VH. While the opacity curve OC is a standard graph showing the distribution of the opacity α suitable for observing a predetermined observation target, the field of view histogram VH shows the tomographic image of a specific individual patient. This is because the information of the signal value histogram H of the group is included.

As already described in § 1.2, it is presumed that the deviation between the peak positions Soc and Svh is one of the factors that cause blurring of the observation target on the volume rendering image 30. . For example, as in the example shown in FIG. 5(b), if the peak positions Soc and Svh match, the blurring of the observed object is improved. The field-of-view histogram-considered correction unit 141 corrects the opacity curve OC so as to eliminate such a shift in peak position.

In other words, the basic concept of the correction processing performed by the field-of-view histogram consideration correction unit 141 is that the signal value histogram VH obtained by weighting the signal value histogram H with the opacity curve OC is compared to the distribution shape of the field-of-view histogram VH. The opacity curve OC is corrected so that the distribution shape of the corrected field of view histogram OCC obtained by weighting the value histogram H with the corrected opacity curve OCC is closer to the "predetermined predetermined shape". That's what it is. Here, the "predetermined shape" means "a shape that minimizes the deviation between the peak positions Soc and Svh".

Since the field of view histogram VH is originally defined as the product of the signal value histogram H and the opacity curve OC, the field of view histogram VH cannot be directly corrected. , the correction indirectly affects the field of view histogram VH. Therefore, the field of view histogram consideration correction unit 141 creates a corrected opacity curve OCC by correcting the opacity curve OC, and the corrected field of view histogram VHC ( The C at the end means "corrected") so that the shape approaches "a predetermined shape".

In other words, the field of view histogram consideration correction unit 141 corrects the opacity curve OC by a specific method that brings the shape of the corrected field of view histogram VHC closer to the "predetermined predetermined shape". An OCC will be created. Here, two types of specific methods, a fitting method and a flattening method, are given as specific examples.

<3.1.1 Fitting method>
FIG. 10 is a graph showing the principle of the correction process by the fitting method performed by the field-of-view-histogram-considered correction unit 141 shown in FIG. Here, the signal value histogram generator 120 generates a signal value histogram H as shown in FIG. Consider a case where the map C is input and the field of view histogram generator 130 generates the field of view histogram VH as the product of H×OC as indicated by the dashed line in FIG. 10(c).

As shown in FIG. 10(c), both the opacity curve OC and the field of view histogram VH are graphs in the form of raised mountains. There is a deviation on the horizontal axis between them. Therefore, in order to eliminate this deviation, some correction processing is performed on the opacity curve OC shown in FIG. 10(b) to create a corrected opacity curve OCC having a shape as shown in FIG. specification. Then, when the product H×OCC of the signal value histogram H shown in FIG. 10(a) and the corrected opacity curve OCC shown in FIG. Assume that a corrected field of view histogram VHC as indicated by the dashed line in FIG. 10(e) is created.

As shown in FIG. 10(e), both the opacity curve OC and the corrected visual field histogram VHC are graphs in the form of raised mountains. , and the deviation between Soc and Svh shown in FIG. 10(c) is eliminated. In other words, the peak shape of the corrected field of view histogram VHC fits the peak shape of the opacity curve OC.

The basic principle of the fitting method described here is to obtain a corrected field of view histogram VHC having a shape that fits the mountain-like protuberant shape of the opacity curve OC (correction as shown in FIG. 10(e)). 10(b) to 10(d)) to obtain the corrected opacity curve OCC. 10(a), (b), and (c), the peak position Svh of the field of view histogram VH obtained by multiplying the signal value histogram H by the opacity curve OC is the peak position Svh of the opacity curve OC. 10(a), (d), and (e), the signal value histogram H is multiplied by the corrected opacity curve OCC, The peak position Svhc of the corrected field of view histogram VHC obtained by the above method should coincide with the peak position Soc of the opacity curve OC.

In other words, the field of view histogram consideration correction unit 141 performs a predetermined correction on the opacity curve OC shown in FIG. 10(b) so as to obtain the corrected field of view histogram VHC as shown in FIG. 10(e). , a correction process for creating a corrected opacity curve OCC shown in FIG. 10(d). As for the corrected visual field histogram VHC created based on this corrected opacity curve OCC, as shown in FIG. 10(e), the peak position Svhc matches the peak position Soc of the opacity curve OC. It is possible to suppress the blurring of the observation target on the volume rendering image 30 caused by the

The field of view histogram VH shown in FIG. 10(c) does not accurately represent the product H×OC of the signal value histogram H shown in FIG. 10(a) and the opacity curve OC shown in FIG. 10(b). It is deformed for convenience of explanation. Similarly, the corrected field of view histogram VHC shown in FIG. 10(e) is the product H×OCC of the signal value histogram H shown in FIG. 10(a) and the corrected opacity curve OCC shown in FIG. 10(d). It is deformed for convenience of explanation, not for illustration. The same applies to FIGS. 12, 14 and 16 shown later.

In this way, the field-of-view histogram consideration correction unit 141 that employs the fitting method as a correction method obtains the distribution shape of the field-of-view histogram VH obtained by weighting the signal value histogram H using the opacity curve OC ( c), the distribution shape of the corrected field of view histogram VHC obtained by weighting the signal value histogram H with the corrected opacity curve OCC (the dashed-dotted line graph in FIG. 10(e)). ) corrects the opacity curve OC so as to be closer to a predetermined shape. Here, the "predetermined shape" is, as described above, "a shape that minimizes the deviation between the peak positions Soc and Svh".

More specifically, the field-of-view histogram consideration correction unit 141 that employs the fitting method compares the peak position Svh of the field-of-view histogram VH obtained by weighting the signal value histogram H with the opacity curve OC. Then, the opacity curve OC is adjusted so that the peak position Svhc of the corrected field of view histogram VHC obtained by weighting the signal value histogram H with the corrected opacity curve OCC is closer to the peak position Soc of the opacity curve OC. should be corrected.

In the example shown in FIG. 10, the field histogram consideration correcting unit 141 corrects the opacity curve OC so that the peak position Svhc of the corrected field histogram VHC matches the peak position Soc of the opacity curve OC, and corrects the corrected opacity curve OC. Although the curve OCC is created, the peak position Svhc does not necessarily have to match the peak position Soc. The effect of the present invention can be obtained as long as the correct correction is performed. That is, if the amount of shift between the two peak positions can be made smaller, blurring of the observation target on the volume rendering image 30 can be reduced.

Next, an example of a specific processing method for correcting the opacity curve OC shown in FIG. 10(b) to the corrected opacity curve OCC shown in FIG. 10(d) will be described. As shown in FIG. 6, the corrected histogram consideration correction unit 141 is provided with the field of view histogram VH created by the field of view histogram creation unit 130 and the opacity curve OC input by the color map input unit 150 . Therefore, the corrected histogram consideration correction unit 141 uses the field of view histogram VH shown in FIG. 10(c) to correct the opacity curve OC shown in FIG. Processing for creating OCC is performed.

This processing can actually be performed by computation using mathematical formulas. 11A and 11B are diagrams for explaining mathematical formulas used in the correction process by the fitting method performed by the field-of-view-histogram-considered correction unit 141. FIG. Calculation of this formula requires the opacity curve OC and the field of view histogram VH. Therefore, FIG. 11(a) shows a specific graph of the opacity curve OC and the graph of the field of view histogram VH. The opacity curve OC and the field of view histogram VH shown in this graph respectively correspond to the opacity curve OC and the field of view histogram VH shown in FIG. It has been replaced by a graph with a shape.

That is, the opacity curve OC shown in FIG. 11A is an isosceles triangle with its apex at the peak position Soc, and the field of view histogram VH is an isosceles triangle with its apex at the peak position Svh. Both peak positions Soc and Svh are shifted on the horizontal axis. Now, let OCmax be the peak value of the opacity curve OC, and VHmax be the peak value of the visual field histogram VH, as shown in the figure. Here, OCmax is a numerical value whose unit is the opacity α shown on the vertical axis on the left side of the graph, whereas VHmax is a numerical value whose unit is the product H×OC shown on the vertical axis on the right side of the graph. Therefore, it is meaningless to compare the absolute values of both, but since the signal value S on the horizontal axis is a common parameter, it is sufficiently meaningful to compare the peak positions Soc and Svh.

Now, let OC(S) be the opacity (value of α) of the opacity curve OC at an arbitrary signal value S, and VH(S) be the value of the field of view histogram VH at an arbitrary signal value S, then the correction coefficient γ at the signal value S (S), as shown in FIG. 11(b),
γ(S)=(VHmax/OCmax)·(OC(S)/VH(S)) Equation (1)
If defined by the following formula, the value OCC(S) indicating the opacity (value of α) of the corrected opacity curve OCC at an arbitrary signal value S is as shown in FIG.
OCC(S)=γ(S)・OC(S) Formula (2)
It can be calculated by the following formula.

As shown in equation (2), the correction coefficient γ(S) serves as a parameter for converting the opacity curve OC into the corrected opacity curve OCC, and the opacity indicated by the opacity curve OC at any signal value S OC(S) is corrected to opacity OCC(S) by multiplying it by a correction factor γ(S). The corrected opacity curve OCC becomes a graph showing the opacity OCC(S) corrected in this way.

Note that since equation (1) cannot be defined when VH(S)=0, the correction coefficient γ(S) is not defined for the signal value S at which VH(S)=0, and equation (2) operation is not performed. That is, the opacity OC(S) for the signal value S at which VH(S)=0 becomes the opacity OCC(S) as it is without being corrected.

In practice, a predetermined upper limit value is set for the correction coefficient γ(S) (for example, the upper limit value is 10), and when the value of the correction coefficient γ(S) calculated by Equation (1) exceeds the upper limit value Preferably, the value of the correction coefficient γ(S) is replaced with the upper limit value. Further, since the upper limit of the opacity (α value) indicated by the corrected opacity curve OCC is 1, when the value OCC(S) calculated by the formula (2) exceeds 1, the value of OCC(S) It is necessary to perform processing to replace the value with the upper limit value of 1.

The meaning of equation (1) defining the correction coefficient γ(S) can be easily understood by considering the cases of signal value S=Soc and signal value S=Svh.

For example, the value of the correction coefficient γ(Soc) when the signal value S=Soc is as shown in FIG.
γ(Soc)=(VHmax/OCmax)·(OC(Soc)/VH(Soc)) Equation (3)
is given by Here, as shown in FIG. 11(a), OC(Soc)=OCmax and VH(Soc)=a, so γ(Soc) shown in Equation (3) is
γ(Soc)=(VHmax/a)
become.

At the position Soc, the value of the corrected field of view histogram VHC obtained by multiplying the signal value histogram H by the corrected opacity curve OCC is higher than the value of the field of view histogram VH obtained by multiplying the signal value histogram H by the opacity curve OC. is γ(Soc) times (that is, (VHmax/a) times). means to become

On the other hand, the value of the correction coefficient γ (Svh) when the signal value S=Svh is as shown in FIG.
γ(Svh)=(VHmax/OCmax)·(OC(Svh)/VH(Svh)) Equation (4)
is given by Here, since OC(Svh)=b and VH(Svh)=VHmax as shown in FIG. 11(a), γ(Svh) shown in Equation (4) is
γ(Svh)=(b/OCmax)
become.

At the position Svh, the value of the corrected field of view histogram VHC obtained by multiplying the signal value histogram H by the corrected opacity curve OCC is higher than the value of the field of view histogram VH obtained by multiplying the signal value histogram H by the opacity curve OC. is multiplied by γ(Svh) (that is, (b/OCmax) times), and the value at the position Svh of the corrected field of view histogram VHC is VHmax·(b/OCmax). means that

Thus, the corrected field of view histogram VHC has a peak value VHmax at the position Soc and a value of VHmax·(b/OCmax) at the position Svh. In the above, only the case of signal value S=Soc and the case of signal value S=Svh were considered. , the graph of the corrected field of view histogram VHC has a peak value VHmax at the position Soc and has a mountain-like shape in which the value changes continuously. That is, the peak position Svhc of the corrected field of view histogram VHC coincides with the peak position Soc of the opacity curve OC, and the shape of the corrected field of view histogram VHC fits the shape of the opacity curve OC.

In the example shown in FIG. 11(a), for convenience of explanation, the field of view histogram VH is shown as a simple isosceles triangle having an apex at the peak position Svh. Of course, the field of view histogram VH that is actually obtained does not have such a simple triangular shape, but generally has a mountain-like shape in which the values gradually decrease toward the left and right. However, there are cases where sharp peaks appear at the left and right ends of the horizontal axis (the axis of the signal value S), completely separate from the peaks appearing in the central portion of the mountain shape. Such peaks appearing in the vicinity of both left and right ends can be said to be noise components caused by a unique algorithm adopted when processing the signal values of each pixel of the group of tomographic images 10 obtained from the tomographic imaging apparatus 50. be.

For example, as will be described later in Section 6 "Gradation conversion processing", when processing for compressing a 16-bit signal value to 8 bits is executed, the occurrence frequency h of pixel values corresponding to the lower limit value Lmin of compression processing, The appearance frequency h of the pixel value corresponding to the upper limit value Lmax of the compression process may become abnormally high. This is because, of the signal values represented by 16 bits, the signal values corresponding to less than the lower limit value Lmin of the compression processing are replaced with the lower limit value Lmin, and the signal values corresponding to the values exceeding the upper limit value Lmax of the compression processing are replaced with the lower limit value Lmin. This is because it is replaced with the upper limit value Lmax. Therefore, in the signal value histogram H as shown in FIG. 4(a), the appearance frequency h of pixels having the minimum value Smin and the maximum value Smax becomes abnormally high, and unexpected peaks appear near both left and right ends. It will be.

Therefore, in practice, when searching for the peak position Svh exhibiting the maximum value VHmax in the visual field histogram VH shown in FIG. It is preferable to set the search range within the range (for example, within the range excluding the neighboring portions on the left and right ends). Specifically, in order to obtain the maximum value VHmax within the predetermined signal value range of the field of view histogram VH, the field of view histogram consideration correction unit 141 adds 1 to the minimum signal value of the field of view histogram VH having a value other than 0. A range equal to or greater than the value obtained by subtracting 1 from the maximum signal value having a value other than 0 in the field of view histogram may be set as the predetermined signal value range. By doing so, the search range for the maximum value VHmax can be the signal value range excluding the unexpected peak positions appearing near the left and right ends on the signal value S axis.

11, the maximum value within a predetermined signal value range of the field of view histogram VH is defined as VHmax, the signal value giving the maximum value VHmax is defined as Svh, and the opacity curve OC is defined as Let OCmax be the maximum value within a predetermined signal value range, Soc be the signal value giving the maximum value OCmax, and OC(Svh) be the opacity of the opacity curve OC at the signal value Svh (in the case of FIG. 11, OC (Svh)=b), the value of the corrected field of view histogram VHC at the signal value Soc is VHmax, and the value of the corrected field of view histogram VHC at the signal value Svh is VHmax·OC(Svh)/OCmax (in the case of FIG. 11, VHmax·b /OCmax) to obtain a corrected opacity curve OCC.

<3.1.2 Planarization method>
Next, correction processing by the flattening method will be described. FIG. 12 is a graph showing the principle when the field-of-view-histogram-considered correction unit 141 shown in FIG. 6 performs correction processing using the flattening method. Again, the signal value histogram generator 120 generates a signal value histogram H as shown in FIG. Consider a case where the map C is input and the field of view histogram generator 130 generates the field of view histogram VH as the product of H×OC as indicated by the dashed line in FIG. 12(c). FIGS. 12(a)-(c) are exactly the same graphs as FIGS. 10(a)-(c), and the materials used in the correction process are exactly the same as in the fitting method described above.

As shown in FIG. 12(c), there is a deviation on the horizontal axis between the peak position Soc of the opacity curve OC and the peak position Svh of the field of view histogram VH. In order to eliminate this deviation, the fitting method described above employs a method of moving the peak position Svh of the field of view histogram VH to match the peak position Soc of the opacity curve OC. In the flattening method described here, instead of moving the peak position Svh of the field of view histogram VH, by flattening the shape near the peak of the field of view histogram VH, as a result, the deviation between the peak positions Soc and Svh is reduced. Adopt a method of dissolution.

In this flattening method, a corrected opacity curve OCC as shown in FIG. 12(d) is created by correcting the opacity curve OC shown in FIG. 12(b). This corrected opacity curve OCC is designed to have the following characteristics. That is, when the product H×HCC of the signal value histogram H shown in FIG. 12(a) and the corrected opacity curve OCC shown in FIG. As indicated by the one-dot chain line in FIG. 12(e), a corrected field of view histogram VHC in which the shape near the peak is flattened is created.

As shown in FIG. 12(e), the opacity curve OC is a graph having a triangular ridge with the peak position Soc as the apex, whereas the corrected field of view histogram VHC is a graph with a flattened top. The graph has shape ridges. Moreover, the flat portion (the upper side of the trapezoid) is arranged at a position that includes the signal value section from the peak position Soc of the opacity curve OC to the peak position Svh of the field of view histogram VH. In other words, the point J1 corresponding to the peak position Soc and the point J2 corresponding to the peak position Svh on the graph of the corrected visual field histogram VHC both belong to the flat portion (the upper side of the trapezoid).

The basic principle of the flattening method described here is to obtain a corrected field of view histogram VHC whose upper part is flattened (to obtain a corrected field of view histogram VHC as shown in FIG. 12(e)). To obtain a corrected opacity curve OCC by performing correction processing for the opacity curve OC (correction processing from FIG. 12(b) to FIG. 12(d)). 12(a), (b), and (c), the peak position Svh of the field of view histogram VH obtained by multiplying the signal value histogram H by the opacity curve OC is the peak position Svh of the opacity curve OC. 12(a), (d), and (e), the signal value histogram H is multiplied by the corrected opacity curve OCC, The upper portion of the corrected field of view histogram VHC obtained by the above is flattened so that at least the interval from the corresponding point J1 of the peak position Soc to the corresponding point J2 of the peak position Svh is included in the flat portion.

In other words, the field of view histogram consideration correction unit 141 performs a predetermined correction on the opacity curve OC shown in FIG. 12(b) so as to obtain the corrected field of view histogram VHC as shown in FIG. 12(e). , a correction process for creating a corrected opacity curve OCC shown in FIG. 12(d). If such a correction process is performed, the peak position of the corrected field of view histogram VHC will be distributed over at least the section from the position Soc to the position Svh instead of one point. Therefore, although it cannot be said that the peak position of the corrected field of view histogram VHC coincides with the peak position Soc of the opacity curve OC, the maximum value distribution interval includes the peak position Soc. It is possible to eliminate the deviation that causes blurring of the observation target.

As described above, the field of view histogram VH shown in FIG. 12(c) is the product H×OC of the signal value histogram H shown in FIG. 12(a) and the opacity curve OC shown in FIG. 12(b). It is not exactly shown and is distorted for convenience of explanation. Similarly, the corrected field of view histogram VHC shown in FIG. 12(e) is the product H×OCC of the signal value histogram H shown in FIG. 12(a) and the corrected opacity curve OCC shown in FIG. 12(d). It is deformed for convenience of explanation, not for illustration.

In this way, the field-of-view histogram consideration correction unit 141, which employs the flattening method as a correction method, obtains the distribution shape of the field-of-view histogram VH (see FIG. 12 ( c) The dashed line graph in Fig. 12(e) shows the distribution shape of the corrected field of view histogram VHC obtained by weighting the signal value histogram H with the corrected opacity curve OCC (the dashed line graph in Fig. 12(e) ) corrects the opacity curve OC so as to be closer to a predetermined shape. Here, the "predetermined shape" is, as described above, "a shape that minimizes the deviation between the peak positions Soc and Svh".

More specifically, the field-of-view histogram consideration correction unit 141 that employs the flattening method adjusts the shape near the peak of the field-of-view histogram VH obtained by weighting the signal value histogram H with the opacity curve OC. In comparison, the opacity curve OC is corrected so that the shape near the peak of the corrected field of view histogram VHC obtained by weighting the signal value histogram H with the corrected opacity curve OCC is flattened. You can do it.

In the example shown in FIG. 12, the field of view histogram consideration correction unit 141 determines that the value of the corrected field of view histogram VHC at the peak position Soc of the opacity curve OC (the vertical axis coordinate value of the corresponding point J1 shown in FIG. The opacity curve OC is corrected so as to match the value of the corrected field of view histogram VHC at the peak position Svh of the histogram VH (the vertical coordinate value of the corresponding point J2 shown in FIG. 12(e)) to create the corrected opacity curve OCC. are doing.

Next, an example of a specific processing method for correcting the opacity curve OC shown in FIG. 12(b) to the corrected opacity curve OCC shown in FIG. 12(d) will be described. As shown in FIG. 6, the corrected histogram consideration correction unit 141 is provided with the field of view histogram VH created by the field of view histogram creation unit 130 and the opacity curve OC input by the color map input unit 150 . Therefore, the corrected histogram consideration correction unit 141 uses the field of view histogram VH shown in FIG. 12(c) to correct the opacity curve OC shown in FIG. Processing for creating OCC is performed.

This processing can actually be performed by computation using mathematical formulas. 13A and 13B are diagrams for explaining mathematical formulas used in the correction process by the flattening method performed by the field-of-view-histogram-considered correction unit 141. FIG. Calculation of this formula requires the opacity curve OC and the field of view histogram VH. Therefore, FIG. 13(a) shows a specific graph of the opacity curve OC and the graph of the field of view histogram VH. The opacity curve OC and the field of view histogram VH shown in this graph respectively correspond to the opacity curve OC and the field of view histogram VH shown in FIG. It has been replaced by a graph with a shape.

That is, the opacity curve OC shown in FIG. 13(a) is an isosceles triangle with its apex at the peak position Soc, and the field of view histogram VH is an isosceles triangle with its apex at the peak position Svh. Both peak positions Soc and Svh are shifted on the horizontal axis. Again, as shown in the figure, the peak value of the opacity curve OC is OCmax, and the peak value of the field of view histogram VH is VHmax.

Now, let OC(S) be the opacity (α value) of the opacity curve OC at an arbitrary signal value S, and VH(S) be the value of the field of view histogram VH at an arbitrary signal value S, then the correction coefficient γ at the signal value S (S), as shown in FIG. 13(b),
γ(S)=VHmax/VH(S) Equation (5)
If defined by the following formula, the value OCC(S) indicating the opacity (value of α) of the corrected opacity curve OCC at an arbitrary signal value S is, as shown in FIG.
OCC(S)=γ(S)・OC(S) Formula (6)
It can be calculated by the following formula.

As shown in equation (6), the correction coefficient γ(S) serves as a parameter for converting the opacity curve OC into the corrected opacity curve OCC, and the opacity indicated by the opacity curve OC at any signal value S OC(S) is corrected to opacity OCC(S) by multiplying it by a correction factor γ(S). The corrected opacity curve OCC becomes a graph showing the opacity OCC(S) corrected in this way.

Note that equation (5) cannot be defined when VH(S)=0. Therefore, for signal value S where VH(S)=0, correction coefficient γ(S) is not defined, and equation (6) operation is not performed. That is, the opacity OC(S) for the signal value S at which VH(S)=0 becomes the opacity OCC(S) as it is without being corrected.

In practice, a predetermined upper limit value is set for the correction coefficient γ(S) (for example, the upper limit value is 10), and when the value of the correction coefficient γ(S) calculated by Equation (5) exceeds the upper limit value Preferably, the value of the correction coefficient γ(S) is replaced with the upper limit value. Further, since the upper limit value of the opacity (value of α) indicated by the corrected opacity curve OCC is 1, when the value OCC(S) calculated by Equation (6) exceeds 1, the value of OCC(S) It is necessary to perform processing to replace the value with the upper limit value of 1.

The meaning of equation (5) defining the correction coefficient γ(S) can be easily understood by considering the cases of signal value S=Soc and signal value S=Svh.

For example, the value of the correction coefficient γ(Soc) when the signal value S=Soc is as shown in FIG.
γ(Soc)=(VHmax/VH(Soc)) Equation (7)
is given by Here, as shown in FIG. 13(a), since VH(Soc)=a, γ(Soc) shown in Equation (7) is
γ(Soc)=(VHmax/a)
become.

At the position Soc, the value of the corrected field of view histogram VHC obtained by multiplying the signal value histogram H by the corrected opacity curve OCC is higher than the value of the field of view histogram VH obtained by multiplying the signal value histogram H by the opacity curve OC. is γ(Soc) times (that is, (VHmax/a) times). means to become In other words, the corresponding point J0 on the field of view histogram VH shown in FIG. 13A is corrected to the corresponding point J1 on the corrected field of view histogram VHC.

On the other hand, the value of the correction coefficient γ (Svh) when the signal value S=Svh is as shown in FIG.
γ(Svh)=(VHmax/VH(Svh)) Equation (8)
is given by Here, as shown in FIG. 13(a), since VH(Svh)=VHmax, γ(Svh) shown in Equation (8) is
γ(Svh)=(VHmax/VHmax)=1
become. That is, the value of the correction coefficient γ(S) at the position Svh becomes 1, and the value of the corrected opacity curve OCC at this position is the same as the value of the opacity curve OC before correction. In other words, the corresponding point J2 on the field of view histogram VH shown in FIG. 13A is the same corresponding point J2 on the corrected field of view histogram VHC.

Thus, the corrected field of view histogram VHC has a peak value VHmax at the position Soc and also has a peak value VHmax at the position Svh. In the above, only the case of signal value S=Soc and the case of signal value S=Svh were considered. , VH(S) in which the value of VH(S) is not 0 has a peak value VHmax. That is, the graph of the corrected visual field histogram VHC maintains the peak value VHmax at least for the section from the signal value Soc to Svh. Thus, as shown in FIG. 12(e), a corrected field of view histogram VHC having a trapezoidal raised portion whose upper side is at least the interval from corresponding points J1 to J2 is obtained.

Even in this flattening method, as described above, there are cases where unexpected peaks appear near the left and right ends of the field histogram VH. Instead of using the entire range of S as the search range, it is preferable to set the search range within a predetermined signal value range (for example, within a range excluding the vicinity of both left and right ends).

Specifically, as described in the fitting method, in order to obtain the maximum value VHmax within the predetermined signal value range of the field of view histogram VH, 1 is added to the minimum non-zero signal value of the field of view histogram VH. A range equal to or greater than the value obtained by subtracting 1 from the maximum signal value having a value other than 0 in the field of view histogram may be set as the predetermined signal value range. By doing so, the search range for the maximum value VHmax can be the signal value range excluding the unexpected peak positions appearing near the left and right ends on the signal value S axis.

13, the maximum value within a predetermined signal value range of the visual field histogram VH is VHmax, the signal value giving the maximum value VHmax is Svh, and the opacity curve OC Let OCmax be the maximum value within a predetermined signal value range, and Soc be the signal value that gives the maximum value OCmax. can be said to be a process of obtaining a corrected opacity curve OCC by correcting the opacity curve OC so that VHmax becomes VHmax.

<3.2 Power correction unit 142>
The opacity curve correction unit 140 shown in FIG. 6 includes a view histogram consideration correction unit 141 and a power correction unit 142 . Here, as described in § 3.1, the field of view histogram consideration correction unit 141 is a component that uses the field of view histogram VH to correct the opacity curve OC. explained the method.

The exponentiation correction unit 142 described here is a component that performs the same function as the field histogram consideration correction unit 141 in terms of correcting the opacity curve. A curve is sufficient, and the field of view histogram VH is unnecessary.

In § 1.2, as factor 1 of the problem of the conventional technology, if there is a deviation between the peak position of the field of view histogram VH and the peak position of the opacity curve OC, the object to be observed on the volume rendering image 30 will be blurred. explained what happens. The correction processing by the field-of-view histogram consideration correction unit 141 is performed for the purpose of eliminating the shift between the two peak positions in order to suppress this blurring. On the other hand, in Section 1.2, it was also described that blurring occurs based on the "Weber-Fechner law" that the sensitivity of human sensory organs has a logarithmic characteristic, as Factor 2 of the problem in the prior art. The correction processing by the exponentiation correction unit 142, which will be described below, is performed for the purpose of eliminating blur based on this "Weber-Fechner law".

In the case of the color map optimization device 101 according to the basic embodiment shown in FIG. 6 , the power correction by the power correction unit 142 is superimposed after the correction by the field histogram consideration correction unit 141 . Therefore, actually, the opacity curve OC included in the color map C input by the color map input unit 150 is first given to the field of view histogram consideration correcting unit 141, where correction processing using the field of view histogram VH is performed. applied. Then, the opacity curve obtained by the correction by the view histogram consideration correction unit 141 is given to the power correction unit 142, where the power correction is performed. The opacity curve correction unit 140 outputs a corrected color map C obtained by adding the color map CP to the power-corrected opacity curve.

However, the power correction by the power correction unit 142 does not necessarily have to be performed on the opacity curve corrected by the field histogram consideration correction unit 141, and any opacity curve (for example, the color map C input by the color map input unit 150 It is also possible to do this for the opacity curve OC contained in the . Therefore, a case where the power correction unit 142 performs power correction on an arbitrary opacity curve to be corrected will be described below as an example.

The concept of the correction processing by the exponentiation correction unit 142 is obtained by weighting the signal value histogram H by the opacity curve OC, similarly to the correction processing by the field histogram consideration correction unit 141 described above. Compared to the distribution shape of the field of view histogram VH, the distribution shape of the corrected field of view histogram VHC obtained by weighting the signal value histogram H with the corrected opacity curve OCC is closer to a predetermined shape. Second, it is to correct the opacity curve OC. Here, the "predetermined shape" is "a shape that matches the logarithmic characteristics of the sensitivity of human sensory organs based on the Weber-Fechner law" as will be described later. Here, as specific power correction methods, an embodiment of uniform power correction and an embodiment of S-shaped power correction will be described in order.

<3.2.1 Uniform exponentiation correction>
FIG. 14 is a graph showing the principle when the power correction unit 142 shown in FIG. 6 performs uniform power correction. 14(a) and the opacity curve OC shown in FIG. 14(b). Consider the case where the histogram VH is obtained. In power correction, the field of view histogram VH is not used as a material for correction processing, so it is not necessary to create a field of view histogram VH when performing power correction. Therefore, the field of view histogram VH is also shown.

As shown in FIG. 14(c), there is a deviation on the horizontal axis between the peak position Soc of the opacity curve OC and the peak position Svh of the field of view histogram VH. As described above, the fitting method and the flattening method performed by the view histogram consideration correction unit 141 perform correction for the purpose of eliminating this deviation, but the purpose of power correction is not to eliminate this deviation, but to " To correct the shape of the field of view histogram VH so as to eliminate blur based on the "Weber-Fechner law". Although the example of FIG. 14(c) shows a state in which a deviation occurs, in the case of the color map optimization device 101 according to the basic embodiment shown in FIG. Since the field-of-view histogram-considered correction unit 141 performs the correction before the correction, the deviation has already been eliminated.

In the uniform exponentiation correction described here, a corrected opacity curve OCC as shown in FIG. 14(d) is created by applying a correction that adds a logarithmic characteristic to the opacity curve OC shown in FIG. 14(b). . Comparing FIG. 14(b) and FIG. 14(d), the former is a triangular graph with corners, while the latter is a graph with rounded corners. . Therefore, the corrected field of view histogram VHC corresponding to the product H×HCC of the signal value histogram H shown in FIG. 14(a) and the corrected opacity curve OCC shown in FIG. Thus, it has a shape obtained by adding logarithmic characteristics to the shape of the field of view histogram VH shown in FIG. 14(c). As described above, since the power correction does not change the peak position, in the illustrated example, the peak position of the corrected field of view histogram VHC is the same as the peak position Svh of the field of view histogram VH.

According to experiments conducted by the inventors of the present application, the opacity shown by the corrected visual field histogram VHC in FIG. It is known that the volume rendering image 30 with distribution eliminates blurring of the observation target. This corresponds to the logarithmic characteristics of the sensitivity of human sensory organs based on the "Weber-Fechner's law" described in § 1.2 because the logarithmic characteristics due to power correction are added to the corrected field of view histogram VHC. It is thought that this is because a smooth opacity distribution can be obtained.

14(a), (b), and (c), as indicated by white arrows that change from one to another, compared to the shape of the field of view histogram VH obtained by multiplying the signal value histogram H by the opacity curve OC, compared to the shape of FIG. As indicated by black arrows that change from (a), (d), and (e), the shape of the corrected field of view histogram VHC, which is obtained by multiplying the signal value histogram H by the corrected opacity curve OCC, is more like the "Weber-Fechner It can be said that the shape is more in line with the logarithmic characteristics of the sensitivity of human sensory organs based on the "law of .

A triangular opacity curve OC having corners as shown in FIG. For example, in the case of the example shown in FIG. 8, the operator can define the triangular opacity curve OC by simply specifying the positions of three specified points Q1, Q2, and Q3 on the two-dimensional coordinate system. can be done. However, in light of the "Weber-Fechner's law", such a polygonal opacity curve OC obtained by combining straight lines does not match the logarithmic sensitivity characteristic of human sensory organs. The exponentiation correction can add logarithmic characteristics to such a polygonal opacity curve OC, and is an effective correction for suppressing blurring of the observation target.

Also in this case, the field of view histogram VH shown in FIG. 14(c) is the product H×OC of the signal value histogram H shown in FIG. 14(a) and the opacity curve OC shown in FIG. 14(b). It is deformed for convenience of explanation, not for illustration. Similarly, the corrected field of view histogram VHC shown in FIG. 14(e) is the product H×OCC of the signal value histogram H shown in FIG. 14(a) and the corrected opacity curve OCC shown in FIG. 14(d). It is deformed for convenience of explanation, not for illustration.

Next, an example of a specific processing method (specific method of uniform exponentiation correction) for correcting the opacity curve OC shown in FIG. 14(b) to the corrected opacity curve OCC shown in FIG. 14(d) will be described. This processing can actually be performed by computation using mathematical formulas. 15A and 15B are diagrams for explaining mathematical expressions used in correction processing by uniform power correction performed by the power correction unit 142. FIG. In order to perform calculation based on this formula, it is sufficient to prepare only the opacity curve to be corrected. FIG. 15(a) shows a graph of a specific opacity curve to be corrected. Here, for the sake of convenience, an example in which uniform exponentiation correction is performed on a correction target opacity curve having a simple isosceles triangle shape will be described.

In the case of the color map optimization device 101 according to the basic embodiment shown in FIG. 6, the power correction by the power correction unit 142 is superimposed after the correction by the view histogram consideration correction unit 141, as described above. In this case, the correction target opacity curve shown in FIG. 15 actually corresponds to the corrected opacity curve OCC (capable of correcting the deviation of the peak position) corrected by the view histogram consideration correction unit 141 .

In the uniform power correction process, when the opacity at an arbitrary signal value S of the opacity curve to be corrected is α old (S) and the opacity at the signal value S of the opacity curve after correction is α new (S), As shown in FIG. 15(b),
α new (S) = α old (S) ^γ formula (9)
It is done by performing the following calculation. The word “uniformly” in the uniform exponentiation correction means that the formula (9) is uniformly applied. Here, the correction coefficient γ in this equation (9) takes a value within the range of 0<γ<1 (in the embodiment shown here, γ is set to 0.4). Ultimately, this uniform power correction can be said to be correction in which the opacity indicated by the correction target opacity curve is raised to the power of the correction coefficient γ (where 0<γ<1).

For example, in the case of the opacity curve to be corrected shown in FIG. 15(a), since the opacity for an arbitrary signal value S1 is αold(S1), the opacity αnew(S) for the signal value S1 after power correction is , as shown in FIG.
α new (S1) = α old (S1) ^γ formula (10)
is given by

When such uniform exponentiation correction is performed, logarithmic characteristics can be added to the opacity curve to be corrected, so blurring of the observation target can be suppressed. Note that the uniform exponentiation correction does not necessarily have to be performed on the entire portion of the opacity curve to be corrected, and may be performed only on a portion. For example, in the case of the opacity curve to be corrected shown in FIG. Power correction may be applied.

In the case of the color map optimization device 101 according to the basic embodiment shown in FIG. Power correction is further performed on the corrected opacity curve. In this case, the opacity curve correction unit 140 raises the opacity of part or all of the opacity curve corrected by the field histogram consideration correction unit 141 to the power of a correction coefficient γ (where 0<γ<1). Power correction is performed.

Equation (9) shown in FIG. 15(b) is based on the premise that the opacity α is defined as a numerical value within the range of 0≦α≦1, but it can be executed on a computer. In actual calculation, α is treated as digital data. For example, when the opacity α is expressed as 8-bit digital data, the range that α should take is 0≦α≦255. In that case, the formula used for the actual calculation is not formula (9), but as shown in FIG. 15(b):
α new (S) = 255 × (α old (S) / 255) ^γ formula (11)
becomes a formula.

<3.2.2 S-curve exponentiation correction>
Next, S-curve exponentiation correction will be described. FIG. 16 is a graph showing the principle when the power correction unit 142 shown in FIG. 6 performs S-shaped power correction. Here, FIGS. 16(a) to 16(c) are exactly the same as FIGS. 14(a) to 14(c) used for explaining the uniform exponentiation correction. The difference between FIG. 14 and FIG. 16 is that the correction processing from FIG. 14(b) to FIG. , is the S-curve exponentiation correction described below. As a result, the shape of the corrected opacity curve OCC shown in FIG. 16(d) is slightly different from the shape of the corrected opacity curve OCC shown in FIG. 14(d). The shape is also slightly different from the shape of the corrected visual field histogram VHC shown in FIG. 14(e).

The purpose of this S-curve power correction is to correct the shape of the field of view histogram VH so as to eliminate blurring based on the "Weber-Fechner's law", similar to the purpose of the uniform power correction described above. Therefore, even if there is a deviation between the peak positions Soc and Svh as shown in FIG. 16(c), the deviation remains without being corrected as shown in FIG. 16(e). However, in the case of the color map optimization device 101 according to the basic embodiment shown in FIG. It means that it has been canceled.

In the S-curve exponentiation correction described here as well, a corrected opacity curve OCC as shown in FIG. 16(d) is created by applying a correction that adds logarithmic characteristics to the opacity curve OC shown in FIG. 16(b). be done. When the logarithmic characteristic is added by power correction in this way, it becomes possible to obtain an opacity distribution that matches the logarithmic characteristic of the sensitivity of human sensory organs based on the "Weber-Fechner law", and the blurring of the observation target can be obtained. It is the same as uniform exponentiation correction that the suppression effect can be obtained. Therefore, this S-shaped exponentiation correction is also performed, for example, by adding a logarithmic characteristic to the triangular opacity curve OC created by designating three points Q1, Q2, and Q3 as shown in FIG. This is an effective correction for suppressing blurring of the target.

Also in this case, the field of view histogram VH shown in FIG. 16(c) is the product H×OC of the signal value histogram H shown in FIG. 16(a) and the opacity curve OC shown in FIG. 16(b). It is deformed for convenience of explanation, not for illustration. Similarly, the corrected field of view histogram VHC shown in FIG. 16(e) is the product H×OCC of the signal value histogram H shown in FIG. 16(a) and the corrected opacity curve OCC shown in FIG. 16(d). It is deformed for convenience of explanation, not for illustration.

The difference between the uniform power correction described above and the S-curve power correction described here is, as described above, only in the details of the correction processing to be specifically performed and the shapes of the corrected opacity curve OCC and the corrected field of view histogram VHC obtained as a result. be. Therefore, the difference will be described below.

First, comparing the shape of the corrected opacity curve OCC shown in FIG. 14(d) with the shape of the corrected opacity curve OCC shown in FIG. , the slope near the foot is gentler. Therefore, comparing the shape of the corrected field of view histogram VHC shown in FIG. 14(e) with the shape of the corrected field of view histogram VHC shown in FIG. However, the gradient near the foot is gentler.

The shapes of the corrected opacity curve OCC and the corrected field of view histogram VHC obtained by the S-curve power correction have such unique features because the uniform power correction shown in FIG. 9) is uniformly used, whereas in the S-curve power correction shown in FIG. Therefore, correction is performed to make the gradient near the peak of the mountain steeper and the gradient near the base gentler. The reason why the power correction described here is called S-shaped power correction is that the shape of the slope of the mountain of the opacity curve is corrected to resemble an S-shape. Experiments conducted by the inventors of the present application have shown that such S-curve power correction is more effective than the uniform power correction described above in eliminating the blurring of the observation target.

Next, an example of a specific processing method (a specific method of S-curve exponentiation correction) for correcting the opacity curve OC shown in FIG. 16(b) to the corrected opacity curve OCC shown in FIG. 16(d) will be described. . This processing can actually be performed by computation using mathematical formulas. 17A and 17B are diagrams for explaining mathematical formulas used in correction processing by S-shaped power correction performed by the power correction unit 142. FIG. In order to perform calculation based on this formula, it is sufficient to prepare only the opacity curve to be corrected. FIG. 17(a) shows a graph of a specific opacity curve to be corrected. For the sake of convenience, an example in which the correction target opacity curve having the shape of a simple isosceles triangle is subjected to S-shaped exponentiation correction will be described here as well.

In the case of the color map optimization device 101 according to the basic embodiment shown in FIG. 6, the power correction by the power correction unit 142 is superimposed after the correction by the view histogram consideration correction unit 141, as described above. In this case, the correction target opacity curve shown in FIG. 17 actually corresponds to the corrected opacity curve OCC (capable of correcting the deviation of the peak position) corrected by the view histogram consideration correction unit 141 .

In the S-curve power correction process, αold(S) is the opacity at an arbitrary signal value S of the opacity curve to be corrected, and αnew(S) is the opacity at the signal value S of the corrected opacity curve. , as shown in FIG.
When αold(S)≧T αnew(S)=αold(S) ^γ Formula (12)
When αold(S)<T αnew(S)=αold(S) ^1/γ formula (13)
It is done by performing the following calculation.

Here, the correction coefficient γ in the equations (12) and (13) takes a value within the range of 0<γ<1 (in the embodiment shown here, γ is set to 0.4). . Further, the threshold value T serving as a selection criterion for the formula to be applied may be set to an arbitrary value that satisfies the condition 0<T<OCmax, where OCmax is the maximum value of the opacity curve to be corrected. In the embodiment shown here, T=OCmax/2 is set.

As shown in FIG. 17(a), when T is set to OCmax/2, the threshold value T is half the peak height (OCmax) of the opacity curve to be corrected. The indicated horizontal line serves as a dividing point for selecting which one of the above equations (12) and (13) is applied for calculation. That is, for a certain signal value S, if the value of the opacity αold(S) indicated by the corrected opacity curve is greater than or equal to this horizontal line, the value of the opacity αnew(S) after correction using Equation (12) is If the calculated opacity αold(S) value is less than this horizontal line, the corrected opacity αnew(S) value is calculated using equation (13).

For example, in the case of the opacity curve to be corrected shown in FIG. 17(a), the opacity for the signal value S1 is αold(S1), and αold(S1)≧T. The opacity α new (S1) is obtained by applying the formula (12) shown in FIG. 15(b).
α new (S1) = α old (S1) ^γ formula (14)
is given by On the other hand, the opacity for the signal value S2 is αold(S2), and αold(S2)<T. b) By applying equation (13) shown in
αnew(S2)=αold(S2) ^1/γ formula (15)
is given by

The opacity α takes a value within the range of 0≦α≦1, and the correction coefficient γ takes a value within the range of 0<γ<1. The effect of making the gradient steeper is obtained, and the effect of making the gradient gentler is obtained by performing the correction based on the above equation (13). Therefore, when such S-curve exponentiation correction is performed, logarithmic characteristics are added to the opacity curve to be corrected, and the slope near the peak of the mountain (the portion above the threshold value T) is reduced with respect to the shape of the slope of the mountain. Correction is performed to make the slope steeper and to make the slope near the foot of the mountain (the portion below the threshold value T) gentler. Such correction is very effective in eliminating blurring of the observation target.

It should be noted that this S-shaped exponentiation correction does not necessarily have to be performed on the entire part of the opacity curve to be corrected, and may be performed only on a part of it. For example, in the case of the opacity curve to be corrected shown in FIG. 17(a), the S-shaped exponentiation correction may be applied only to portions where the opacity α satisfies α≧T/2.

In the case of the color map optimization device 101 according to the basic embodiment shown in FIG. Power correction is further performed on the corrected opacity curve. In this case, the opacity curve correction unit 140 corrects the opacity of part or all of the opacity curve corrected by the field histogram consideration correction unit 141 if the opacity is equal to or greater than a predetermined threshold value T. Correction is performed by raising the coefficient γ (where 0<γ<1), and if the opacity is less than the threshold value T, correction is performed by raising the reciprocal of the correction coefficient γ.

The formulas (12) and (13) shown in FIG. 17(b) are based on the premise that the opacity α is defined as a numerical value within the range of 0≦α≦1. In the actual computations performed above, α is treated as digital data. For example, when the opacity α is expressed as 8-bit digital data, the range that α should take is 0≦α≦255. In that case, when performing the S-curve exponentiation correction, an equation based on equation (11) shown in FIG. 15(b) is used instead of equations (12) and (13).

<<< §4. Creating a volume rendered image ＞＞＞
So far, we have mainly described specific embodiments of the color map optimization device 101 included in the volume rendering device 100 shown in FIG. As shown in FIG. 6, the color map optimization device 101 is composed of a tomographic image input unit 110, a signal value histogram generation unit 120, a field of view histogram generation unit 130, an opacity curve correction unit 140, and a color map input unit 150. .

The volume rendering device 100 shown in FIG. 6 is obtained by adding a color map storage unit 161, a color map creation unit 162, a voxel image creation unit 170, and a rendering processing unit 180 to the color map optimization device 101. . An example of the color map storage unit 161 has already been described in §2.2.2. Also, an embodiment of the color map creation unit 162 has already been described in §2.2.3. Accordingly, examples of the voxel image creation unit 170 and the rendering processing unit 180 will be described in order below.

<4.1 Voxel Image Creation Unit 170>
As shown in FIG. 6, the voxel image generating unit 170 receives the tomographic image group 10 (collection of pixels S (x, y, z)) input by the tomographic image input unit 110, and the color map optimization device 101. Optimized correction color map CC (correction color map CC output from opacity curve correction section 140) is provided. The voxel image creating unit 170 performs processing to create the voxel image 20 based on these data.

The voxel image 20 is a three-dimensional image model in which voxels V (x, y, z) assigned color values (R, G, B) and opacity α are arranged in a three-dimensional space. FIG. 18 is a conceptual diagram showing the voxel configuration of the voxel image 20 created by the voxel image creating section 170. As shown in FIG. As shown, a plurality of voxel planes 11v, . are placed. Each voxel V(x, y, z) is specified by coordinate values (x, y, z) on the XYZ three-dimensional orthogonal coordinate system.

, 16v, 17v, and 18v are only partially shown in FIG. 17 and 18, and individual voxels V(x, y, z) on each voxel plane correspond to individual pixels S(x, y, z) on each tomographic image shown in FIG. It is something to do. Therefore, in reality, a total of 370 voxel planes corresponding to a total of 370 tomographic images are stacked in the Z-axis direction at predetermined intervals (imaging intervals of the tomographic imaging apparatus 50).

18 are equal to the axial sizes Sx, Sy, Sz of the tomographic image group 10 shown in FIG. In the examples shown in FIGS. 2 and 18, Sx=6, Sy=6, and Sz=8 for convenience of explanation, but in the case of the example shown in FIG. , Sy=512 and Sz=370.

Each pixel S (x, y, z) forming the tomographic image group 10 shown in FIG. It is On the other hand, each voxel V (x, y, z) forming the voxel image 20 shown in FIG. 18 is given a predetermined color value (R, G, B) and an opacity α. That is, each voxel V(x, y, z) is given a voxel value (R, G, B, α). These voxel values (R, G, B, α) can be represented, for example, by 8-bit data indicating values within the range of 0-255.

The voxel image generator 170 performs processing for converting the tomographic image group 10 shown in FIG. 2 into the voxel image 20 shown in FIG. In order to perform such conversion, the signal value S given as a pixel value to each pixel S (x, y, z) constituting the group of tomographic images 10 is converted to a voxel value (R, G, B, α ) must be converted to This conversion processing is performed using the corrected color map CC provided from the color map optimization device 101 .

, 16, 17, 18 input by the tomographic image input unit 50. is converted into voxel values (R, G, B, α) indicating a predetermined color value and opacity using the corrected color map CC output from the opacity curve correction unit 140, and at each pixel position of the tomographic image By assigning the voxel value to each defined voxel V(x, y, z), a voxel image 20 consisting of a set of voxels arranged in a three-dimensional space is created. Similar to the color map C shown in FIG. 7, the corrected color map CC is a table that associates each signal value S with a predetermined voxel value (R, G, B, α). 170 can use this table to convert the signal values S into voxel values (R, G, B, α) to create the voxel image 20 .

<4.2 Rendering Processing Unit 180>
The rendering processing unit 180 shown in FIG. 6 generates a two-dimensional image obtained when the voxel image 20 (three-dimensional image model) illustrated in FIG. Execute the rendering process that creates the image. A two-dimensional image created in this way is output as a volume rendering image 30 . The volume rendering image 30 shown in FIG. 1(c) is the volume rendering image 30 output from the rendering processing section 180 displayed on the screen of the display device 60. FIG.

FIG. 19 is a flowchart showing the procedure of rendering processing performed by the rendering processing unit 180. As shown in FIG. As illustrated, first, in a voxel image input step S1, a voxel image 20 created by the voxel image creation unit 170 is input. Each of the following steps is sequentially executed based on this voxel image 20 .

In the subsequent smoothing processing step S2, smoothing processing is performed to smooth the distribution of the opacity α given to each voxel. For B), a shadow calculation is performed to add shadow information generated according to a predetermined lighting environment. Details of the smoothing processing stage of step S2 are described in §4.2.1, and details of the shadow calculation stage of step S3 are described in §4.2.2.

Next, in a display condition setting step S4, display conditions are set for displaying the voxel image 20 given as a three-dimensional image model as a two-dimensional volume rendering image 30. FIG. Specifically, the rendering processing unit 180 sets display conditions such as the viewpoint position, line-of-sight direction, and display magnification based on the input from the operator.

In the subsequent step S5, a processing method is selected. In the embodiment shown here, two methods are prepared for volume rendering processing. The first method is volume rendering processing by the ray casting method, which is executed by a CPU-version program entrusting processing to the CPU. The second method is a volume rendering process using a 3D texture mapping method, which is executed by a GPU version of a program entrusting processing to the GPU. Either method is selected and executed when the operator provides input specifying the desired method.

If the first method, that is, the rendering using the CPU version program is selected, the process proceeds from step S5 to step S6 to execute volume rendering processing by the ray casting method, and further to step S7 to display the data on the display. is performed. Since the volume rendering processing by the ray casting method executed in step S6 is software processing in which the CPU executes the application program, it is difficult to speed up the processing. Therefore, the stage of display on the display in step S7 is a progressive display (a gradual display method in which an incomplete display during calculation gradually transitions to a complete display as the calculation progresses).

On the other hand, if the second method, that is, rendering using the GPU version of the program is selected, the process advances from step S5 to step S8 to execute volume rendering processing using the 3D texture mapping method, and then advances to step S9. , the presentation stage on the display is performed. Since the volume rendering processing by the 3D texture mapping method executed in step S8 is processing executed by the GPU, which is hardware specialized for graphic processing, it is possible to speed up the processing. Therefore, the stage of display on the display in step S9 is a real-time display (a display method of presenting a complete rendered image immediately each time the display conditions are set in step S4).

In this way, the processes of steps S4 to S9 are repeatedly executed until it is determined in step S10 that the operator has given an end instruction. Therefore, the operator can display the desired volume rendering image 30 on the screen of the display device 60 while variously changing the display condition settings in step S4. Details of the volume rendering stages of steps S6 and S8 are described in §4.2.3.

<4.2.1 Smoothing processing step>
Next, the smoothing processing step S2 in the flowchart of FIG. 19 will be described. The smoothing process is a process for smoothing the distribution of the opacity α given to each voxel forming the voxel image 20 . Specifically, a process of replacing the opacity α of each voxel V (x, y, z) that constitutes the voxel image 20 with the average value of the opacities of the voxel and adjacent voxels adjacent to the voxel is performed. .

FIG. 20 is a diagram for explaining the specific contents of the smoothing process performed in this smoothing process step S2. FIG. 20(a) shows a specific target voxel V(x, y, z) and 26 adjacent voxels arranged around it. In the drawing, the hatched voxel V(x, y, z) is the voxel of interest, which is located at the coordinate (x, y, z) on the XYZ three-dimensional orthogonal coordinate system. In the case of the embodiment described here, the opacity α in the voxel values (R, G, B, α) given to this target voxel V(x, y, z) is A smoothing process for correction is performed by referring to the voxel values (opacity α) of a total of 27 voxels.

Z(-1), Z(0), and Z(+1) shown in FIG. 20(a) are voxel planes on which voxels are arranged, and voxel plane Z(0) is voxel plane Z(-1). Voxel plane Z(+1) is the plane immediately above and adjacent to voxel plane Z(0). Further, here, each coordinate value of the coordinate system is defined according to the arrangement pitch of voxels. (x+1, y, z) and the neighboring voxel to the left becomes voxel V(x-1, y, z). The same applies to adjacent voxels adjacent in the Y-axis direction and the Z-axis direction.

Here, the opacity α in the voxel values (R, G, B, α) assigned to each voxel is represented as αold, and the opacity α after correction is represented as αnew. In addition, the opacity α for a specific voxel is shown together with the coordinate values of the specific voxel. For example, the opacity α before modification of the target voxel V (x, y, z) located at coordinates (x, y, z) is α old (x, y, z), and the opacity α after modification is α old (x, y, z). becomes α new (x, y, z).

Then, in the smoothing process according to the embodiment shown here, the corrected opacity α new (x, y, z) of the voxel V (x, y, z) of interest is expressed as follows, as shown in FIG.
α new (x, y, z) = Σ _{k = -1 to 1} Σ _{j = -1 to 1} Σ _{i = -1 to 1}
α old (x+i, y+j, z+k)/27 Equation (16)
It can be said that this is a process of obtaining based on the following formula. The right side of this equation (16) indicates the average value of the old opacity α of a total of 27 voxels including the voxel of interest and adjacent voxels, and this average value is the new opacity α of the voxel of interest.

Note that edge voxels located at the outer edges of the voxel image 20 (for example, in the case of the example shown in FIG. 18, voxels located at the left and right edges, the front and rear edges of each voxel plane, the top voxel plane, and the bottom voxel plane voxels located at ), there are no voxels adjacent to the outside, so if these end voxels are set as voxels of interest, the calculation of equation (16) cannot be performed. Therefore, these edge voxels are excluded from the smoothing process.

In this way, the rendering processing unit 180 shown in FIG. 6 performs smoothing processing that replaces the opacity α of each voxel forming the voxel image 20 with the average value of the opacity α of the voxel and adjacent voxels adjacent to the voxel. and generate a volume rendering image based on the smoothed voxel image. Such smoothing processing is effective in obtaining a more appropriate volume rendering image 30 .

<4.2.2 Shadow calculation stage>
Next, the shadow calculation step S3 in the flowchart of FIG. 19 will be described. The shadow calculation is a process of adding shadow information generated according to a predetermined lighting environment to the color values (R, G, B) given to each voxel forming the voxel image 20 .

FIG. 21 is a diagram for explaining specific details of the shadow calculation performed in the shadow calculation step S3. Now, let us assume that a voxel image 20 (shown as a rectangular parallelepiped block in the figure for convenience) is arranged on an XYZ three-dimensional orthogonal coordinate system, as shown in FIG. 21(a). In the embodiment shown here, it is assumed that the voxel image 20 is illuminated with parallel illumination light L from a predetermined direction. Then, a light source vector L (Lx, Ly, Lz) is defined to indicate the direction of the parallel illumination light L, and shadow calculation is performed using this light source vector.

A specific formula for performing such shadow calculation is shown in FIG. 21(b). First, the axial components Gx, Gy, Gz of the gradient vector G (Gx, Gy, Gz) are defined by the following equations shown in the upper part of FIG. 21(b).
Gx=(α(x+1, y, z)-α(x-1, y, z))・(Rxy/Rz)
Equation (17)
Gy=(α(x,y+1,z)-α(x,y−1,z))・(Rxy/Rz)
Equation (18)
Gz=(α(x, y, z+1)−α(x, y, z−1))
Equation (19)

Here, Gx indicates the gradient of the opacity α in the X-axis direction for a given target voxel V(x, y, z), and the opacity of two pairs of adjacent voxels in the X-axis direction. defined as the difference between Similarly, Gy and Gz indicate the gradient of opacity α in the Y-axis direction and Z-axis direction, respectively, and are also defined as the difference in opacity of two pairs of adjacent voxels in each axis direction. there is

The term (Rxy/Rz) at the end of equations (17) and (18) is a correction term for correcting the difference between the resolution Rxy in the X-axis and Y-axis directions and the resolution Rz in the Z-axis direction. . As described above, the voxel image generators 170 are arranged in the X-axis direction at a predetermined X-axis direction pitch, in the Y-axis direction at a predetermined Y-axis direction pitch, and in the Z-axis direction at a predetermined Z-axis direction pitch. A voxel image 20 consisting of a three-dimensional grid-like voxel assembly arranged in the order of . This voxel arrangement conforms to the arrangement of pixels forming the tomographic image group 10 shown in FIG.

As shown in FIG. 2, the group of tomographic images 10 is configured as a stack of a large number of tomographic images arranged at predetermined intervals in the Z-axis direction. Here, the pitch of each tomographic image in the Z-axis direction corresponds to the imaging interval of the tomographic imaging apparatus 50 . On the other hand, the pitches in the X-axis direction and the Y-axis direction of pixels constituting one tomographic image are determined by the two-dimensional resolution of the tomographic image at the time of imaging. set equal. Therefore, in the embodiment shown here, the resolutions in the X-axis direction and the Y-axis direction of the voxels included in the voxel image 20 are set to the same value Rxy, and the resolution in the Z-axis direction Rz is different from Rxy. set to different values. Therefore, in defining the gradient vector G, it is necessary to correct for these differences in axial pitch. The correction term (Rxy/Rz) is a term for such correction. Note that the formula (19) is a formula that defines the Z-axis direction component Gz, so the correction term is 1.

In short, in the case of the embodiment shown here, the rendering processing unit 180 generates an X-axis difference value indicating the difference in opacity between a pair of adjacent voxels arranged on both sides along the X-axis direction for a given voxel of interest. , a Y-axis difference value indicating the difference in opacity between a pair of adjacent voxels arranged on both sides along the Y-axis direction, and a difference in opacity between a pair of adjacent voxels arranged on both sides along the Z-axis direction is calculated, and the product of the X-axis difference value multiplied by the coefficient (Rxy/Rz) corresponding to the X-axis direction pitch is taken as the X-axis direction component Gx, and the Y-axis difference value is the Y-axis direction A vector having the Y-axis direction component Gy as the product of the product of the coefficient (Rxy/Rz) corresponding to the pitch and the Z-axis direction component Gz as the product of the Z-axis difference value multiplied by the coefficient 1 corresponding to the Z-axis direction pitch. , a gradient vector G (Gx, Gy, Gz) for the voxel of interest.

The magnitude G of the gradient vector G (Gx, Gy, Gz) is, as shown in the upper part of FIG. 21(b),
G=(Gx ² +Gy ² +Gz ² ) ^1/2 Formula (20)
defined by

The gradient vector G(Gx, Gy, Gz) is then used to define the brightness value E(x, y, z) for the voxel of interest V(x, y, z). That is, the luminance value E(x, y, z) is, as shown in the middle part of FIG. 21(b),
E (x, y, z) = (1-Ab) (Gx · Lx + Gy · Ly + Gz · Lz)
/G + Ab Formula (21)
is given by

Here, Ab is an ambient light component coefficient, and is set to a value in the range of 0≦Ab≦1. This ambient light component coefficient Ab indicates the ratio of components of ambient light (for example, light emitted by the voxel itself) not caused by the parallel illumination light L shown in FIG. 21(a). Therefore, the ratio of the contribution of the parallel illumination light L to the luminance value is (1-Ab). The term (Gx · Lx + Gy · Ly + Gz · Lz) in equation (21) corresponds to the inner product of the gradient vector and the light source vector, and the value obtained by dividing this by the magnitude G of the gradient vector was due to the parallel illumination light L luminance value. The final luminance value E (x, y, z) is calculated as the sum of the luminance value due to the parallel illumination light L and the luminance value due to the ambient light, weighted by the ambient light component coefficient Ab. be done.

Note that edge voxels located at the outer edges of the voxel image 20 (for example, in the case of the example shown in FIG. 18, voxels located at the left and right edges, the front and rear edges of each voxel plane, the top voxel plane, and the bottom voxel plane voxels located at ), there are no voxels adjacent to the outside. , the gradient vector G(Gx, Gy, Gz) cannot be determined. Therefore, for these edge voxels, the luminance value E(x, y, z) may be set to 0 for convenience.

After the brightness value E(x, y, z) is calculated for each voxel in this way, as shown in the lower part of FIG. 21(b),
R new (x, y, z)=E (x, y, z) R old (x, y, z) Equation (22)
G new (x, y, z) = E (x, y, z) · Gold (x, y, z) Equation (23)
B new (x, y, z)=E (x, y, z) Bold (x, y, z) Equation (24)
An operation based on the following formula is performed. Here, Rold (x, y, z), Gold (x, y, z), and Bold (x, y, z) are the color values R, G, and B before shading calculation, and Rnew ( x, y, z), Gnew (x, y, z), and Bnew (x, y, z) are the color values R, G, and B after shading calculation, respectively.

As a result, the rendering processing unit 180 according to the present embodiment creates a gradient vector G (Gx, Gy, Gz ), and the luminance value E (x, y, z) is calculated in consideration of the direction of the light source vector L (Lx, Ly, Lz) and the gradient vector G (Gx, Gy, Gz) pointing in a predetermined direction. , performs a shading process for correcting the color value (R, G, B) of each voxel by multiplying the luminance value E (x, y, z), and based on the voxel image after this shading process , the volume rendering image 30 is generated.

<4.2.3 Volume Rendering Stage>
Next, the volume rendering steps S6 and S8 in the flowchart of FIG. 19 will be described. Both are processes for creating a two-dimensional volume rendering image 30 obtained when a voxel image 20 (three-dimensional image model) that has undergone smoothing processing and shadow calculation (three-dimensional image model) is observed from a predetermined viewing direction. As described above, in the volume rendering stage S6, the ray casting method using the CPU version of the program is performed, and in the volume rendering stage S8, the 3D texture mapping method using the GPU version of the program is performed. Since any processing is well-known processing, only the outline will be briefly described here.

First, volume rendering processing by the ray casting method executed in step S6 will be described using the basic conceptual diagram of FIG. Now, as shown in FIG. 22(a), it is assumed that the voxel image generator 170 has generated a voxel image 20 consisting of a set of voxels arranged on the XYZ three-dimensional orthogonal coordinate system, and the operator has set Consider a case where a volume rendering image 30 is generated on the projection plane P1 shown in the figure based on display conditions. FIG. 22(a) shows an XYZ three-dimensional orthogonal coordinate system in which the X axis is in the right direction of the figure, the Z axis is in the upper direction of the figure, and the Y axis is in the direction perpendicular to the paper surface of the figure. Since the projection plane P1 is set each time based on the display conditions specified by the operator, it is a plane oriented in an arbitrary direction with respect to the coordinate system.

The basic principle of the ray casting method is to irradiate a virtual ray (a dashed-dotted arrow in the figure) from a projection point F on the projection plane P1 toward the voxel image 20 in a direction orthogonal to the projection plane P1, and on this virtual ray A volume rendering image 30 is generated on the projection plane P1 by giving the color value calculated based on the color value and opacity for each voxel of , as the pixel value of the pixel located at the projection point F. . More specifically, first, along the virtual ray from the projection point F, the degree of transmission of the virtual ray is calculated from the front voxel to the back voxel in order with reference to the opacity α. Find the deepest point D that cannot be reached from the point onwards. Then, this time, following the path of the virtual ray in reverse (solid arrow in the figure), from the deepest point D to the projection point F, taking into account the opacity α of each voxel, the color value of each voxel ( (R, G, B)) on the projection point F is sequentially calculated, and finally the color value (R, G, B) of the pixel at the position of the projection point F is determined.

In practice, in order to reduce the computational burden of such processing, when generating the volume rendering image 30 on a predetermined projection plane P1, the XYZ cubic It is preferable to perform a coordinate transformation that rotates the original orthogonal coordinate system. FIG. 22(b) is a diagram showing an example in which the above-described ray casting method is applied to the voxel image 25 after such coordinate transformation. For the sake of convenience, the coordinate system shown in FIG. 22(a) and the coordinate system shown in FIG. 22(b) are drawn in the same direction, and the projection plane P1 and the projection plane P2 are drawn in different directions. However, in reality, the coordinate system is rotated, and the projection planes P1 and P2 are arranged at the same position and in the same direction. That is, the coordinate system shown in FIG. 22(b) is obtained by rotating the coordinate system shown in FIG. 22(a) in a predetermined direction. , is an image that has undergone coordinate transformation according to the rotation of the coordinate system.

If such coordinate transformation is performed, the projection plane P2 becomes a plane parallel to the XY plane. When a virtual ray (a dashed-dotted line arrow in the drawing) is irradiated, the virtual ray always becomes a ray parallel to the Z-axis. Therefore, in the above-described ray casting method, a virtual ray is irradiated from the projection point F on the projection plane P2 to the voxel image 25 after the coordinate transformation in a direction parallel to the Z axis, and each voxel on this virtual ray By giving the color values calculated based on the color values and the opacity as the pixel values of the pixels located at the projection point F, the volume rendering image 30 can be generated.

That is, as shown in FIG. 22(b), if the ray casting method is performed using the voxel image 25 after coordinate transformation, the process of obtaining the deepest point D using virtual rays and the projection point D from this deepest point D can be performed. Since the process of sequentially calculating the effects of the color values (R, G, B) of each voxel on the projection point F toward F can be performed along the direction parallel to the Z-axis, the operation The burden can be greatly reduced.

Next, the volume rendering processing by the 3D texture mapping method executed in step S8 will be explained using the basic conceptual diagram of FIG. First, as shown in FIG. 23(a), a large number of mapping planes M are prepared on the XYZ three-dimensional orthogonal coordinate system. All of the mapping planes M are planes parallel to the XY plane, and are arranged at predetermined intervals in the Z-axis direction. Then, the voxel images 20 created by the voxel image creation unit 170 are mapped onto these many mapping planes M as three-dimensional texture images. A thick line portion on the mapping plane M shown in FIG. 23(a) indicates an image of each mapped texture image.

Also in this embodiment, coordinate transformation is performed to reduce the computational load. Therefore, again, consider the case where the volume rendering image 30 is generated on the projection plane P1 shown in the figure based on the display conditions set by the operator. In this case also, coordinate transformation is performed to rotate the coordinate system so that the projection plane is parallel to the XY plane. FIG. 23(b) shows the state after such coordinate conversion, and the projection plane P2 is a plane parallel to the XY plane.

However, in the case of the embodiment shown here, the mapping plane M is not rotated and remains in its original position, the texture mapped on this mapping plane M is rotated, and the projection plane is also rotated accordingly. I'm trying The difference between the thick line portion on the mapping plane M shown in FIG. 23(a) and the thick line portion on the mapping plane M shown in FIG. 23(b) is that the texture to be mapped has been rotated. showing. Of course, such texture rotation can also be performed by a coordinate transformation operation from the first coordinate system to the second coordinate system.

As described above, volume rendering processing by this 3D texture mapping method is executed by the GPU, which is hardware specialized for graphic processing. Therefore, relatively high-speed processing can be expected. Therefore, the process of determining the color values (R, G, B) of the pixels positioned at the respective projection points on the projection plane P2 is performed on the rear mapping plane M (shown in the lower part of FIG. 23(b)). From the texture mapped on the mapping plane M) to the texture mapped on the mapping plane M on the front side (mapping plane M shown in the upper part of FIG. 23(b)), one by one, By referring to the voxel values (R, G, B, α) of each voxel on the texture, the process of accumulating the influence on the projection point may be executed.

As a result, when performing volume rendering processing by the 3D texture mapping method, the voxel image 20 created by the voxel image creating unit 170 is a three-dimensional texture image composed of a set of voxels arranged on the first XYZ three-dimensional orthogonal coordinate system. will be treated as Then, when generating the volume rendering image 30 on a predetermined projection plane, the rendering processing unit 180 defines a second XYZ three-dimensional orthogonal coordinate system in which the XY plane is parallel to the projection plane. , on this second XYZ three-dimensional orthogonal coordinate system, a plurality of mapping planes M parallel to the XY plane are defined at predetermined intervals in the Z-axis direction, and from the first XYZ three-dimensional orthogonal coordinate system to the second XYZ A three-dimensional texture image is mapped onto a plurality of mapping planes M by performing coordinate transformation to a three-dimensional orthogonal coordinate system, and color values of projection points having XY coordinate values (x, y) on the projection plane are XY coordinate values (x, y) on the nearest mapping plane from the color value of the pixel mapped at the position of XY coordinate values (x, y) on the mapping plane farthest from the projection point The volume rendering image 30 is generated by sequentially blending and determining the color values of the pixels mapped to the positions in consideration of the opacity of each pixel.

In this way, in volume rendering processing by the 3D texture mapping method, calculations are performed for all mapping planes M from the back side to the front side, so the calculation results for the mapping plane M on the back side may be wasted. Although there are many, since it is hardware processing by GPU, high-speed processing is possible, and as described above, real-time display is possible.

<<< §5. Additional Embodiments of the Invention >>>
Having thus far described the volume rendering device 100 according to the basic embodiment of the invention shown in FIG. 6, a volume rendering device according to an additional embodiment of the invention will now be described.

<5.1 Additions to Basic Embodiment>
FIG. 24 is a block diagram showing the configuration of a volume rendering device 100' (including a color map optimization device 101) according to an additional embodiment of the present invention. The difference between the device 100 according to the basic embodiment shown in FIG. 6 and the device 100' according to the additional embodiment shown in FIG. and that the former voxel image creation unit 170 is changed to the voxel image creation unit 175 in the latter. Therefore, this change will be described below.

In §1.2 “Problems of the Prior Art and Their Factors”, three factors have been explained as factors for generating an inappropriate volume rendering image. Of these factors, the first factor, "deviation of the field of view histogram VH with respect to the opacity curve OC", is described in §3.1 as being able to be resolved by correction by the field of view histogram consideration correction unit 141 using a fitting method, a flattening method, or the like. Indicated. As for the second factor "Weber-Fechner's law", it was shown in §3.2 that it can be solved by correction by the power correction unit 142 using uniform power correction, S-curve power correction, or the like. Additional embodiments described herein propose a solution to the third factor, "mixing of different bodies with proximity signal values".

As described above, the signal values of pixels included in a tomographic image have different distribution ranges for each tissue such as internal organs, bones, skin, muscle, and fat. Therefore, in order to create a volume rendering image 30 suitable for a specific observation target, an opacity curve OC having a raised portion in a signal value distribution range specific to the observation target is used to selectively A display method is used.

However, in reality, there are different bodies whose signal value distribution ranges are close to each other in tissues other than the observation target or in internal substances such as air and water. The same applies to the bed, fixing belt, and the like used when performing imaging with the tomographic imaging apparatus 50 . In this way, when different objects having proximity signal values coexist, on the volume rendering image 30, the different objects are displayed together with the observation object, and as a result, the observation object is difficult to observe. is displayed incorrectly. In particular, optimizing the color map using the method described in §3 improves the resolution of the organs to be observed, but also improves the resolution of the bed and body surface areas (skin, muscle, fat, etc.). This may result in the target organ region being obscured.

The additional embodiments described here are new proposals for solving such problems, and the basic concept thereof is "the position of the observation object in the three-dimensional space, not the distribution range of the signal value. It is based on the idea that

For example, when the observation target is the lungs of a human body, in order to limit the lungs in the signal value distribution range, a sieve based on the lung observation opacity curve OC having a raised portion in the signal value distribution range of the lung tissue is applied. However, if there is another object having a signal value distribution range similar to that of the lung tissue, the volume rendering image 30 includes not only the lung but also the other object. On the other hand, sifting based on the position occupied by the lungs in the three-dimensional space can prevent the mixture of different bodies distant from the lungs. An additional embodiment described here uses a sieve based on the position in the three-dimensional space in addition to the sieve based on the opacity curve OC, and suppresses the mixture of different objects with proximity signal values. .

For this reason, the volume rendering device 100' shown in FIG. 24 is provided with an opacity correction magnification setting unit 190, and the voxel image creation unit 175 uses the correction color map CC (opacity sieve based on the curve OC) and the opacity correction magnification ξ set in the opacity correction magnification setting unit 190 (sieve based on the position in the three-dimensional space) to create the voxel image 20. ing.

A predetermined opacity correction magnification ξ is set in the opacity correction magnification setting unit 190 according to the position in the three-dimensional space. Then, the opacity α indicated by the corrected color map CC is multiplied by the opacity correction magnification ξ set according to the position of the voxel, and the voxel value including the opacity α after this magnification correction is A process to be assigned to the voxel is performed.

As shown in FIG. 18, the voxel image generating unit 170 arranges voxel planes 11v to 18v corresponding to the tomographic images 11 to 18 so as to be parallel to the XY plane, and each voxel plane 11v to 18v A voxel image 20 consisting of a set of voxels arranged three-dimensionally on an XYZ three-dimensional orthogonal coordinate system by defining a voxel V (x, y, z) corresponding to a pixel S (x, y, z) of a tomographic image. to create

Here, it is assumed that a tomographic image group 10 is given, with the human body as an object to be imaged, with the Z axis along the spine of the human body, the X axis in the lateral direction of the human body, and the Y axis in the longitudinal direction of the human body. specification. In this case, one tomographic image is an image of a cross section parallel to the XY plane obtained by slicing the human body, as shown in FIG. 1(b). , is close to an ellipse with the minor axis in the Y-axis direction. Of course, each coordinate axis of the voxel image 20 shown in FIG. 18 is also defined in the same direction as the tomographic image. Also, here, as in the example shown in FIG. 18, it is assumed that voxels are arranged in the first quadrant of the XY coordinate system.

FIG. 25 is a diagram showing a specific example of the opacity correction magnification ξ set in the opacity correction magnification setting unit 190 shown in FIG. 24, taking into account such coordinate axis definition and voxel image arrangement. FIG. 25(a) shows one voxel plane parallel to the XY plane. is taken. As described above, each voxel on this voxel plane is arranged in the first quadrant, and its coordinate values x and y are both positive values.

The black circle shown in the center of this voxel plane indicates the cross section of the reference through line U. As shown in FIG. This reference penetration line U is a straight line parallel to the Z-axis that penetrates a plurality of voxel planes forming the voxel image 20 . In the illustrated example, the reference penetration line U is a straight line passing through the center point of the voxel plane having the horizontal size Sx and the vertical size Sy, but it can be defined at any position according to the organ to be observed. can do. In the voxel plane shown in FIG. 25(a), a large number of concentric ellipses W are defined with the penetrating position of the reference penetrating line U as the center point. In the embodiment shown here, a predetermined opacity correction magnification ξ is set based on this concentric ellipse W. Such a setting is optimal when a specific organ is to be observed based on the tomographic image group 10 of the human body.

For example, to voxels close to the i-th concentric ellipse Wi from the inside, the i-th opacity correction magnification ξi associated with the concentric ellipse Wi is applied, and the opacity α given to the voxel is Then, correction using the opacity correction magnification ξi may be performed. Then, the opacity α can be corrected according to the position of each voxel in the three-dimensional space, and the observation target can be limited according to the position in the three-dimensional space. Here, the reason why the opacity correction magnification ξ is set using a concentric ellipse instead of a concentric circle is that the sliced image of the human body becomes an image based on the ellipse as shown in FIG. 1(b). .

However, the opacity correction magnification setting unit 190 does not necessarily have to use a concentric ellipse to set the opacity correction magnification ξ, and sets the opacity by separately considering the distance in the X-axis direction and the distance in the Y-axis direction. It is sufficient that the correction magnification ξ can be set. That is, after defining a reference penetration line U that penetrates a plurality of voxel planes and is parallel to the Z-axis at a predetermined position, for each voxel, the distance between the voxel and the reference penetration line U in the X-axis direction and the Y-axis It suffices if the opacity correction magnification ξ can be set according to the distance in the direction.

For example, for a voxel placed at the position of point P(x, y) shown in FIG. Instead of setting the transparency correction magnification ξ, the distance between the point P (x, y) and the reference penetration line U in the X-axis direction and the distance between the point P (x, y) and the reference penetration line U in the Y-axis direction It is sufficient to set the opacity correction magnification ξ by separately considering the distance and . In the case of the illustrated example, setting weighting the distance in the Y-axis direction enables setting close to setting using concentric ellipses.

Practically, it is preferable to set using concentric ellipses. That is, the opacity correction magnification setting unit 190 sets a long axis parallel to the X-axis and a short axis parallel to the Y-axis on each voxel plane with a predetermined vertical-to-horizontal ratio and a central point at the intersection with the reference penetration line U. A plurality of concentric ellipses W having axes may be defined, and the same opacity correction magnification ξ may be set for voxels located on the same ellipse. At this time, a larger opacity correction magnification ξ is set for an ellipse closer to the center point. Then, a voxel located closer to the center point (reference penetration line U) is given a higher opacity α, and an organ to be observed located near the center point is selectively selected. be able to display.

If the opacity correction magnification setting unit 190 sets the opacity correction magnification ξ(x, y) for an arbitrary point P(x, y) on the voxel plane, the voxel image generation unit 175 When giving opacity α to voxel V(x, y, z), correction using opacity correction magnification ξ(x, y) set for point P(x, y) can be performed. In the case of the embodiment shown here, as shown in FIG. 25(a), the opacity correction magnification ξ(x, y) is the point P(x, y) on any voxel plane parallel to the XY plane. , the opacity correction magnification for voxel V(x, y, z) is always ξ(x, y) regardless of its coordinate value z.

Magnification correction using the opacity correction magnification ξ(x, y) performed in the voxel image generation unit 175 is as shown in FIG.
α new (x, y) = ξ (x, y) · α old (x, y) Equation (25)
It can be done by the following formula. Here, α old (x, y) is the old opacity before magnification correction, that is, the opacity corresponding to the signal value S (x, y, z) obtained by referring to the corrected opacity curve OCC. , α new (x, y) is the new opacity after magnification correction. The new opacity α new (x, y) is obtained by multiplying the old opacity α old (x, y) by the opacity correction magnification ξ(x, y).

The setting of the opacity correction magnification ξ(x, y) using the concentric ellipse described above is actually as shown in FIG.
ξ(x, y)=1−k・r(x, y)/R formula (26)
It can be done with the following formula.

where ξ(x,y) is the opacity correction factor to be applied to the voxel located at point P(x,y) shown in FIG. It is a weighted distance considering the vertical and horizontal weights of the line U and the point P(x, y). This weighted distance r(x, y) is
r (x, y) = ((Sx/2-x)/n) ² + (Sy/2-y) ² Formula (27)
is defined by the expression

In the example shown in FIG. 25(a), the reference penetration line U is a straight line passing through the center point of the voxel plane having the horizontal size Sx and the vertical size Sy. indicates the X-coordinate value of the reference penetration line U, and "Sy/2" indicates the Y-coordinate value of the reference penetration line U. n is the major axis ratio of the concentric ellipse W and is given by the length in the X-axis direction/the length in the Y-axis direction. The concentric ellipse W drawn in FIG. 25(a) is an example where n=2 and the length of the major axis along the X axis is twice the length of the minor axis along the Y axis. It has become. The first term on the right side of equation (27) contains n as a divisor because the distance between the point P(x, y) and the reference penetration line U in the X-axis direction is greater than that of the point P(x, y). This is because the distance in the Y-axis direction between U and the reference penetration line U is weighted twice.

Thus, r(x, y) defined by Equation (27) is not a simple geometric distance, but a weighted distance obtained by weighting the distance in the Y-axis direction by n times.

Also, R in formula (26) is a weighted reference distance,
R=((Sx/2)/n) ² +(Sy/2) ² Equation (28)
is defined by the expression Since "Sx/2" in this equation (28) indicates the X coordinate value of the reference penetration line U, and "Sy/2" indicates the Y coordinate value of the reference penetration line U, the weighted reference distance is calculated as shown in FIG. ) is the weighted distance between the reference through line U and the origin O shown in FIG.

As a result, the term "r(x, y)/R" in equation (26) is the weighted distance (weighted reference penetration line U ), and the farther the point P(x, y) is from the reference penetration line U, the larger the value of "r(x, y)/R". The opacity correction magnification ξ(x, y) defined by Equation (26) is calculated by subtracting from 1 the value obtained by multiplying this "r(x, y)/R" by the coefficient k. Here, k is an attenuation coefficient, and when k is set large, the degree of attenuation of the value of the opacity correction magnification ξ(x, y) increases according to the distance.

If the opacity correction magnification ξ(x, y) is defined by equation (26), ξ(x, y)<0 may occur. If so, set ξ(x,y)=0. Then ξ(x, y) will always be in the range of 0≦ξ(x, y)≦1. The value of the attenuation coefficient k can be set arbitrarily. When set, skin, muscle, fat on the body surface, and parts of the spine and ribs became invisible. Using equation (26), the opacity correction magnification ξ(x, y) can be set to attenuate in inverse proportion to the square of the length of the major axis or minor axis of the ellipse. With such a setting, a very favorable result was obtained when displaying a specific organ as an observation target based on the tomographic image group 10 of the human body as an imaging target.

As described above, as exemplified in equation (26), the value ξ(x, y) given as a function of the distance in the X-axis direction and the Y-axis direction from the reference penetration line U parallel to the Z-axis is expressed as the voxel V(x , y, z) as the opacity correction scaling factor ξ. However, since it is sufficient if the correction magnification ξ can be defined as a value corresponding to the position of the voxel V, it can be defined by various other methods. For example, the correction magnification ξ can be changed according to the position of the voxel V in the Z-axis direction. Specifically, in the above-described embodiment, a correction magnification ξ of ξ(x, y, z) may be defined in consideration of the position of the voxel V(x, y, z) in the Z-axis direction. good. Alternatively, based on the signal value S of each pixel S (x, y, z) constituting the tomographic image group, the shape of a predetermined object in each tomographic image is estimated, and the object having the estimated shape It is also possible to set a smaller correction magnification ξ for the outside of the object than for the inside.

<5.2 Independent Embodiment>
The volume rendering device 100' shown in FIG. After converting the signal value S into the opacity α using the corrected opacity curve OCC, the obtained opacity α is subjected to magnification correction using the correction magnification ξ. However, the magnification correction using this correction magnification ξ does not necessarily need to be used in combination with the color map optimization device 101, and can be used alone.

For example, in the volume rendering device 100' shown in FIG. The resulting color map C) is given to the voxel image generator 175, and the signal values S (x, y, z) of each pixel constituting the tomographic image group 10 given from the tomographic imaging apparatus 50 are converted into After conversion to opacity α using the opacity curve OC, a magnification using the opacity correction magnification ξ(x, y) set by the opacity correction magnification setting unit 190 for the converted opacity α A voxel image 20 can also be created with corrections.

As described above, the function of magnification correction using the opacity correction magnification ξ(x, y) generates the volume rendering image 30 based on the tomographic image group 10 captured by the tomographic imaging apparatus 50 and a predetermined color map. It can be widely used in a volume rendering apparatus for tomographic image display having a function of

Components (i) to (v) below are essential for such a volume rendering apparatus for displaying tomographic images.
(i) A tomographic image input unit for inputting a plurality of tomographic images obtained by tomography of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value S.
(ii) including an opacity curve OC in which opacity α is associated with each signal value S within a predetermined range, and a color palette CP in which predetermined color values (R, G, B) are associated with each other; A color map provider that provides color map C
(iii) an opacity correction magnification setting unit that sets a predetermined opacity correction magnification ξ according to the position in the three-dimensional space;
(iv) The signal value S given to each pixel of the plurality of tomographic images input by the tomographic image input unit is converted to the color map C provided by the color map providing unit and the opacity set in the opacity correction magnification setting unit. Using the correction magnification ξ, the color values (R, G, B) and the voxel values (R, G, B, α) indicating the opacity α are converted into voxel values (R, G, B, α). A voxel image creation unit that creates a voxel image consisting of a set of voxels V (x, y, z) arranged in a three-dimensional space by assigning the voxel values to the voxels.
(v) a rendering processing unit that generates a volume rendering image showing a state in which the voxel image is observed from a predetermined line-of-sight direction as a two-dimensional image;

Here, when the voxel image creation unit assigns voxel values (R, G, B, α) to individual voxels V (x, y, z), the opacity αold indicated by the color map C is added to the corresponding Correction is performed by multiplying the opacity correction magnification ξ that is set according to the position of the voxel, and the voxel value including the post-correction opacity αnew is given to the voxel.

More specifically, in the case of the above-described embodiment, in the above-described tomographic image display volume rendering device, the voxel image creating unit arranges a plurality of tomographic images so as to be parallel to the XY plane, and converts them into XYZ three-dimensional orthogonal A voxel image composed of a set of voxels arranged on a coordinate system is created, and an opacity correction magnification setting unit defines a reference penetration line U parallel to the Z-axis penetrating a plurality of tomographic images at a predetermined position. , for each voxel V(x, y, z), set the opacity correction magnification ξ(x, y) according to the distance in the X-axis direction and the distance in the Y-axis direction between the voxel and the reference penetration line U You should do it like this.

Practically, it is preferable to set .xi.(x, y) using a concentric ellipse W as shown in FIG. 25(a). That is, the opacity correction magnification setting unit sets the intersection point with the reference penetration line U having a predetermined aspect ratio on each tomographic image as the center point, and sets the long axis parallel to the X axis and the short axis parallel to the Y axis. is defined, the voxels V(x, y, z) located on the same ellipse W are set to the same opacity correction magnification ξ(x, y), and the center point A larger opacity correction magnification ξ(x, y) should be set for an ellipse closer to .

As described above, according to the experiment conducted by the inventor of the present application, the opacity correction magnification ξ(x, y) is set so as to attenuate in inverse proportion to the square of the length of the major axis or the minor axis of the ellipse. As a result, very favorable results were obtained when a specific organ was displayed as an observation target based on a group of tomographic images of the human body as an imaging target. Of course, such a volume rendering apparatus for displaying tomographic images can actually be configured by installing a dedicated program in a computer.

<<< §6. Gradation conversion processing >>>
6 and 24, the tomographic image input unit 110 in the apparatus shown in FIG. 6 and FIG. input as a collection of z). A general tomographic imaging apparatus 50 currently in use can output a tomographic image composed of a set of pixels having 16-bit signal values. Therefore, the explanation of each process in the embodiments described so far has been based on the premise of a tomographic image having this 16-bit signal value.

In practice, however, in cases where it is necessary to reduce the computational load, it may be preferable to convert the 16-bit signal value into, for example, an 8-bit signal value. In such a case, the tomographic image input unit 110 may be provided with a gradation conversion processing function for reducing the gradation of the input gradation image. By doing so, the gradation image having 16-bit signal values given from the tomography apparatus 50 is converted into a gradation image having 8-bit signal values, and the signal value histogram generating unit 120 and the voxel image generating unit 120 convert the gradation image having 8-bit signal values. Since it can be given to the units 170 and 175, it is possible to perform color map optimization processing and voxel image creation processing using a gradation image having an 8-bit signal value.

A specific example of tone conversion processing for converting a 16-bit signal value into an 8-bit signal value will be described below. This gradation conversion process is a process in which the tomographic image input unit 110 inputs a plurality of tomographic images as monochrome images each having a 16-bit signal value, and then converts these to monochrome images having 8-bit signal values. In other words, the signal value histogram generator 120 and the voxel image generators 170 and 175 are supplied with a monochrome image having 8-bit signal values.

FIG. 26 is a diagram showing a basic procedure of gradation conversion processing performed by the gradation image input unit 110 shown in FIGS. 6 and 24 as necessary. Here, the 16-bit signal value of each pixel constituting the group of tomographic images 10 captured by the tomographic imaging apparatus 50 is represented by the symbol SS, and the 8-bit signal value obtained by converting the gradation is represented by the symbol SS. It will be represented by S.

The upper part of FIG. 26A shows the signal width of the 16-bit signal value SS input by the gradation image input unit 110 . As shown, the 16-bit signal value SS typically ranges from -32768 to +32767 with a sign. In the gradation conversion process, first, the minimum value Dmin and the maximum value Dmax of the 16-bit signal values SS assigned to the pixels forming the input tomographic image are obtained. The obtained minimum value Dmin and maximum value Dmax are exemplified in the middle part of FIG. 26(a).

When obtaining the minimum value Dmin and the maximum value Dmax, pixels included in all of the plurality of input tomographic images can be used as a population. Pixels included in a part of the image are used as the population. Specifically, as illustrated in FIG. 2, one tomographic image (Z coordinate value (Sz/2) ) is extracted as an intermediate tomographic image, the signal values SS of all pixels included in this intermediate tomographic image are used as a population, and the minimum value Dmin and the maximum value Dmax are obtained from the population.

Generally, since the original image data of the tomographic image group 10 having 16-bit signal values SS is a large amount of data, a large-capacity memory is required to hold the data in memory at once. As described above, if the method of determining the minimum value Dmin and the maximum value Dmax using only a part of the tomographic images as a population is adopted, the memory capacity required for holding the original image data can be saved, and the gradation can be reduced. The advantage is that the image data after conversion can be built directly in the memory. In the case of the above example, the minimum value Dmin and the maximum value Dmax are obtained using only the pixels included in one intermediate tomographic image as a population. Since the intermediate tomographic image obtained includes a representative part to be imaged, there is no practical problem.

Subsequently, a lower limit value Lmin and an upper limit value Lmax are determined using a predetermined contrast adjustment coefficient β. Specifically, as shown in FIG. 26(b),
Lmin = Dmin + (Dmax - Dmin) β Formula (29)
Lmax = Dmin + (Dmax - Dmin) (1-β) Equation (30)
The lower limit value Lmin and the upper limit value Lmax are obtained from the following equation. The contrast adjustment coefficient β may be set to any value that satisfies 0<β<0.3. In the example shown here, β is set to 0.1. The lower limit of FIG. 26(a) shows the lower limit Lmin and the upper limit Lmax thus set. As shown, the lower limit Lmin is set to a value slightly larger than the minimum value Dmin, and the upper limit Lmax is set to a value slightly smaller than the maximum value Dmax.

Finally, the 8-bit signal value S is obtained from the 16-bit signal value SS by linearly converting the range from the lower limit value Lmin to the upper limit value Lmax to 0-255. Specifically, the 16-bit signal value SS (x, y, z) of the pixel located at the coordinate (x, y, z) is converted into an 8-bit signal value S (x, y, z). As shown in FIG. 26(b), the conversion formula for
S(x,y,z)=(SS(x,y,z)−Lmin)/
(Lmax-Lmin)×255 Equation (31)
can be defined by the formula

However, the signal value SS less than the lower limit Lmin is converted to 0, and the signal value SS exceeding the upper limit Lmax is converted to 255. In other words, when S(x, y, z)>255 as a result of the calculation according to the above formula (31), S(x, y, z)=255 and S(x, y, z)<0, then S(x, y, z)=0. As a result, in the chart shown in the lower part of FIG. 26(a), the signal value SS belonging to the area A3 is converted into a signal value S of 0 to 255, and the signal value SS belonging to the areas A1 and A2 is a signal of 0. The signal value SS converted to the value S and belonging to the areas A4 and A5 will be converted to a signal value S of 255. Thus, a monochrome image with 8-bit signal value S is generated based on a monochrome image with 16-bit signal value SS.

It should be noted that when such tone conversion processing is performed, the appearance frequency of pixels having a signal value S=0 and pixels having a signal value S=255 increases. As a result, for example, in the visual field histogram VH shown in FIG. 11(a), unexpected peaks occur at the positions of signal value S=0 and signal value S=255. Therefore, as already described, when searching for the peak position Svh exhibiting the maximum value VHmax in the field of view histogram VH, instead of using the entire range of the signal value S as the search range, the range excluding the neighboring portions on the left and right ends is used. is preferably used as the search range.

<<< §7. Understanding as a method invention >>>
So far, the present invention has been described as a device invention, such as a color map optimization device for tomographic image display or a volume rendering device for tomographic image display. Here, the present invention will be described as a method invention.

(1) Method of optimizing a color map for displaying tomographic images The present invention optimizes a color map used to generate a volume rendering image based on a group of tomographic images captured by a tomographic imaging apparatus. It can also be regarded as a method of optimizing a color map for tomographic image display. This optimization method consists of the following computer-implemented steps:

(i) A tomographic image input step of inputting a plurality of tomographic images obtained by tomography of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value.
(ii) Inputting a color map including an opacity curve in which opacity is associated with each signal value within a predetermined range and a color palette in which predetermined color values are associated with each other as a target for optimization processing. colormap input stage.
(iii) A signal value histogram creating step of creating a signal value histogram indicating the appearance frequency of each signal value for each pixel constituting a plurality of tomographic images.
(iv) A visual field histogram creation step of creating a visual field histogram by weighting the signal value histogram with the opacity curve.
(v) a correction color map creation step of correcting the opacity curve to create a correction opacity curve, and creating a correction color map including the correction opacity curve instead of the opacity curve;

Here, the feature of the present invention is that, in the opacity curve correction stage, at least the opacity curve is corrected using the field of view histogram.

By using this method for optimizing a color map for tomographic image display, it is possible to manufacture a computer-readable information recording medium on which data for a color map for tomographic image display are recorded. In this manufacturing method, in addition to the steps (i) to (v) constituting the optimization method described above, the computer stores the correction color map data created in the correction color map creating step on an information recording medium (for example, , a magnetic information recording medium such as a hard disk, an optical information recording medium, etc.).

(2) Volume rendering method for displaying tomographic images (Part 1)
The embodiment of the present invention described in Section 4 can also be regarded as a volume rendering method for tomographic image display for generating a volume rendering image based on a group of tomographic images captured by a tomographic imaging apparatus. This volume rendering method comprises the steps (i) to (v) of the method for optimizing a color map for tomographic image display described above, and further adding the following steps to be executed by a computer.

(vi) Using the correction color map created in the correction color map creation step in the optimization method, the signal value given to each pixel of the plurality of tomographic images input in the tomographic image input step in the optimization method , to a voxel value indicating a predetermined color value and opacity, and assigning the above voxel value to each voxel defined at each pixel position of the tomographic image, thereby creating a collection of voxels arranged in a three-dimensional space. A voxel image creation stage that creates a voxel image consisting of:
(vii) A rendering processing step of generating a volume rendering image showing a state of observing the voxel image from a predetermined viewing direction as a two-dimensional image.

By using this volume rendering method for tomographic image display, it is possible to manufacture a computer-readable information recording medium on which volume rendering image data is recorded. In this manufacturing method, in addition to the steps (i) to (vii) constituting the volume rendering method, the computer stores data of the volume rendering image created in the rendering processing step in an information recording medium (for example, a hard disk). It can be configured by adding a volume rendering image recording step for recording on a magnetic information recording medium such as a magnetic information recording medium or an optical information recording medium.

(3) Volume rendering method for tomographic image display (Part 2)
An additional embodiment of the present invention described in Section 5 can also be regarded as a volume rendering method for tomographic image display for generating a volume rendering image based on a group of tomographic images captured by a tomographic imaging apparatus. This volume rendering method consists of the following computer-implemented steps:

(i) A tomographic image input step of inputting a plurality of tomographic images obtained by tomography of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value.
(ii) A color map preparation stage of preparing a color map including an opacity curve in which opacities are associated with signal values within a predetermined range, and a color palette in which predetermined color values are associated.
(iii) an opacity correction magnification setting step for setting predetermined opacity correction magnifications according to positions in the three-dimensional space;
(iv) The signal value assigned to each pixel of multiple tomographic images input in the tomographic image input stage is converted to the color map prepared in the color map preparation stage and the opacity correction magnification set in the opacity correction magnification setting stage. is used to convert to a voxel value indicating a predetermined color value and opacity, and the voxel value is given to each voxel defined at each pixel position of the tomographic image, thereby obtaining voxel values arranged in a three-dimensional space. A voxel image creation stage for creating a voxel image consisting of aggregates.
(v) A rendering processing step of generating a volume rendering image showing a state of observing the voxel image from a predetermined viewing direction as a two-dimensional image.

Here, in the voxel image creation stage, when giving voxel values to individual voxels, correction is performed by multiplying the opacity indicated by the color map by the opacity correction magnification set according to the position of the voxel. , the voxel value including the corrected opacity is given to the voxel.

By using this volume rendering method for tomographic image display, it is possible to manufacture a computer-readable information recording medium on which volume rendering image data is recorded. In this manufacturing method, in addition to the steps (i) to (v) constituting the volume rendering method, the computer stores data of the volume rendering image created in the rendering processing step in an information recording medium (for example, a hard disk). It can be configured by adding a volume rendering image recording step for recording on a magnetic information recording medium such as a magnetic information recording medium or an optical information recording medium.

<<< §8. Display example of volume rendering image ＞＞＞
Finally, some display examples of volume rendering images created using the volume rendering apparatus for tomographic image display according to the present invention will be presented. 27 to 30 are diagrams showing specific examples of volume rendering images generated by the volume rendering apparatuses 100, 100' shown in FIG. 6 or 24. FIG.

Figure 27(a) is an example of correction by the fitting method described in §3.1.1, and Figure 27(b) is an example of correction by the flattening method described in §3.1.2. This is an example of what I did. In both cases, the color map input unit 150 inputs the color map C for lung observation and creates the voxel image using the corrected color map CC obtained by correcting this. Regarding this example, the correction by the fitting method is more favorable than the correction by the flattening method in terms of generating a display image of the lung as an observation target.

FIG. 28(a) is an example in which the uniform exponentiation correction described in §3.2.1 is added to the example shown in FIG. 27(a), and FIG. 28(b) is an example of FIG. 2 is an example in which the S-curve exponentiation correction described in §3.2.2 is added to the example shown in FIG. Due to the limitation of the resolution of the application drawing, it is not possible to fully grasp the difference between the result shown in FIG. 27 and the result shown in FIG. 28 from the photograph on the drawing, but in the result shown in FIG. Therefore, the object of observation is expressed more clearly.

FIGS. 29(a) and 29(b) show the examples shown in FIGS. 28(a) and 28(b), respectively, in addition to the opacity magnification correction (unnecessary) described as an additional embodiment in §5.1. This is an example in which a correction that is multiplied by a transparency correction magnification ξ is added. This opacity magnification correction is correction using the opacity correction magnification ξ set based on the concentric ellipse W as shown in FIG. can be applied. Therefore, even if there is another object having a signal value distribution similar to that of the lungs, if the other object is located away from the lungs, it will be excluded from the displayed image.

Comparing FIGS. 28(a) and 28(b) with FIGS. 29(a) and 29(b), it can be seen that most of the bodies other than the lungs mixed in the former are excluded in the latter. It can be seen that For example, the patient fixation belt that appeared in FIGS. 28(a) and 28(b) is completely removed in FIGS. 29(a) and 29(b). In this way, the opacity magnification correction is very effective in solving the blurring of the observed object due to the mixture of different objects having proximity signal values.

In both FIGS. 30(a) and 30(b), the color map input unit 150 inputs the color map C for heart observation and corrects it to obtain voxel images using the corrected color map CC. was created. FIG. 30(a) is an example of correction by the fitting method, and FIG. 30(b) is an example of correction by the flattening method, S-curve power correction, and opacity magnification correction. When the color map C for cardiac observation is used, the signal value distribution of the heart and the signal value distribution of the bones are close to each other, resulting in an image in which the bones are mixed. ) excludes some ribs, making the image more suitable for viewing the heart.

10: Group of tomographic images 11 to 18: Individual tomographic images 11v to 18v: Voxel planes of voxel images 20: Voxel image 25: Voxel image after coordinate transformation 30: Volume rendering image 31: Volume rendering suitable for lung observation Image 32: Volume rendering image suitable for observing the heart 50: Tomographic imaging device (CT, MRI, PET, etc.)
60: Display device 100, 100': Volume rendering device 101: Color map optimization device 110: Tomographic image input unit 120: Signal value histogram creation unit 130: View histogram creation unit 140: Opacity curve correction unit 141: Consider view histogram Correction unit 142: power correction unit 150: color map input unit 161: color map storage unit 162: color map creation unit 170: voxel image creation unit 175: voxel image creation unit 180: rendering processing unit 190: opacity correction magnification setting unit Ab: Ambient light component ratio A1 to A5: Gradation value area B: Blue pixel value Bnew: Blue pixel value after shadow calculation Bold: Blue pixel value before shadow calculation C: Color map CC: Correction color map CP: Color palette D: Deepest point Dmax: Maximum value of pixels in the intermediate tomographic image Dmin: Minimum value of pixels in the intermediate tomographic image E (x, y, z): Brightness value F: Projection point G: Green pixel value/ Magnitude of gradient vector Gnew: Green pixel value after shadow calculation Gold: Green pixel value before shadow calculation Gx, Gy, Gz: Coordinate axis components of gradient vector H: Signal value histogram h: Frequency of appearance of signal value S J0, J1, J2: Corresponding points k: Attenuation coefficient L: Parallel illumination light L (Lx, Ly, Lz): Light source vector Lmax: Upper limit of compression process Lmin: Lower limit of compression process M: Mapping plane n: Ellipse Long and short axis ratio O: Origin of XYZ three-dimensional orthogonal coordinate system OC: Opacity curve OCmax: Peak value (maximum value) of opacity curve OC
OC(S): Value of opacity α corresponding to signal value S OCC: Corrected opacity curve P: One point on two-dimensional image P1, P2: Projection plane Q1, Q2, Q3: Specified point R: Red pixel value / Weighted reference distance Rnew: Red pixel value after shadow calculation Rold: Red pixel value before shadow calculation Rxy: Resolution in X-axis direction and Y-axis direction Rz: Resolution in Z-axis direction r(x, y): Reference Weighted distances S, S1 to S2, Sa, Sb from the penetration line U: Signal values S1 to S10: Each step in the flowchart Smax: Maximum signal value Smin: Minimum signal value Soc: Peak position signal of opacity curve OC Value Svh: Peak position signal value of field of view histogram VH Svhc: Peak position signal value of corrected field of view histogram VHC S (x, y, z): Tomographic image pixel/pixel signal value SS (x, y, z): Pixel Signal values before compression Sx, Sy, Sz: Size of tomographic image s1, s2, s3: Signal value T: Threshold value U: Reference penetration line V (x, y, z): Voxel/voxel value of voxel image VH: Visibility histogram VHmax: Peak value (maximum value) of visibility histogram VH
VH(S): Value of field of view histogram corresponding to signal value S VHC: Corrected field of view histogram W: Concentric ellipse X: Coordinate axis of XYZ three-dimensional orthogonal coordinate system Y: Coordinate axis of XYZ three-dimensional orthogonal coordinate system Z: XYZ three-dimensional orthogonal Coordinate axes Z(0), Z(+1), Z(-1) of coordinate system: Voxel planes α, α1, α2, α3 of voxel image: Opacity αnew: New opacity after correction αold: Old before correction Opacity β: Contrast adjustment coefficient γ: Correction coefficient ξ(x, y): Correction magnification of opacity α at the position of point P(x, y)

Claims (33)

A device for optimizing a color map used for generating a volume rendering image based on a group of tomographic images captured by a tomographic imaging device,
a tomographic image input unit for inputting a plurality of tomographic images obtained by performing tomographic imaging of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value;
A color map that inputs a color map that includes an opacity curve that associates opacity with each signal value within a predetermined range and a color palette that associates a predetermined color value with each other as a target for optimization processing. an input unit;
a signal value histogram creation unit for creating a signal value histogram indicating the frequency of appearance of each signal value for each pixel constituting the plurality of tomographic images;
a field of view histogram creation unit that creates a field of view histogram by weighting the signal value histogram with the opacity curve;
an opacity curve correction unit that corrects the opacity curve to create a corrected opacity curve and outputs a corrected color map containing the corrected opacity curve instead of the opacity curve;
with
A tomographic image display color map optimizing apparatus, wherein the opacity curve correction unit includes at least a field histogram consideration correction unit that performs correction on the opacity curve using the field histogram. In the tomographic image display color map optimization device according to claim 1,
The field-of-view histogram consideration correction unit weights the signal value histogram with the corrected opacity curve compared to the peak position of the field-of-view histogram obtained by weighting the signal value histogram with the opacity curve. A tomographic image display color map optimization apparatus characterized in that the opacity curve is corrected so that the peak position of the corrected field of view histogram obtained by performing the above is closer to the peak position of the opacity curve. . In the tomographic image display color map optimization device according to claim 2,
Optimizing a color map for displaying a tomographic image, wherein the field histogram consideration correcting unit corrects the opacity curve so that the peak position of the corrected field histogram matches the peak position of the opacity curve. Device. In the tomographic image display color map optimization device according to claim 3,
The field of view histogram consideration correcting unit defines the maximum value of the field of view histogram within a predetermined signal value range as VHmax, the signal value giving the maximum value VHmax as Svh, and the maximum value of the opacity curve within the predetermined signal value range. is OCmax, the signal value giving the maximum value OCmax is Soc, and the opacity of the opacity curve at the signal value Svh is OC(Svh), the value of the corrected visual field histogram at the signal value Soc is VHmax. and correcting the opacity curve so that the value of the corrected visual field histogram at the signal value Svh becomes VHmax.OC(Svh)/OCmax. In the tomographic image display color map optimization device according to claim 4,
The field-of-view histogram consideration correcting unit sets the opacity of the opacity curve at the signal value S to OC(S), the value of the field-of-view histogram at the signal value S to VH(S), and a correction coefficient γ( S)
γ(S)=(VHmax/OCmax
)・(OC(S)/VH(S))
The opacity OCC (S) of the corrected opacity curve at the signal value S is defined by the following formula:
OCC(S)=γ(S)・OC(S)
A tomographic image display color map optimizing apparatus characterized in that calculation is performed by the following equation. In the tomographic image display color map optimization device according to claim 1,
The field-of-view-histogram-considering correcting unit performs weighting on the signal value histogram with the opacity curve, compared to the shape near the peak of the field of view histogram obtained by weighting the signal value histogram with the corrected opacity curve. The corrected field of view at the peak position of the opacity curve is corrected so that the shape near the peak of the corrected field of view histogram obtained by weighting is flattened. The value of the histogram matches the value of the corrected field of view histogram at the peak position of the field of view histogram , and the value of the corrected field of view histogram between the peak position of the opacity curve and the peak position of the field of view histogram is A tomographic image display color map optimizing apparatus, wherein the opacity curve is corrected so as to match the value of the corrected field of view histogram at the peak position of the field of view histogram . In the tomographic image display color map optimization device according to claim 6,
The field of view histogram consideration correcting unit defines the maximum value of the field of view histogram within a predetermined signal value range as VHmax, the signal value giving the maximum value VHmax as Svh, and the maximum value of the opacity curve within the predetermined signal value range. is OCmax and the signal value giving the maximum value OCmax is Soc, the opacity A tomographic image display color map optimization apparatus characterized by correcting a curve. In the tomographic image display color map optimization device according to claim 7,
The field-of-view histogram consideration correcting unit sets the opacity of the opacity curve at the signal value S to OC(S), the value of the field-of-view histogram at the signal value S to VH(S), and a correction coefficient γ( S)
γ(S)=(VHmax/VH(S))
The opacity OCC (S) of the corrected opacity curve at the signal value S is defined by the following formula:
OCC(S)=γ(S)・OC(S)
A tomographic image display color map optimizing apparatus characterized in that calculation is performed by the following equation. In the tomographic image display color map optimization device according to any one of claims 1 to 8,
The tomographic image input unit inputs the plurality of tomographic images as monochrome images each having a 16-bit signal value, and then performs a gradation conversion process for converting these to monochrome images having an 8-bit signal value. Having a function of providing the subsequent tomographic image to the signal value histogram creation unit,
In the gradation conversion process, the minimum value Dmin and the maximum value Dmax of the signal values for some or all of the plurality of input tomographic images are obtained, and a predetermined contrast adjustment coefficient β (where 0<β<0. 3), the lower limit value Lmin and the upper limit value Lmax are obtained by the following equations: Lmin = Dmin + (Dmax - Dmin).β, Lmax = Dmin + (Dmax - Dmin).(1-β). ~ linearly transforming the range of the upper limit value Lmax to 0 to 255, converting signal values less than the lower limit value Lmin to 0, and converting signal values exceeding the upper limit value Lmax to 255, thereby converting the 8-bit A tomographic image display color map optimization apparatus characterized by generating a monochrome image having signal values. In the tomographic image display color map optimization device according to claim 4, 5, 7, or 8,
The field of view histogram consideration correcting unit determines the maximum value VHmax within a predetermined signal value range of the field of view histogram by adding 1 to the minimum non-zero signal value of the field of view histogram. A tomographic image display color map optimizing apparatus, wherein a range equal to or less than a value obtained by subtracting 1 from a maximum signal value having a value other than 0 in a histogram is defined as the predetermined signal value range. In the tomographic image display color map optimization device according to any one of claims 1 to 10,
The opacity curve correction unit corrects the opacity of part or all of the opacity curve corrected by the field histogram consideration correction unit to the power of a correction coefficient γ (where 0<γ<1). A color map optimization apparatus for tomographic image display, further comprising a power correction unit. In the tomographic image display color map optimization device according to any one of claims 1 to 10,
The opacity curve correction unit corrects the opacity of a part or all of the opacity curve corrected by the field histogram consideration correction unit by a correction coefficient γ ( However, it is characterized by further comprising a power correcting unit that performs correction to the power of 0<γ<1) and performs correction to the power of the reciprocal of the correction coefficient γ if the opacity is less than the threshold value T. A device for optimizing a color map for tomographic image display. A program for causing a computer to function as the tomographic image display color map optimizing device according to any one of claims 1 to 12. A volume rendering device comprising the tomographic image display color map optimizing device according to any one of claims 1 to 12,
In addition to the tomographic image input unit, the color map input unit, the signal value histogram creation unit, the field histogram creation unit, and the opacity curve correction unit, which are components of the tomographic image display color map optimization device,
Signal values given to each pixel of the plurality of tomographic images input by the tomographic image input unit are indicated as predetermined color values and opacities using the correction color map output from the opacity curve correction unit. A voxel image that creates a voxel image consisting of a set of voxels arranged in a three-dimensional space by converting the values into voxels and assigning the voxel values to each voxel defined at each pixel position of the tomographic image. creation department,
a rendering processing unit that generates a volume rendering image showing a state of observing the voxel image from a predetermined viewing direction as a two-dimensional image;
A volume rendering apparatus for tomographic image display, further comprising: In the tomographic image display volume rendering device according to claim 14,
further comprising a color map storage unit storing a plurality of pre-created standard color maps,
A volume rendering apparatus for tomographic image display, wherein the color map input unit inputs a predetermined standard color map from the color map storage unit in accordance with an operator's selection instruction. In the tomographic image display volume rendering device according to claim 14 or 15,
further comprising a color map creation unit that creates a new color map based on an operator's creation instruction,
A volume rendering apparatus for tomographic image display, wherein the color map input unit inputs the color map created by the color map creation unit. In the tomographic image display volume rendering device according to claim 16,
The color map creation unit inputs a plurality of representative signal values, an opacity corresponding to each of these representative signal values, and a color value corresponding to each of a plurality of signal value intervals as a creation instruction from the operator. Then, based on this creation instruction, the opacity curve is created by interpolating the opacity for continuous signal values within a predetermined range, and the color palette in which the color values are associated with the signal value intervals. A volume rendering apparatus for tomographic image display, characterized in that it creates a In the tomographic image display volume rendering device according to claim 17,
a first representative signal value, a second representative signal value, and a third representative signal value discretely arranged in ascending order, and an opacity corresponding to each of these representative signal values; is input as a creation instruction by the operator, and the signal value interval between the first representative signal value and the second representative signal value and the interval between the second representative signal value and the third representative signal value are input. linearly interpolate the opacity for each signal value interval between the third a volume rendering apparatus for tomographic image display, wherein the opacity curve is created by performing interpolation to give an opacity corresponding to the third representative signal value for a signal value interval exceeding the representative signal value of the third representative signal value. In the tomographic image display volume rendering device according to claim 17,
The color map creation unit uses the first representative signal value and the second representative signal value discretely arranged in ascending order and the opacity corresponding to each of these representative signal values as the operator's creation instruction. linearly interpolating opacity for a signal value interval between the first representative signal value and the second representative signal value; By performing interpolation that gives an opacity corresponding to one representative signal value, and performing interpolation that gives an opacity corresponding to the second representative signal value for a signal value section exceeding the second representative signal value, A volume rendering apparatus for tomographic image display, characterized by creating an opacity curve. In the tomographic image display volume rendering device according to any one of claims 14 to 19,
The rendering processing unit performs a smoothing process that replaces the opacity of each voxel that constitutes the voxel image with the average value of the opacities of the voxel and adjacent voxels adjacent to the voxel, and after the smoothing process A volume rendering apparatus for tomographic image display, wherein the volume rendering image is generated based on a voxel image. In the tomographic image display volume rendering device according to any one of claims 14 to 20,
The rendering processing unit obtains a gradient vector representing a three-dimensional gradient of opacity for each voxel that constitutes the voxel image, and calculates a luminance value considering the direction of the light source vector pointing in a predetermined direction and the gradient vector. is calculated, shading addition processing is performed to correct the color value of each voxel by multiplying the luminance value, and the volume rendering image is generated based on the voxel image after the shading addition processing. A volume rendering device for tomographic image display. In the tomographic image display volume rendering device according to claim 21,
The voxel image generators are arranged side by side at a predetermined X-axis direction pitch in the X-axis direction, at a predetermined Y-axis direction pitch in the Y-axis direction, and at a predetermined Z-axis direction pitch in the Z-axis direction. Create a voxel image consisting of a grid-like voxel aggregate,
The rendering processing unit includes an X-axis difference value indicating a difference in opacity between a pair of adjacent voxels arranged on both sides of a voxel of interest along the X-axis direction, and A Y-axis difference value indicating a difference in opacity between a pair of adjacent voxels arranged on both sides, and a Z-axis indicating a difference in opacity between a pair of adjacent voxels arranged on both sides of the voxel of interest along the Z-axis direction. The product of multiplying the X-axis difference value by a coefficient corresponding to the X-axis direction pitch is defined as an X-axis component, and the Y-axis difference value is a coefficient corresponding to the Y-axis direction pitch. A vector having the multiplied product as a component in the Y-axis direction and the product of the Z-axis difference value multiplied by a coefficient corresponding to the pitch in the Z-axis direction as a component in the Z-axis direction is used as the gradient vector for the voxel of interest. A volume rendering device for displaying tomographic images. In the tomographic image display volume rendering device according to any one of claims 14 to 22,
The voxel image creation unit creates a voxel image consisting of a voxel aggregate arranged on an XYZ three-dimensional orthogonal coordinate system,
When the rendering processing unit generates a volume rendering image on a predetermined projection plane, coordinate transformation is performed by rotating the XYZ three-dimensional orthogonal coordinate system so that the XY plane is parallel to the projection plane. After that, a virtual ray is irradiated from the projection point on the projection plane to the voxel image after the coordinate transformation in a direction parallel to the Z-axis, and based on the color value and opacity of each voxel on the virtual ray A volume rendering apparatus for displaying a tomographic image, wherein the volume rendering image is generated by giving the calculated color value as a pixel value of a pixel positioned at the projection point. In the tomographic image display volume rendering device according to any one of claims 14 to 22,
The voxel image creation unit creates a three-dimensional texture image consisting of a voxel aggregate arranged on a first XYZ three-dimensional orthogonal coordinate system,
The rendering processing unit defines a second XYZ three-dimensional orthogonal coordinate system such that the XY plane is parallel to the projection plane when generating a volume rendering image on a predetermined projection plane, and A plurality of mapping planes parallel to the XY plane are defined on the XYZ three-dimensional orthogonal coordinate system of No. 2 at predetermined intervals in the Z-axis direction, and the second XYZ three-dimensional The three-dimensional texture image is mapped onto the plurality of mapping planes by performing coordinate transformation to an orthogonal coordinate system, and the color values of projection points having XY coordinate values (x, y) on the projection plane are converted to the XY coordinate values (x, y) on the nearest mapping plane from the color value of the pixel mapped at the position of XY coordinate values (x, y) on the mapping plane farthest from the projection point A volume rendering apparatus for displaying a tomographic image, wherein the volume rendering image is generated by sequentially blending and determining up to the color value of pixels mapped to a position, taking into consideration the opacity of each pixel. In the tomographic image display volume rendering device according to any one of claims 14 to 24,
further comprising an opacity correction magnification setting unit that sets a predetermined opacity correction magnification according to the position in the three-dimensional space,
The voxel image creation unit performs correction by multiplying the opacity indicated by the correction color map by an opacity correction magnification set according to the position of the voxel when giving a voxel value to each voxel. 1. A volume rendering apparatus for displaying a tomographic image, wherein a voxel value including post-correction opacity is given to the voxel. In the tomographic image display volume rendering device according to claim 25,
The voxel image creation unit arranges the voxel planes corresponding to the respective tomographic images so as to be parallel to the XY plane, arranges the voxels in each voxel plane two-dimensionally, and arranges the voxels three-dimensionally on the XYZ three-dimensional orthogonal coordinate system. Create a voxel image consisting of the original arrayed voxel collection,
The opacity correction magnification setting unit defines a reference penetration line that penetrates the plurality of voxel planes and is parallel to the Z-axis at a predetermined position, and for each voxel, the X-axis direction between the voxel and the reference penetration line. A volume rendering apparatus for displaying a tomographic image, wherein an opacity correction magnification is set in accordance with a distance with respect to and a distance with respect to the Y-axis direction. In the tomographic image display volume rendering device according to claim 26,
The opacity correction magnification setting unit provides a plurality of concentric ellipses on the plurality of voxel planes, each having a center point at an intersection with the reference penetration line and having a major axis parallel to the X axis and a minor axis parallel to the Y axis. is defined, the same opacity correction magnification is set for voxels located on the same ellipse, and a larger opacity correction magnification is set for the ellipse closer to the center point. Volume rendering device. In the tomographic image display volume rendering device according to claim 27,
A volume rendering apparatus for tomographic image display, wherein the opacity correction magnification setting unit sets an opacity correction magnification that attenuates in inverse proportion to the square of the length of the long axis or the short axis of the ellipse. A program for causing a computer to function as the tomographic image display volume rendering apparatus according to any one of claims 14 to 28. A method of optimizing a color map used for generating a volume rendering image based on a group of tomographic images captured by a tomography apparatus, comprising:
A tomographic image input step in which a computer inputs a plurality of tomographic images obtained by tomography of a predetermined object at predetermined intervals as a two-dimensional array of pixels each assigned a predetermined signal value;
The computer selects a color map including an opacity curve in which opacities are associated with respective signal values within a predetermined range and a color palette in which predetermined color values are associated with each other as a target of optimization processing. an input color map input stage;
a signal value histogram creation step in which the computer creates a signal value histogram indicating the appearance frequency of each signal value for the signal values of each pixel constituting the plurality of tomographic images;
A visibility histogram creation step in which the computer creates a visibility histogram by weighting the signal value histogram with the opacity curve;
a correction color map creation step in which the computer corrects the opacity curve to create a correction opacity curve, and creates a correction color map including the correction opacity curve instead of the opacity curve;
has
A method for optimizing a color map for displaying a tomographic image, wherein at least the opacity curve is corrected using the field histogram in the corrected color map creation step. A manufacturing method for manufacturing a computer-readable information recording medium on which data of a tomographic image display color map is recorded using the method for optimizing a tomographic image display color map according to claim 30,
In addition to the steps constituting the optimization method, the computer further has a correction color map recording step of recording the correction color map data created in the correction color map creation step on an information recording medium. A method of manufacturing an information recording medium on which color map data for tomographic image display is recorded. A volume rendering method comprising steps constituting the method for optimizing a color map for displaying a tomographic image according to claim 30,
The computer uses the correction color map created in the correction color map creation step in the optimization method to convert the signal values assigned to each pixel of the plurality of tomographic images input in the tomographic image input step in the optimization method. to voxel values indicating a predetermined color value and opacity, and assigning the voxel values to each voxel defined at each pixel position of the tomographic image, thereby obtaining voxel values arranged in a three-dimensional space. a voxel image creation step of creating a voxel image consisting of aggregates;
a rendering processing step in which the computer generates a volume rendering image showing a state in which the voxel image is observed from a predetermined viewing direction as a two-dimensional image;
A volume rendering method for tomographic image display, further comprising: 33. A manufacturing method for manufacturing a computer-readable information recording medium on which the data of the volume rendering image is recorded by using the volume rendering method for displaying a tomographic image according to claim 32,
In addition to the steps constituting the volume rendering method, the computer further has a volume rendering image recording step of recording data of the volume rendering image generated in the rendering processing step on an information recording medium. A method for manufacturing an information recording medium on which data of a volume rendering image for tomographic image display is recorded.

Images