JPH09259298A - Pseudo three-dimensional image constitution method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform improved density gradient calculation at the time of applying a volume rendering method to a center projection method, to faithfully constitute pseudo three-dimensional images from three-dimensional images, to substantially save the capacity of a memory used in the density gradient calculation and to accelerate a processing. SOLUTION: In this method, piled-up three-dimensional images for which plural topographic images including volume images are piled up are shaded and projected on a projection surface 30 set beforehand by using the volume rendering method and the pseudo three-dimensional images are constituted. In this case, the center projection method for which one view point (e) is an origin is used at the time of projection to the projection surface 30 and an image constitution voxel used in the volume rendering method is turned to a quadrangular pyramid base type voxel definable by a quadrangular pyramid constituted with the view point (e) as a vertex. That is, the quandrangular pyramid base type voxel is applied to projecting and shading algorithm to the projection surface in the volume rendering method and the processing similar to an image processing in the normal volume rendering method is realized even in the center projection method.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to an X-ray CT apparatus, MR
The present invention relates to a pseudo three-dimensional image constructing method for constructing a pseudo three-dimensional image by projecting a stacked three-dimensional image obtained from an I device or the like on a projection surface, and particularly, the inside of an endoscope or a laparoscope. The present invention relates to a pseudo three-dimensional image constructing method suitable for displaying a pseudo three-dimensional image for observation.

[0002]

2. Description of the Related Art Conventionally, as a pseudo three-dimensional image forming method of this kind, X-ray CT images are stacked to obtain a three-dimensional image.
There is a technique of projecting this on a projection surface to obtain a pseudo three-dimensional image. As a projection method, a parallel projection method and a central projection method are known. In the parallel projection method, the viewpoint is composed of planes (the viewpoint is set to infinity), and the projection is performed in parallel from this viewpoint on the projection surface, which is suitable for constructing images of external shapes and cut shapes. ing. In the central projection method, the viewpoint is a point, not a surface, and the line segment connecting this viewpoint and the three-dimensional image, or the point at which its extension line intersects the projection surface, becomes the coordinates on the projection surface, and moves like an endoscope. This is suitable when you want to obtain a projected image that simulates. For example, with an endoscope or a laparoscope, as it is advanced, the field of view changes rapidly. Moreover, the tip of the endoscope is small, and it is necessary to observe the changing visual field from this small viewpoint. The central projection method is optimal for obtaining a simulated pseudo three-dimensional image as if observing with such an endoscope (Japanese Patent Laid-Open No. 7-21070).
No. 4, Japanese Patent Application Laid-Open No. 8-16813).

A volume rendering method is known as a technique for projecting a three-dimensional image by shading it on a projection surface. In this volume rendering method, each pixel of a stacked three-dimensional image is called a voxel having a cubic shape. Therefore, the three-dimensional image is an image in which these voxels are three-dimensionally stacked. If the data of the voxel is an X-ray CT apparatus, the CT value calculated from the X-ray absorption value of the part of the human body corresponding to the voxel, M
In the case of an RI device, the measured value is a value such as a proton density.

The volume rendering method extracts a CT value that meets a threshold condition that is appropriately set for the three-dimensionally stacked voxel data (hereinafter referred to as a CT value), thereby extracting bones, skin, and the like. A desired subject is distinguished and extracted, and its CT value is used as a parameter for shading on the projection surface. Also, as a parameter for shading by the volume rendering method, CT
The transmittance / reflectance of light set with respect to the value, the gradient of the contour surface formed by the CT value (hereinafter referred to as the density gradient), and the like are also used. That is, the light traveling from the viewpoint to the projection plane passes through each voxel in the depth direction, but at that time, it is attenuated according to the transmittance / reflectance determined by the CT value of each voxel. Therefore, the magnitude of the reflected light in each voxel can be obtained, and in the shading algorithm by the volume rendering method, the larger the reflected light, the brighter the voxel. Further, it is considered that a virtual surface exists where the CT value of the voxel greatly changes, and the more the virtual surface faces the front, that is, the larger the concentration gradient, the brighter the voxel is.

Then, the information indicating the brightness of each pixel on the projection plane is calculated by accumulating the information of the brightness of the voxel which can be obtained as described above over each voxel in the depth direction from the viewpoint (hereinafter referred to as "value"). , Pixel values), and the pixel values of all pixels on the projection surface are obtained.

[0006]

However, when the volume rendering method is applied to the parallel projection method, it is possible to calculate the density gradient independent of the distance from the viewpoint when calculating the density gradient. When applied to the central projection method, there is a problem that the volume of voxels increases according to the distance from the point light source, and the concentration gradient calculation method that is not affected by the distance from the point light source cannot be used.

The present invention has been made in view of the above circumstances. When the volume rendering method is applied to the central projection method, good density gradient calculation is possible, and a pseudo three-dimensional image is faithfully converted from a three-dimensional image. It is an object of the present invention to provide a pseudo three-dimensional image constructing method constructing method that can be constructed, can significantly save the capacity of the memory used in the density gradient calculation, and can speed up the processing.

[0008]

In order to achieve the above object, the present invention provides a method of shadowing a stacked three-dimensional image in which a plurality of tomographic images including a volume image are stacked on a projection plane preset by using a volume rendering method. In a pseudo three-dimensional image constructing method for attaching and projecting to construct a pseudo three-dimensional image, a central projection method starting from one viewpoint is used for projection onto the projection plane, and the volume rendering method is used. It is characterized in that the image forming voxels to be used are quadrangular pyramid truncated voxels which can be defined by a quadrangular pyramid having the viewpoint as a vertex. That is, the present invention proposes a new voxel shape. Although the conventional voxel has the shape of a cube, this voxel is not suitable for the central projection method using a point light source. Adopt a truncated pyramid voxel). Then, by applying this quadrangular truncated pyramid voxel to the projection onto the projection surface and the shading algorithm in the volume rendering method, even the central projection method realizes the same processing as the image processing in the normal volume rendering method. Is possible.

However, the truncated pyramid-shaped voxels change in volume arranged in the depth direction. Therefore, the present invention is suitable for a case where a truncated pyramid-shaped voxel is adopted by using the volume ratio according to the distance from the viewpoint of the truncated pyramid-shaped voxel as a parameter for shading calculation in the volume rendering method. I am trying to shade it.

Further, according to the present invention, the density values of the quadrangular truncated pyramid-shaped voxels arranged from the viewpoint toward a certain pixel coordinate on the projection surface are sequentially captured from the viewpoint side in the depth direction, and the captured density is obtained. Of the values, only the density values within a predetermined range are sequentially stored in the memory table, and the distance information from the viewpoint of the truncated pyramidal voxel having the density value that was first captured is stored, and then the shading in the volume rendering method is performed. When calculating the concentration gradient of each truncated pyramid-shaped voxel used as a parameter for calculation based on the concentration value of the truncated pyramid-shaped voxel and the concentration value stored in the memory table adjacent thereto, The density values stored in the memory table are corrected by the distance information, and the distance values in the memory table are corrected to calculate the density values. It is way. This allows
The capacity of the memory table can be saved.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of a pseudo three-dimensional image constructing method according to the present invention will be described in detail below with reference to the accompanying drawings. As shown in FIG. 1, the pseudo three-dimensional image constructing method according to the present invention stacks a plurality of tomographic images including a volume image to obtain a stacked three-dimensional image, and shades this on a two-dimensional image viewed from an arbitrary direction. In the method of adding the images to form a pseudo three-dimensional image, when projecting each tomographic image onto the projection surface, the pixel coordinates of each tomographic image are converted into coordinates on the projection surface using the central projection (step S11). ), A pixel value is given to each pixel coordinate on the projection surface according to a shading algorithm, and shading is performed to form a pseudo three-dimensional image (step S12).

Here, an MRI apparatus or the like does not two-dimensionally measure an image slice by slice, unlike the X-ray CT apparatus.
Images of a plurality of slices can be measured three-dimensionally, and thus an image in which two-dimensional images are three-dimensionally arranged is obtained, which is called a volume image. This volume image (three-dimensionally arrayed image) can be decomposed into a two-dimensional array (slice array) image, and the “plurality of tomographic images” in the method of the present invention also includes the volume image.

The coordinate conversion by the central projection will be described in detail below. The conversion of the pixel coordinates of each tomographic image to the coordinates on the projection surface when projecting each tomographic image on the projection plane by the central projection is performed as follows. In the example shown in FIG. 2, in order to simplify the description, the coordinate system is set so that the projection plane and the tomographic image plane, and further the xy plane are parallel to each other.

In FIG. 2, x, y, and z are axes of the three-dimensional coordinate system (x, y, z) and point e (x1, y1, d1).
Is the position of the viewpoint e, point P (X, Y) is the point on the projection surface (corresponding to the display screen) 21, and point S (x0, y0, d0) is e.
A straight line 2 passing through the point (x1, y1, d1) and the point P (X, Y)
2 and the tomographic image 23A intersect.

D is the position of the projection surface 21 (on the z-axis)
Can be arbitrarily set, d0 is the position (on the z axis) of the tomographic image 23A, and d1 is the z coordinate of the viewpoint e. According to this, the following equation holds.

[0016]

X = {(D-d1) / (d0-d1)} × (x0-x1) + x1 (1)

[0017]

## EQU00002 ## Y = {(D-d1) / (d0-d1)}. Times. (Y0-y1) + y1 (2)

[0018]

## EQU00003 ## x0 = {(d0-d1) / (D-d1)}. Times. (X-x1) + x1 (3)

[0019]

Y0 = {(d0-d1) / (D-d1)} × (Y-y1) + y1 (4) The projected image is displayed on the display screen (not shown) corresponding to the projection surface 21. , 512 pixels in the vertical direction × 512 pixels in the horizontal direction, X and Y take values from 0 to 512. On the tomographic image 23A of d0 for each pixel coordinate (X, Y), x0 and y0 are determined by the above equations (3) and (4), and which point should be projected. There are multiple tomographic images 23,
Since there are a plurality of d0s, a plurality of points (x0, y0) to be projected are determined for one pixel coordinate (X, Y).

Next, the principle of the pseudo three-dimensional image constructing method according to the present invention will be described with reference to FIGS.
The method of the invention proposes a new voxel. In this voxel, as shown in FIG. 3, a quadrangular pyramid having a point light source e as a vertex and a bottom surface in the pixel direction of the projection surface 30 is represented by ray (ray).
It is a solid that is divided at equal intervals on a plane perpendicular to, and this is called a truncated pyramid voxel.

Opacity and density gradient calculation are mentioned as important parameters for realizing the processing using the volume rendering method. For a truncated pyramidal voxel with a uniform height, similar to the conventional voxel, the same opacity calculation as the conventional method can be used. However, in the concentration gradient calculation, since the volume of the truncated pyramidal voxel varies depending on the distance from the point light source, the concentration gradient calculation method that is not influenced by the distance as in the conventional method cannot be used.

Therefore, as shown in FIG. 4, the volume ratio based on the quadrangular pyramid is used to obtain the volume ratio of the front and rear quadrangular pyramid truncated voxels. The length of one side of the bottom of the quadrangular pyramid is a, the height is h,
The height of the truncated pyramidal voxel is 1, and the truncated pyramids A, B, C
The volume of is calculated by the following formula.

[0023]

(Equation 5)

Then, the volume ratios of the front and rear quadrangular pyramid-shaped voxels are calculated from these three formulas, and the following formula is obtained.

[0025]

(Equation 6)

In the above equation, if h is the distance from the point light source to the projection surface and 1 is the interpolation width, the volume ratio of the front and rear quadrangular truncated pyramid voxels can be obtained. Therefore, the calculation formula of the concentration gradient in consideration of the volume of the truncated pyramidal voxel according to the distance from the point light source can be, for example, the following formula.

[0027]

(Equation 7)

Here, PROP is a volume ratio coefficient including depth information, and for the voxel located in the depth direction n from the point light source, from the volume ratio prop calculated by the equation (6),
It is defined as follows.

[0029]

## EQU00008 ## PROP = prop.times.n (n.gtoreq.1) (8) As a result, it becomes possible to construct the truncated pyramidal voxel, and the pseudo three-dimensional image constructing method using the voxel will be described below. Will be described. Image processing often requires more advanced techniques and enormous amounts of memory to compose more delicate images. In pseudo three-dimensional image reconstruction using the depth method and the voxel method, rendering is performed by preparing a memory for storing a density value, for example, a CT value, which is a z-buffer for shading calculation. , Two z-buffer memory tables 40 are used as shown in FIG.

Then, as shown in FIG. 6, the first table 40a has pixel coordinates (0, 0) to (51) on the projection surface.
All the CT values of the truncated pyramidal voxels corresponding to (2, 0) are stored, and the four tables corresponding to the pixel coordinates (0, 1) to (512, 1) of the projection surface are stored in the second table 40b. Only the minimum necessary CT value, which is the CT value of the truncated pyramid voxel, is calculated.

Here, the minimum required CT value is a CT value that meets a preset threshold condition, and C that meets the threshold condition based on the transmittance set according to the CT value.
It is the CT value until the calculated light transmittance is calculated to be 0 by calculating the light transmittance according to the T value. That is, it is not necessary to completely fill the memory tables of the second and subsequent sheets with CT values, and it is sufficient that the light is completely attenuated and stored in the storage positions a and b of the memory table 40 as shown in FIG. The light attenuation is calculated in the order of the CT values, and the CT values are stored up to the point c where the rendering condition is satisfied (light is completely attenuated). This eliminates waste of storing CT values in the depth direction, saves memory, and speeds up because less wasteful processing is performed.

Based on the CT values stored in the two tables in this manner, as shown in FIG. 5, for each quadrangular truncated pyramid voxel, the CT value of the quadrangular truncated pyramid voxel and
Adjacent to this, the differences dx, dy, and dz of the CT values in each direction are obtained based on the CT values stored in the memory table.
Then, these dx, dy, dz and the volume ratio coefficient PR
The concentration gradient gg is calculated from OP (equation (8)) based on the above equation (7).

Thereafter, a value (hereinafter referred to as a voxel value) indicating the brightness of the square truncated pyramid voxel is obtained based on the CT value of the square truncated pyramid voxel, the density gradient calculated above, and the magnitude of reflected light. Rendering calculation processing is performed, and the sum of the voxel values subjected to the rendering calculation processing with respect to one or a plurality of truncated pyramidal voxels corresponding to the pixel coordinates of the projection surface is set as a pixel value indicating the brightness of the pixel coordinates in the display memory ( (See FIG. 10).

Next, the CT values stored in the second table 40b are used as they are, and the empty contents of the first table 40a are stored in the next CT value as the third table 40a '. Used (see FIG. 6). In this way,
By alternately using the two tables, memory is saved. By the way, when the CT values satisfying the threshold value are sequentially stored in the memory table, even if the points A and B are on the same tomographic image 20a as shown in FIG. The distance from the light source increases, and the data will be distorted. That is, in FIG.
The arc-shaped equidistant line n from is a straight line in the memory table. On the contrary, the tomographic image 20a is stored while being curved in the memory table array as shown by 20a '.
Further, when the concentration gradient calculation is performed, it is not always guaranteed that the adjacent CT value is the CT value located next to the original data.

Therefore, in the present invention, as shown in FIG. 9, first, an array d for storing the distance from the point light source of the voxel having the CT value satisfying the threshold value condition is provided. As a result, it is possible to eliminate the waste of securing an unnecessary storage area up to the position that satisfies the threshold value condition, and to accurately provide three-dimensional position (distance) information. Corrects the adjacent CT value of the so that it will be the CT value located next to the original data (interpolation)
However, an appropriate concentration gradient can be calculated. Also, by giving distance information, on the memory map,
The surface of the data can be created three-dimensionally and can be applied to surface rendering.

FIG. 10 is a block diagram showing a hardware configuration to which the method of the present invention is applied. In the figure, 53
Is a central processing unit (CPU), 51 is a main memory, 52 is a magnetic disk, 54 is a display memory, 56 is a controller, and these are connected to a common bus 58. The magnetic disk 52 stores a volume image formed by stacking a large number of tomographic images.

The CPU 53 executes a predetermined process on the volume image stored in the magnetic disk 52 according to the projection display software for coordinate conversion and shading stored in the main memory 51, and shading is performed. A pseudo three-dimensional image subjected to the above is constructed, and the result is displayed in the display memory 54.
To display on CRT55. In this processing, input / output processing and processing operations are performed using the mouse 57 and the keyboard attached to the controller 56. Further, the display content is stored in the magnetic disk 52 and used for re-display.

Next, the processing contents of the CPU 53 will be described with reference to the flowchart shown in FIG.
First, the pixel coordinates (X, Y) on the projection plane are set to X = 0, Y
= 0 (step S20). This coordinate value becomes the starting point. The projection surface is divided into pixels of X = 0 to 512 and Y = 0 to 512. Moreover, the projection plane and the viewpoint are preset.

Next, the coordinate conversion program receives the coordinate value of the target position to be rendered along the ray from the viewpoint to the coordinate on the projection surface (step S2).
2) It is determined whether or not the CT value at the rendering target position meets a preset threshold condition (TH ₁ <CT value <TH ₂ ) (step S24). If they match, the CT value is stored in the memory table (step S
26). It is determined whether the stored CT value is the CT value that first satisfies the threshold value condition (step S28). If the stored CT value is the CT value that first satisfies the threshold value condition, the distance d from the viewpoint position is also stored. Yes (step S30).

Then, the light attenuation calculation is performed by the CT value satisfying the threshold condition (step S32). In addition, CT beforehand
The reflectance / transmittance is set according to the value. And
It is determined whether or not the light is completely attenuated by the attenuation calculation in step S32 (that is, whether or not the amount of incident light on the voxel at the back rendering position is almost 0) (step S34). Is the coordinate (X,
The next (back) rendering position in Y) is obtained (step S36), and the steps S22 to S34 are performed.
Execute the processing of

On the other hand, when the attenuation is complete, the coordinates (X,
The X coordinate of Y) is incremented by 1 (step S3
8), it is determined whether the X coordinate is less than 512 (step S40). If the X coordinate is less than 512, step S
Returning to 22, when the X coordinate becomes 512, the coordinates (X, Y)
X coordinate of X is set to X = 0 (step S42), and then the CT value stored in the memory table in step S26,
Based on the distance d from the viewpoint position stored in step S30, the differences dx, dy, dz of the CT values in each direction and the density gradient gg (see formula (7)) are calculated. Coordinate Y = 0
In this case, the CT values of the first memory table are all 0, for example, and the CT values are stored in the second memory table in step S26.

Next, the CT value is standardized so that the pixel value on the projection surface has a desired dynamic range (step S46), and then the CT of the truncated pyramid voxel is obtained.
Rendering calculation processing for obtaining a voxel value based on the value, the density gradient, and the magnitude of the reflected light is performed (step S4).
8) Store the sum of voxel values that have been subjected to rendering calculation processing for one or more truncated pyramidal voxels until the light is completely attenuated in the display memory as a pixel value indicating the brightness of the coordinates (X, Y). Yes (step S50).

When the storage of 512 pixel values in the X direction is completed, the Y coordinate of the coordinate (X, Y) is then incremented by 1 (step S52), and it is determined whether the Y coordinate is less than 512. Yes (step S54). Y coordinate is 51
If it is less than 2, the process returns to step S22 and the above process is repeated. On the other hand, in step S54, the Y coordinate is 5
When it becomes 12, the pixel values for all the coordinates on the projection plane are stored in the display memory, and the image (pseudo three-dimensional image) corresponding to the pixel values stored in this display memory is CR.
Can be displayed on T.

[0044]

As described above, according to the pseudo three-dimensional image constructing method of the present invention, a new quadrangular truncated pyramid voxel is proposed when reconstructing a pseudo three-dimensional image using the central projection method. As a result, an image faithful to the data can be reconstructed, and particularly when the volume rendering method is applied to the central projection method, good density gradient calculation becomes possible. Further, the capacity of the memory used in the concentration gradient calculation can be greatly saved, and the processing speed can be increased.

[Brief description of drawings]

FIG. 1 is a flow chart showing the outline of the method of the present invention.

FIG. 2 is a diagram used for explaining conversion of pixel coordinates of a tomographic image into coordinates on a projection surface in the method of the present invention.

FIG. 3 is an explanatory diagram of a truncated pyramid voxel according to the present invention.

FIG. 4 is an explanatory diagram for performing volume ratio calculation of a truncated pyramid voxel according to the present invention.

FIG. 5 is an explanatory view of the outline of the method of the present invention.

FIG. 6 is a diagram for explaining a method of using a CT value storage array.

FIG. 7 is a diagram for explaining a method of storing CT values in a CT value storage array.

FIG. 8 is a diagram for explaining distance distortion that occurs in the CT value storage array.

FIG. 9 is a diagram for explaining a method of correcting distance distortion that occurs in the CT value storage array.

FIG. 10 is a block diagram showing a hardware configuration to which the method of the present invention is applied.

FIG. 11 is a flowchart used to explain the processing of the CPU of FIG.

[Explanation of symbols]

 20 ... tomographic image 30 ... projection plane 40 ... memory table 51 ... main memory 52 ... magnetic disk 53 ... CPU 54 ... display memory 55 ... CRT 56 ... controller 57 ... mouse e ... viewpoint (point light source) n ... equidistant from viewpoint Line 20a '... Equidistant line from viewpoint in memory table

Claims (6)

[Claims] 1. A stacked three-dimensional image obtained by stacking a plurality of tomographic images including a volume image is projected on a projection plane set in advance by shading using a volume rendering method,
In a pseudo three-dimensional image constructing method for constructing a pseudo three-dimensional image, a center projection method starting from one viewpoint is used for projection onto the projection plane, and an image constructing voxel used in the volume rendering method is used. A pseudo-three-dimensional image constructing method, characterized in that it is a truncated pyramidal voxel which can be defined by a quadrangular pyramid having the viewpoint as a vertex. 2. A shading suitable for the truncated pyramidal voxel is obtained by using a volume ratio of the truncated pyramidal voxel according to a distance from the viewpoint as a parameter for shading calculation in the volume rendering method. The method for constructing a pseudo three-dimensional image according to claim 1, wherein 3. The density values of the truncated pyramidal voxels arranged from the viewpoint toward a certain pixel coordinate on the projection surface are sequentially captured from the viewpoint side in the depth direction, and the captured density values are acquired. Of these, only the density values within a predetermined range are sequentially stored in the memory table, and the distance information from the viewpoint of the truncated pyramidal voxel having the density value that is first captured is stored, and then the shadow in the volume rendering method is stored. When calculating the concentration gradient of each truncated pyramid-shaped voxel used as a parameter for the attachment calculation based on the concentration value of the truncated pyramid-shaped voxel and the concentration value stored in the memory table adjacent thereto, the adjacent Then, the density value stored in the memory table is corrected by the distance information to correct the distance distortion in the memory table. Pseudo three-dimensional image constructing method as claimed in claim 1, wherein the calculating using. 4. The correction is performed so that the calculated concentration gradient increases as the volume of the truncated pyramid voxel increases based on the volume ratio of the truncated pyramid voxel from the viewpoint. The pseudo three-dimensional image constructing method according to claim 3, characterized in that. 5. When the density value of the truncated pyramidal voxel is stored in the memory table, the light transmittance according to the stored density value is calculated based on the transmittance set according to the density value. 4. The pseudo three-dimensional image according to claim 3, wherein the storage of the density value is executed in the memory table until the calculated transmittance becomes approximately 0 to save the storage capacity of the memory. How to configure. 6. A voxel value in the volume rendering method is calculated based on density values stored in the two memory tables, and then the density values of one of the memory tables are used as the memory table. Is calculated, and the voxel value in the volume rendering method is calculated based on the rewritten density value and the density value stored in the other memory table, and the stored contents of the two memory tables are alternately rewritten. The pseudo three-dimensional image constructing method according to claim 3 or 5, wherein the storage capacity of the memory is saved by repeating the processing sequentially.