JP3326154B2 - Outline extraction method, image forming method, and image forming apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to the contour extraction method of enabling contour extraction and image formation of 3-dimensional objects with less A Chifakuto <br/> (擬像), an image forming method and an image forming apparatus.

[0002]

2. Description of the Related Art In recent years, with the development of image processing technology using a computer, a technology for visualizing the inside of a three-dimensional object has attracted attention. In particular, in the medical field, lesions can be found early by visualizing the inside of a living body.
CT (Computed Tomography) device or MRI (Mag
Medical diagnosis using a netic resonance imaging device is widely performed.

As a method for obtaining a three-dimensional image of the inside of an object, a method called volume rendering is known. In this volume rendering, an image is projected on a projection surface by irradiating light (ray) to a three-dimensional voxel (micro volume element) space. This operation is called ray casting. In this ray casting, sampling is performed at regular intervals along a ray path, and voxel values are obtained from voxels at each sampling point.

[0004] The image projected on the projection surface shows the result of the ray passing through each voxel being attenuated or reflected according to the opacity α corresponding to the voxel value unique to the voxel.
The voxel value is called a CT value in the case of a CT image.

The opacity α is a value indicating the transmittance of a ray, and is set in the range of 0.0 (transmission) to 1.0 (blocking), and is assigned corresponding to each voxel value. By changing the opacity α corresponding to this voxel value, a specific voxel in the voxel space becomes opaque,
An arbitrary region of the object can be extracted and visualized. For example, objects such as bones, organs, and blood vessels inside a living body each have a certain range of voxel values.
In the image, a value of 500 to 1000 is used for a bone, and a value of 0 to 500 is used for an organ as voxel values.

Accordingly, when an arbitrary object inside the living body is extracted and visualized, the opacity α corresponding to the voxel value is variably set, and the opacity corresponding to the voxel value of the target object is set to 1. By setting the value close to 0.0, the target object becomes opaque and is extracted.

[0007]

However, the conventional volume rendering has the following problems, and its improvement has been desired. That is, when extracting an arbitrary region inside the object, it is necessary to select a voxel value in a desired range from sampling data of consecutive voxel values.

FIG. 7 is a characteristic diagram showing a filter setting for obtaining the opacity α according to the voxel value V. In the figure, the vertical axis indicates the opacity α, and the horizontal axis indicates the voxel value V. When the voxel value of the object exists from V2 to V4 centering on V3 among consecutive voxel values, the opacity α2, α3 is selected by selecting the range of voxel values V2 to V4 using the filter of FIG. , Α4 to visualize the inside of the object.

FIG. 9 is a diagram showing a state in which ray casting is performed with the filter settings shown in FIG. Ray 1 detects opacity α2, α3, α4 indicating that it is opaque,
Ray 2 detects opacity α1, α2, α4, and ray 3 detects opacity α2, α3.

[0010] Since discrete sampling is performed in ray casting, if the sampling interval is widened or the voxel value or opacity changes rapidly, the nature of the data will not be correctly reflected in the projected image. When the inside of the object is visualized in such a state, a so-called artifact (pseudo image) occurs. In particular, when a living body is visualized, this artifact appears as striped bones and fragmentary capillaries.

FIG. 8 is a characteristic diagram showing the setting of a filter for obtaining only the opacity α near a specific voxel value V. This filter has a sharper shape than the filter of FIG. 7, and extracts only the opacity α near the voxel value V3. In the range of the selected voxel values V2 to V3,
There is only one opacity α3 corresponding to voxel value V3.

FIG. 10 is a diagram showing a state in which ray casting is performed with the filter settings shown in FIG. Rays 1 and 3 detect an opacity α3 indicating opacity, but ray 2 does not detect an opacity indicating opacity.

Since voxel values of an object are inherently continuous, voxel values should exist between each raycast sampling. However, since the voxel values are discretely sampled, sampling is not performed on the area corresponding to the voxel value V3 in ray 2, and the ray casting result is α between ray 1 and ray 3.
It becomes a discontinuous value of 3, 0 and α3. This appears as an artifact.

Such an artifact can be eliminated by narrowing the sampling interval, that is, so-called oversampling. However, performing oversampling leads to an increase in the amount of data processing, and when trying to observe a three-dimensional image inside an object from various directions, it is difficult to follow this movement in real time.

Accordingly, the present invention has been made in view of the above circumstances, and when projecting an image inside an object, a contour extracting method and an image forming method capable of quickly obtaining an object-free outline of an object. Another object of the invention is to provide an image forming equipment.

[0016]

[Means for Solving the Problems]

The contour extraction method according to claim 1 in order to achieve the above object, for the data set of the obtained sampling value by discretely sampling the values of continuously distributed in three-dimensional space, Obtaining a data set of opacity assigned to each minute space obtained by dividing the three-dimensional space, and obtaining an opacity assigned to each adjacent minute space.
A contour extraction method for extracting a spatial continuity or a contour to be analyzed based on the continuity of degrees , wherein two points on a boundary of each of the minute spaces and a path connecting them are defined, and based on the length of the corresponding sampling values and path, the non
And it calculates the transparency.

The image forming method according to claim 2, for the data set of the obtained sampling value by discretely sampling the values of continuously distributed in three-dimensional space, obtained by dividing the three-dimensional space An image forming method for acquiring a data set of opacity assigned to each minute space and forming an image by ray casting based on the opacity assigned to each minute space, wherein a ray path in the ray casting is provided. The opacity assigned to each of the minute spaces is calculated based on the sampling values corresponding to at least two points in the minute space.

The image forming method according to claim 3, further wherein, based on the sampling value corresponding to the two points on the boundary of each minute space, which calculates the opacity claim 4 The image forming method of (1) calculates the opacity based on a distance between two points on a boundary of each of the minute spaces.

According to a fifth aspect of the present invention, the calculated opacity is stored as a table for the sampling values.

According to a sixth aspect of the present invention, in the image forming apparatus for forming an image by ray casting, a ray direction setting means for setting a ray direction in the ray casting; Micro space detecting means for detecting a micro space along the path of the ray along the path of the ray, and sampling value detecting means for detecting sampling values corresponding to two points on the path of the ray and on the boundary of the micro space, Setting means for setting a relationship between the detected sampling value and opacity; and a sampling value for the two points in advance based on the relationship between the set sampling value and the opacity, and 1 assigned to the minute space. Table creation means for creating a table in which the correspondence between the two opacity levels is described, and based on the created table, Opacity obtaining means for obtaining one opacity from sampling values corresponding to two points detected by the pulling value detecting means; image forming means for forming an image by the ray casting based on the obtained opacity; It is provided with.

According to a seventh aspect of the present invention , in the image forming apparatus, the table creating means may determine a relationship between the set sampling value and the opacity and a point between two points on a boundary of a minute space determined by a direction of the ray. The table is created based on the distance.

Hereinafter, the principle of the present invention will be described. First, a description regarding to claim 1. Figure 11 is a diagram representing the analysis target and minute space, 12 analysis typical in the prior art (not
FIG. 13 is a diagram showing the result of calculating the outline ( transparency) , and FIG. In order to recognize the continuity of the target object and analyze the contour (see FIG. 11), a data set of sampling values obtained by discretely sampling values continuously distributed in a three-dimensional space,
A data set of representative values for analysis assigned to each minute space obtained by dividing the three-dimensional space is acquired (see FIG. 12). In this case, if the representative values are derived on a one-to-one basis from the sampling values, artifacts (pseudo-images) may occur, or gaps may occur in places that should be continuous. Therefore, it is not possible to correctly obtain the contour and the spatial continuity of the analysis target (see FIG. 13). Therefore, two points on the boundary of each minute space and a path connecting them are defined, and a representative value for analysis is calculated based on the sampling value and the path length corresponding to the two points.

Next, claims 2 to 7 will be described. FIG. 14 is a diagram showing a relationship between voxel values and opacity based on a conventional method. FIG. 15 is a diagram showing the relationship between voxel values and opacity according to the present invention.

Let the coordinates P in the three-dimensional voxel space be (X, Y,
Z), and the position of an arbitrary voxel is represented by coordinates Pn. In the conventional method, as shown in FIG. 14, the opacity of a voxel is obtained as an opacity αn corresponding to the voxel value Vn at a position Pn, which is a representative point in the voxel, on a one-to-one basis.

On the other hand, in the present invention, as shown in FIG.
The opacity of the voxel is obtained as the opacity αmn corresponding to the voxel values Vm and Vn at two positions Pm and Pn on the boundary of the voxel in a two-to-one correspondence. still,
The opacity of a voxel may be obtained as opacity corresponding to many-to-one with voxel values at three or more positions.

When a projected image is obtained by irradiating a ray on the three-dimensional voxel space, the ray intensity R attenuates as Rout = Rin × (1-α) every time the ray passes through each voxel.

FIG. 16 is a diagram showing a procedure of ray casting. In ray casting in which sampling is performed at intervals A of each voxel (see FIG. 16), positions Pm to Pn
No change in voxel opacity is detected.
On the other hand, in ray casting in which the voxels at the positions Pm to Pn are sampled at intervals B (see FIG. 16) obtained by finely dividing the voxels into v, the voxel values Vm at the divided positions Pmnk are obtained.
k is calculated by an appropriate interpolation process, and the opacity αmnk that changes with respect to Vmk corresponds to k.

FIG. 18 is a graph showing the relationship between voxel values and opacity. FIG. 18 shows voxel values Vm to Vn.
The opacity αmnk that is continuous over is shown. As shown in FIG. 18, when the relationship between the voxel value Vmk and the opacity αmnk is set, the attenuation of the ray intensity from the positions Pm to Pn is calculated by the following equation.

[0030] _{R out = R in {(1} -α mn1) (1-α mn2) (1-α mn3) ... (1-α mnk)} 1 / k

[0031] Thus, the opacity of over the position Pm~Pn, 1 - {(1- α mn1) (1-α mn2) (1-α mn3) ... (1-α mnk)} 1 / k ... (1 ).

FIG. 17 is a diagram showing a procedure of ray casting according to the present invention. The present invention focuses on the fact that the calculation of the above equation (1) does not depend on the position Pn which is the representative point of the voxel, and performs the calculation of the above equation (1) in advance between the voxel values Vm and Vn existing in the two voxels. By performing this calculation and storing it in a look-up table (LUT), the position Pm,
This is to detect a change in opacity of a voxel between Pn.

As described above, according to the present invention, the opacity of the voxel between the positions Pm to Pn is not calculated by narrowing the sampling interval and calculating the product of the opacity of each sampling point every time, as in the related art. , Is immediately obtained from the relationship between the voxel value Vm and the voxel value Vn stored in the reference table in advance.

Further, the opacity of each voxel on the ray path is calculated from the voxel values of two points, an incident position Pm (start point) and an emission position Pn (end point), on the voxel boundary where the ray intersects each voxel, into a lookup table. Obtained based on Further, the opacity may be acquired in consideration of not only the displacement of the two voxel values on the boundary of each voxel, that is, the displacement of the two voxel values, but also the distance between the two points.

In the present invention, there is a case where a corresponding sampling value does not exist exactly. In this case, a value interpolated based on a nearby value (interpolated value) or a nearby value (neighboring value) is used. Used as sampling value. Therefore, the “sampling value” in the present invention includes an interpolation value or a nearby value.

[0036]

Embodiments of the present invention will be described below. The invention according to claim 1 is characterized in that a data set of sampled values obtained by discretely sampling values continuously distributed in a three-dimensional space,
This is applied to a case where a data set of analysis representative values (opacity) assigned to each minute space obtained by dividing the dimensional space is obtained, and data analysis or contour extraction is performed based on the representative values. Further, the invention according to claims 2 to 7 is:
The method is applied to a case where an image is formed on a projection plane by irradiating a ray onto a three-dimensional voxel space obtained by discretely sampling values distributed continuously in a three-dimensional space.

FIGS. 1 and 2 relate to claim 1. FIG. 1 is a diagram showing a result of calculating a representative value (opacity) for analysis according to the present invention. FIG. 2 is a diagram showing a result of contour extraction according to the present invention. FIG. Hereinafter, a procedure for calculating a representative value for analysis and a procedure for extracting a contour based on the representative value for analysis will be described.

First, a data set of sampling values obtained by discretely sampling values distributed continuously in a three-dimensional space is used for analysis data assigned to each minute space obtained by dividing the three-dimensional space. A representative value is defined as two points on the boundary of each of the microspaces and a path connecting them, and based on a sampling value and a path length corresponding to the two points,
calculate. For example, when performing volume rendering, opacity can be used.

Next, a threshold value S is set. When the assigned representative value is equal to or greater than the threshold value, “1” is associated with each minute space.
1 shows a case where the threshold value is set to “10.” As a result, the adjacent minute space V is set.
When the values of 1 and V2 are both "1", it is assumed that V1 and V2 are spatially continuous. Further, in an arbitrary minute space Vi, Vj having a value of “1”, when it is possible to reach Vj through an adjacent continuous minute space starting from Vi, Vi, Vj are spatially continuous. Consider it to be.

In the set of minute spaces that are spatially continuous, the set of minute spaces adjacent to the minute space having the value 0 is
Regarded as a contour. FIG. 2 shows the result of contour extraction based on the representative values calculated in FIG.

FIGS. 3 to 6 relate to claims 2 to 7 . FIG. 3 is a diagram showing a three-dimensional voxel space 100. As shown in FIG. 3, a three-dimensional voxel space 100 constructed by voxels V is obtained as input data for generating a three-dimensional image.

Each voxel is set with a voxel value having shading information. The voxel value is calculated based on the CT image
If it is an image, it is called a CT value. The CT value is a value representing the X-ray absorption rate, and as a moisture reference value, for example, each tissue inside a living body has a CT value in a predetermined range.

Next, based on the projection direction of the image to be generated,
Sets the ray direction for raycasting. In this ray casting, sampling is performed for each voxel in the three-dimensional voxel space 100 on the ray path.

Further, a relation between the voxel value and the opacity is set based on the voxel value of the tissue to be extracted. With respect to the sampled voxel, a voxel value corresponding to the voxel is obtained, and opacity indicating a ray transmittance for the obtained voxel value is obtained.

In ray casting, as a ray passes through each voxel in voxel space, the light passing through each voxel is attenuated and reflected according to the opacity of each voxel. For example, when a three-dimensional image is generated from a voxel space in which tomographic images of internal organs in a living body are stacked, an opacity close to “1.0” is assigned to a range of CT values of the internal organs. For voxels other than this internal organ, the opacity is "0"
The ray is attenuated and reflected only by the voxel corresponding to the viscera, thereby extracting the viscera region of the living body.

Further, the assignment of opacity will be described with reference to FIGS. As shown in FIG. 3, when a ray passes through the three-dimensional voxel space 100, the ray repeatedly enters and exits voxels on the path. FIG. 4 is a diagram showing a relationship between voxel values Vm and Vn and opacity αmn. As shown in FIG. 3, when the voxel values on the boundary with the voxel on the ray path are Vm and Vn,
The voxel value Vm is the start point, the voxel value Vn is the end point, and the opacity α between these voxel values is represented by the voxel values Vm and Vn.
Assign as a function of Two voxel values Vm, Vn
Is set based on the relationship between the voxel values and the opacity in FIGS. 16 and 18 described above, and is stored as a lookup table (LUT) shown in FIG. FIG. 5 is a table showing the relationship between the two voxel values Vm and Vn and the opacity αmn.

The opacity α ranges from 0 to 1.0,
At a value of 0, the ray is completely transmitted, and at a value of 1.0, the ray is completely blocked. In the reference table shown in FIG.
The voxel values Vm and Vn each have a value in the range of values Vmin to Vmax consisting of a predetermined number of steps, and the corresponding opacity α has a value corresponding to the square of the predetermined number of steps.

Since the conventional opacity assignment is based on one point in the voxel, the opacity change cannot be correctly followed depending on the sampling interval of ray casting, and the inherent properties of the data cannot be correctly projected. Could not be reflected in. On the other hand, in order to follow a change in opacity, the sampling interval must be narrowed, which may lead to an increase in the amount of data processing.

On the other hand, as described above, in the present embodiment, the voxel values of two points on the voxel boundary, that is, 2
By obtaining the opacity from the displacement of the voxel value between points, the change in opacity can be correctly reflected in the result. Also, the relationship between the voxel values of two points and the opacity is
Since it can be set in advance, the data processing amount does not increase as compared with the case where the number of sampling points is increased.

In the above embodiment, a case is described in which two points, ie, the incident point (start point) and the exit point (end point) of a ray on each voxel boundary are set, and the opacity is obtained from the voxel values at these two points. Although shown, three or more positions may be set along the ray passage path, and the opacity may be obtained based on a spline function that inputs voxel values at these three or more positions.

FIG. 6 is a diagram showing a configuration of an image forming apparatus to which the image forming method of the present embodiment is applied. The image forming apparatus 200 includes a ray direction setting unit 201 that sets a ray casting ray direction, a voxel detection unit 203 that detects voxels on a path of each ray along the set ray direction, and a ray path. 2 above and on voxel boundaries
A voxel value detection unit 205 that detects a voxel value of a point, a voxel value / opacity relationship setting unit 207 that sets a relationship between a voxel value and an opacity to select a voxel, and a voxel value / opacity value that is set. Based on the relationship, a voxel value / opacity table creation unit 209 creates a reference table describing one opacity for two voxel values, based on the plurality of detected voxel values based on the reference table described above. A voxel value / opacity converter 211 for obtaining opacity, and an image forming unit 213 for forming a projection image by ray casting based on the converted opacity. The projection image is displayed on the external display device 215.

The voxel detector 203 samples the three-dimensional voxel space 100 on the path of each ray at a specified interval in the set ray direction.
Get each voxel. The voxel value detection unit 205 calculates the voxel values Vm, V at two positions at the start point and the end point at which the ray enters the voxel on the boundary of the voxel for each voxel acquired by sampling.
Get n.

Voxel value / opacity table creation unit 20
9 divides the two voxel values Vm and Vn into k equal parts at sufficiently small intervals. Each value equally divided into k is
The voxel values are Vmn0, Vmn1, Vmn2,.... The voxel value / opacity relationship setting unit 207 sets the opacity αmn0, αmn1, αmn2,... Corresponding to the voxel values Vmn0, Vmn1, Vmn2,. The opacity corresponding to the two voxel values Vm and Vn is calculated by the following equation: 1 − ｛(1−αmn0) (1−αmn1)
(1-αmn2)...｝ ^{1 / k} is obtained.

The above formula is executed for all possible combinations of two voxel values, and the result is stored as a reference table.

The voxel value / opacity conversion section 211 acquires the opacity from the two voxel values acquired by the voxel value detection section 205 based on the reference table created by the voxel value / opacity table creation section 209. .

The image forming unit 213 forms a projection image by performing ray casting based on the acquired opacity.
The projection image is displayed on the display device 215.

As described above, by assigning the opacity of each voxel from two voxel values on the boundary of each voxel, a projection image free from artifacts can be obtained at high speed without increasing the data processing amount. It becomes possible.

In the above embodiment, the opacity is obtained from the voxel values of two points on the boundary of each voxel, that is, the displacement of the voxel values of the two points, but the distance between the two points is further taken into consideration. The opacity may be obtained by using

In the above-described embodiment, a case has been described in which an image is formed by performing ray casting on a three-dimensional voxel space. However, the present invention is not limited to this. The present invention can also be applied to a case where a physical quantity such as temperature is measured as a dimensional space, and the measured temperature distribution is expressed in color.

[0060]

According to the present invention, by assigning the opacity of each voxel on the ray path and on each voxel boundary from two voxel values, it is possible to correctly follow the change in the opacity. Moreover, by referring to the table created in advance, an image free from artifacts can be obtained at high speed without increasing the data processing amount by oversampling.

[Brief description of the drawings]

FIG. 1 is a diagram showing a result of calculating a representative value for analysis.

FIG. 2 is a diagram illustrating a result of contour extraction based on a calculated analysis representative value.

FIG. 3 is a diagram showing a three-dimensional voxel space 100.

FIG. 4 is a diagram showing a relationship between voxel values Vm and Vn and opacity αmn.

FIG. 5 shows two voxel values Vm, Vn and opacity αmn
It is a table showing the relationship with.

FIG. 6 is a diagram illustrating a configuration of an image forming apparatus to which the image forming method according to the embodiment is applied.

FIG. 7 is a characteristic diagram showing filter settings for obtaining opacity α according to voxel value V;

FIG. 8 is a characteristic diagram showing filter settings for obtaining only the opacity α near a specific voxel value V;

FIG. 9 is a diagram illustrating a state where ray casting is performed with the filter settings of FIG. 7;

FIG. 10 is a diagram illustrating a state where ray casting is performed with the filter settings of FIG. 8;

FIG. 11 is a diagram illustrating an analysis target and a minute space.

FIG. 12 is a diagram showing a result of calculating a representative value for analysis by a conventional technique.

FIG. 13 is a diagram showing a result of contour extraction based on a representative value for analysis calculated by a conventional technique.

FIG. 14 is a diagram showing a relationship between voxel values and opacity based on a conventional method.

FIG. 15 is a diagram showing a relationship between voxel values and opacity according to the present invention.

FIG. 16 is a diagram showing a procedure of ray casting.

FIG. 17 is a diagram showing a procedure of ray casting according to the present invention.

FIG. 18 is a graph showing the relationship between voxel values and opacity.

[Explanation of symbols]

 Reference Signs List 100 voxel space 200 image forming apparatus 201 ray direction setting unit 203 voxel detection unit 205 voxel value detection unit 207 voxel value / opacity relation setting unit 209 voxel value / opacity table creation unit 211 voxel value / opacity conversion unit 213 image formation Part 215 Display device

Continuation of the front page (56) References JP-A-8-335278 (JP, A) JP-A-8-16814 (JP, A) JP-A-4-10081 (JP, A) JP-A-61-153749 (JP) , A) JP-A-11-7541 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06T ⁷ /00-7/60 G06T 1/00 G06T 15/00-15/50 A61B 5/055

Claims (7)

(57) [Claims] 1. A method in which values distributed continuously in a three-dimensional space are discrete
Sampling obtained by dynamic sampling
For each data set of values, each fine
Get data set of opacity assigned to small space
And the opacity assigned to each of the adjacent minute spaces
Spatial continuity or loop to be analyzed based on the continuity of
A contour extraction method for extracting a contour, wherein two points on a boundary of each of the minute spaces and a path connecting them are defined.
To the sampling value and path length corresponding to the two points.
And calculating the opacity based on the contour.
Extraction method. 2. Discrete values distributed continuously in a three-dimensional space
Sampling obtained by dynamic sampling
For each data set of values, each fine
Get data set of opacity assigned to small space
And based on the opacity assigned to each of the minute spaces.
Image form that forms an image by ray casting
A forming method, lay in the path of the ray-casting, and before
The sampler corresponding to at least two points in each minute space.
Based on the signaling value,
An image forming method comprising calculating opacity. 3. A method according to claim 2, wherein two points on a boundary of each of said minute spaces correspond to two points.
Calculating the opacity based on the sampling value
3. The image forming method according to claim 2, wherein: 4. Further, two points on a boundary of each of the minute spaces.
Calculating the opacity based on a distance between the opacity.
The image forming method according to claim 3, wherein 5. The method according to claim 5, wherein the calculated opacity is calculated using the sump.
The feature is that it is stored as a table for ring values
The image forming method according to any one of claims 2 to 4,
Law. 6. An image is formed by ray casting.
In the image forming apparatus, the direction of the ray in the ray casting is set.
Ray direction setting means, which is on the path of the ray along the set direction of the ray
And infinitesimal space detecting means for detecting a very small space, the path of the ray, and the two points on the boundary of the microspaces
Sampling value detection to detect the corresponding sampling value
Setting means and a relationship between the detected sampling value and opacity
On the basis of the relationship between the set sampling value and the opacity.
Correspondence between the sampling values for the two points in advance and one opacity assigned to the minute space
Table creation means for creating a table in which the sampling value is described, based on the created table.
Sampling corresponding to two points detected by the output means
Opacity obtaining means for obtaining one opacity from a value, and the ray casting based on the obtained opacity.
An image forming apparatus comprising: the image forming means for forming an image, the by. 7. The apparatus according to claim 7, wherein said table creating means is configured to set said table.
The relationship between the sampled value and the opacity,
Distance between two points on the boundary of the micro space determined by the direction of
Creating the table on the basis of
The image forming apparatus according to claim 6.