JP2010152609A - Voxel array visualization device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voxel array visualization device for suitably visualizing a voxel array obtained by nuclear medical diagnosis such as PET (Positron Emission Tomography) and SPECT (Single Photon Emission Computed Tomography) with a small processing loads. <P>SOLUTION: The voxel array visualization device includes steps of: calculating a viewing angle according to a set parameter (S1); arranging a voxel array in a three-dimensional coordinate system (S2); rotating the voxel array according to the calculated view angle (S3); projecting the voxel array by a Z-buffer method using a luminance frame and a Z buffer installed corresponding to the viewing angle (S4); applying shading (S5); and performing grayscale conversion according to a display output means (S6). <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a technique for visualizing volume data of a voxel array configuration obtained mainly by nuclear medicine diagnosis such as PET and SPECT.

  2. Description of the Related Art Conventionally, an imaging technique for creating a cross-sectional image by image reconstruction by non-invasively sensing a three-dimensional structure inside a living body from outside the living body has been put to practical use mainly in the medical field. Such imaging methods include CT (Computed Tomography) that generates images by irradiating X-rays from outside the body, sensing the X-ray transmittance of each tissue, irradiating a high-frequency magnetic field from outside the body, MRI (Magnetic Resonance Imaging), which senses radio waves radiated from water molecules in water, injects radioactive compounds (tracers) into the body, and senses gamma rays emitted from compounds accumulated in specific tissues Then, there are three types of SPECT (Single Photo Emission Tomography) or PET (Positron Emission Tomography) for creating an image.

  CT and MRI can visualize the shape from the difference in the X-ray transmittance of tissues and the density of water molecules, and so are suitable for creating an in-vivo map. On the other hand, PET / SPECT can visualize a quantitative value of a compound accumulated in a specific tissue, and is suitable for creating a traffic distribution map for each area if compared to a map.

  All of the above imaging methods can create cross-sectional images, and voxel images composed of multiple continuous cross-sectional images can be created by moving the scanning head or using multiple sensors. A voxel movie can be created by repeatedly scanning each time. However, even if a large number of collected cross-sectional images are arranged, it is difficult to glance at the three-dimensional structure in the living body. Therefore, for CT and MRI, a technique has been proposed in which volume rendering is performed on a voxel image to visualize it (see Patent Document 1).

Therefore, with regard to PET, a composition method has been developed in which the shape is drawn by CT or MRI and the quantitative values of PET are superimposed, and an apparatus capable of simultaneously scanning both images of PET-CT has also been developed (see Patent Document 2). ).
Japanese Patent Publication No.7-120434 JP 2005-518915 A

  However, when volume rendering as described in Patent Document 1 is applied to information obtained from PET / SPECT, the information regarding the shape is missing, so that it is difficult to grasp the conventional positional relationship. There is a problem. In addition, the ray casting method, which is a standard algorithm for volume rendering, has a problem that the calculation load is heavy.

  In addition, the synthesis method as described in Patent Document 2 has a very large calculation load, and a dedicated mini supercomputer is required to perform real-time reproduction (the PET and CT apparatuses themselves are large-scale, so the current calculation The load is not considered as a problem.) The synthesis method as described in Patent Document 2 has a problem that it is difficult to visualize the shape of the human body when there is no information such as CT and only PET information is obtained.

  Accordingly, an object of the present invention is to provide a voxel array visualization device that can suitably visualize a voxel array obtained by nuclear medicine diagnosis such as PET and SPECT with a small processing load.

  In order to solve the above-described problems, the present invention creates projection image data obtained by projecting a voxel array on a two-dimensional plane at a specified angle with respect to a voxel array in which voxel values are defined in a three-dimensional coordinate system. An image buffer setting means for setting a luminance value buffer and a Z buffer for storing a luminance value and a depth Z value in a two-dimensional coordinate system with the same pixel size, and a voxel in the voxel array. Rotation processing means for executing rotation processing on a three-dimensional coordinate system with an angle specified for each, and each voxel value V (x, y, z) corresponding to the (x, y, z) coordinates after execution of the rotation processing Only if the z value has a value closer to the viewpoint than the value recorded in the Z buffer, the corresponding pixel (x, y) on the luminance value buffer is updated while updating the Z buffer with the z value. Projection processing means for calculating a pixel value V (x, y) of the brightness value buffer by executing a process for reflecting a voxel value corresponding to each z value; and on the brightness value buffer calculated by the projection processing means There is provided a voxel array visualizing device having gradation converting means for performing a predetermined calculation in order to match each pixel value with the output gradation of the display means.

  According to the present invention, when projecting a voxel array onto a two-dimensional plane, a Z buffer that holds a value in the z-axis direction orthogonal to the two-dimensional plane is prepared, and the z value of the projected voxel is recorded in the Z buffer. Only when the value is closer to the viewpoint than the measured value, the Z buffer is updated with the z value, and the degree of reflection is reflected depending on whether the z value of the voxel has a value close to the viewpoint or a value far from the viewpoint. In other words, the voxel value is reflected in the pixel value, so that it is possible to create a projection image in which volume data is projected at high speed with a small processing load.

  According to the present invention, it is possible to create a projection image in which volume data is projected at high speed with a small processing load.

DESCRIPTION OF EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.
(1. Device configuration)
First, the configuration of the voxel array visualization device according to the present invention will be described. FIG. 1 is a configuration diagram of a voxel array visualization device according to the present invention. In FIG. 1, 10 is a voxel array storage unit, 20 is a voxel conversion processing unit, 30 is a pixel conversion processing unit, 40 is an image display unit, and 50 is a parameter setting unit.

  The voxel array storage means 10 stores volume data of a three-dimensional array (voxel array) obtained by a nuclear medicine diagnostic apparatus such as PET or SPECT, and is a storage device such as a hard disk built in or connected to a computer. Realized. The voxel conversion processing unit 20 converts the voxel array in a three-dimensional manner, and includes an in-space arrangement unit 21, a viewpoint angle calculation unit 22, and a rotation processing unit 23. The pixel conversion processing unit 30 converts the voxel array into a two-dimensional image and processes it as a two-dimensional image, and includes a projection processing unit 31, a shadow processing unit 32, and a gradation conversion unit 33. The image display means 40 displays and outputs the luminance frame data that is the result obtained by the pixel conversion processing unit 30, and is realized by a liquid crystal display or other display output device. The parameter setting means 50 sets parameters for specifying processing for the voxel array, and is realized by a keyboard, a mouse, or the like. The voxel array visualization apparatus shown in FIG. 1 is actually realized by incorporating a dedicated program into a general-purpose computer.

  The volume data of the voxel array stored in the voxel array storage means 10 will be described. The volume data is voxel array data obtained by imaging a human body with a nuclear medicine diagnostic apparatus such as PET or SPECT. The volume data used in the present embodiment is 128 × 128 × 47 pixels and is obtained by imaging 61 times in time series. FIG. 2 schematically shows the configuration of volume data used in the present embodiment. In the example of FIG. 2, one rectangle indicates a cross-sectional image in the XY cross section. For example, in the case of a human head, 47 XY cross-sectional images exist in the Z-axis direction (usually in the body axis direction of the human body) at the same time (T), and 61 XYZ voxel arrays (three-dimensional arrays) are different. There are a number corresponding to the time (T). Therefore, when the time T is specified, a voxel array of 128 × 128 × 47 voxels at that time is obtained.

(2. Processing operation)
Next, the processing operation of the voxel array visualization device according to the present invention will be described. First, the user uses the parameter setting unit 50 to set parameters. As specific parameters, time T and viewpoint angle of volume data are set. Time T specifies volume data to be processed. For example, in the example of FIG. 2, by specifying the time T, one three-dimensional array to be processed is determined from 61 three-dimensional arrays. As the viewpoint angle parameter, three types of rotation angle Rxy on the XY plane, rotation angle Rxz on the XZ plane, and rotation angle Ryz on the YZ plane are set. The viewpoint angle parameter is set by designating the viewpoint angle at the initial position and the viewpoint angle at the final position and the number of frames from the initial position to the final position.

  FIG. 3 is a flowchart showing an outline of the processing operation of the voxel array visualization device after parameter setting. When the parameters are set, in the voxel array visualization device, first, the viewpoint angle calculation unit 22 calculates the viewpoint angles according to the set parameters. Specifically, when the initial frame is 1, the viewpoint angle parameters are Rxy1, Rxz1, and Ryz1, the final frame is Fm (specified number of frames), and the viewpoint angle parameters are Rxy2, Rxz2, and Ryz2, the viewpoint angle calculation means 22 calculates viewpoint angles Rxy, Rxz, and Ryz in each frame F by executing processing according to the following [Equation 1].

[Formula 1]
Rxy = (Rxy2-Rxy1). (F-1) / (Fm-1) + Rxy1
Rxz = (Rxz2-Rxz1). (F-1) / (Fm-1) + Rxz1
Ryz = (Ryz2-Ryz1). (F-1) / (Fm-1) + Ryz1

  In [Formula 1], F is a variable indicating a frame number and takes an integer value in the range of 1 to Fm. At first, since the viewpoint angle of the first frame is calculated, the viewpoint angle calculation means 22 performs processing according to the above [Equation 1] with F = 1, and sets the viewpoint angles Rxy, Rxz, Ryz. calculate.

  After setting the viewpoint angles Rxy, Rxz, and Ryz by the parameter setting unit 50, the in-space arrangement unit 21 arranges the volume data in the three-dimensional coordinate system (in the three-dimensional space) (S2). . Specifically, if the size of the voxel array is (Sx, Sy, Sz) and the scaling ratio with respect to the XY coordinates in the Z-axis direction is γ, all Sx × Sy × Sz voxels (x, y, z) at time T ) Is converted and arranged as in the following [Equation 2].

[Formula 2]
(X, y, z) → (x-Sx / 2, y-Sy / 2, (z-Sz / 2) γ)

  In the example of [Formula 2], the arrangement is performed so that the center of the voxel array is located at the origin of (0, 0, 0). The z-coordinate is multiplied by a coefficient γ. This is because the resolution of the detector in the bed moving direction of the scanner device does not match the resolution in the cross-sectional direction, and the former is generally coarser than the latter. In order to make each voxel of the array look like a cubic shape without any spatial distortion, the range that the z value can take is matched with the x coordinate and the y coordinate. For example, when a voxel array of 128 × 128 × 47 voxels is used, γ = 2 is set. By setting γ = 2, the ratio of x and y becomes 128: 94, which is not perfect, but each voxel is close to a cube. However, when the scanned voxel array is the head as in this example, the entire voxel array itself is similar to the geometric shape of the head, and the shape of the voxel array after multiplying by the γ coefficient is generally Z. It becomes a long rectangular parallelepiped in the direction. Further, as in this example, even if it is 128 × 128 × 47 voxels, it is not necessary to multiply the scaling ratio γ if it is not necessary to display the z direction at the same interval as the x and y directions. Usually, volume data obtained by a nuclear medicine diagnostic apparatus such as PET or SPECT is set such that each coordinate of x, y, and z is an integer value of 1 or more. If the center of the volume data stored in the voxel array storage means 10 is set to (0, 0, 0), the process according to [Formula 2] is executed. do not have to.

  Next, the voxel array rotating unit 23 rotates the voxel array in accordance with the change of the viewpoint position (S3). Specifically, with respect to all Sx × Sy × Sz voxels V (x, y, z) at time T, rotation angles Rxy, Rxz, Ryz (with respect to the set XY plane, XZ plane, and YZ plane) are set. The unit of each angle is radians), and the conversion is performed by executing the processing according to the following [Equation 3]. [Expression 3] (x, y, z) in the first expression is after the coordinate conversion by [Expression 2], and (x, y, z) in the second expression is after the coordinate conversion by the first expression. (X, y, z) in the third equation is after the coordinate conversion by the second equation. Therefore, the processing results differ depending on the processing order of the three formulas shown in [Formula 3]. In this embodiment, execution is performed in the order of the following first formula (rotation on the XY plane), second formula (rotation on the XZ plane), and third formula (rotation on the YZ plane). It is not necessary to execute in this order, and it may be executed in another order.

[Formula 3]
(X, y, z) → (x · cosRxy + y · sinRxy, −x · sinRxy + y · cosRxy, z)
(X, y, z) → (x · cosRxz + z · sinRxz, y, −x · sinRxz + z · cosRxz)
(X, y, z) → (x, y · cosRyz + z · sinRyz, −y · sinRyz + z · cosRyz)

  When the processing by the voxel conversion processing unit 20 is completed by executing the rotation processing by the voxel array rotation means 23, the processing by the pixel conversion processing unit 30 is started next. First, the projection processing means 31 performs a process of projecting a three-dimensional voxel array onto a two-dimensional frame (S4). Specifically, the voxel array address is converted into a frame address and the Z buffer method is executed. Since the conversion to the frame address corresponds to the frame size Sfx × Sfy to be rendered, the luminance frame B (x ′, y ′) (integer value) of Sfx × Sfy and the Z buffer Z (x ′, y ′) (actual) Value) in the computer's memory. The luminance value of each pixel of the luminance frame B is B (x ′, y ′) = 0 in all pixels in the initial state, and the value of each pixel in the Z buffer Z is Z (x ′, y in all pixels in the initial state. ′) = − ∞. Note that “−∞” is a theoretical value, and when it is actually initialized, a negative value having a sufficiently large absolute value is set as Z (x ′, y ′).

  When the initial setting of the luminance frame B and the Z buffer Z is completed, the projection processing unit 31 executes processing according to the following [Equation 4], and each voxel (x, y, A frame address (x ′, y ′) corresponding to z) is obtained. The frame address is common to the luminance frame B and the Z buffer Z.

[Formula 4]
(X, y, z) → (x + Sfx / 2, y + Sfy / 2, z)

  Since both the frame address of the luminance frame B and the Z buffer Z are integer values having x and y coordinates of 0 or more, Sfx / 2 and Sfy / 2 are added in the above [Equation 4], respectively. Therefore, x ′ = x + Sfx / 2 and y ′ = y + Sfy / 2.

  Further, the projection processing means 31 defines a transmittance α that takes a real value from 0 to 1, and performs so-called Z buffer method for all Sx × Sy × Sz voxels (x ′, y ′, z). Execute. Specifically, the projection processing unit 31 executes processing according to the following [Equation 5].

[Formula 5]
1) When z> Z (x ′, y ′) B (x ′, y ′) ← αB (x ′, y ′) / (z−Z (x ′, y ′)) ² + V (x ′, y ', z)
Z (x ′, y ′) ← z
2) In the case of z <Z (x ′, y ′) B (x ′, y ′) ← B (x ′, y ′) + αV (x ′, y ′, z) / (z−Z (x ′, y ')) ²
3) When z = Z (x ′, y ′) B (x ′, y ′) ← B (x ′, y ′) + V (x ′, y ′, z)

  As is clear from the above [Equation 5], for the Z buffer Z (x ′, y ′), when z> Z (x ′, y ′), that is, the voxel in the three-dimensional space is represented by the Z buffer Z (x It is updated with the z-coordinate value of the voxel only when it is on the near side (viewpoint side) in the z-axis direction with respect to the z-coordinate value recorded in ', y'). Therefore, the value of the Z buffer Z (x ′, y ′) is constantly updated to a large value. When z ≦ Z (x ′, y ′), the value of the Z buffer Z (x ′, y ′) is not updated.

  Subsequently, the shadow processing means 32 performs a shadow addition process (S5). Specifically, the shading luminance L (x ′, y ′) is calculated with reference to the Z buffer Z (x ′, y ′) for all pixels of Sfx × Sfy of the luminance frame, and has already been calculated. The value of each pixel (x ′, y ′) of the brightness frame B is multiplied and the value of each pixel (x ′, y ′) of the brightness frame B is updated to add a shadow.

Here, a method for calculating the shading luminance L (x ′, y ′) will be described. Consider a case where the shading luminance L (x ′, y ′) of a point P (x ′, y ′) in the Z buffer Z (x ′, y ′) is calculated. At this time, the point P _−1,0 (x′−l, y, zl), P ₊ , which represents the point P adjacent to the point P in the x-axis direction and the point adjacent to the y-axis direction inclusive of the z value. _1,0 (x ′ + l, y ′, zr), P _{0, + 1} (x ′, y ′ + l, zu), P _{0, −1} (x ′, y′−l, zd) Define. The vector of two points in the x-axis direction _{U = P +1,0 -P -1,0 = (} 2,0, zr-zl), the vector V = P ₀ of the two points in the y-axis _{direction, + 1} - P _{0, −1} = (0,2, zu-zd) is calculated.

  A normal vector N at the point P is obtained as an outer product of the vector U and the vector V. Therefore, N = U × V = (− 2 (zr−zl), −2 (zu−zd), 4) = (zl−zr, zd−zu, 2). On the other hand, the relationship shown in [Formula 6] below holds between the shading luminance L (x ′, y ′) and the parallel light source vector and normal vector. The parallel light source vector is set in advance, and is set as (1, 1, 1) in the present embodiment.

[Formula 6]
L (x ′, y ′) · cos θ = (inner product of parallel light source vector and normal vector N)

In the above example, L (x ′, y ′) · cos θ = (zl−zr + zd−zu + 2) / [(zl−zr) ² + (zd−zu) ² +4)] ^1/2 / 3 ^1/2 Become. Therefore, the shading luminance calculation means 33 calculates the shading luminance L (x ′, y ′) at each pixel (x ′, y ′) by executing processing according to this mathematical expression. As described above, the shading luminance calculation unit 33 uses the calculated shading luminance L (x ′, y ′) of each pixel (x ′, y ′) to each pixel (x ′, y ′) on the luminance frame B. ) And the luminance value of each pixel (x ′, y ′) on the luminance frame B is updated as a value including a shadow.

  Next, the gradation conversion means 32 performs gradation conversion of the luminance frame (S6). Specifically, an appropriate coefficient β is set for each pixel value B (x ′, y ′) of Sfx × Sfy of the luminance frame B, and a process according to the following [Equation 7] is executed. It is converted into an integer value of 256 gradations having a value of 0-255. The specific determination of β is to actually display and output the volume data several times and to empirically find that most of the pixels (x ′, y ′) on the luminance frame B are less than 256. To do. Some pixels may exceed 256, but their pixel values are limited to 255 uniformly.

[Formula 7]
When B (x ′, y ′)> 0, B (x ′, y ′) = 256 · β · B (x ′, y ′) + 1
When B (x ′, y ′) ≦ 0, B (x ′, y ′) = 0

  A set of luminance values obtained for the pixels on the luminance frame B in this way becomes projection image data. The obtained projection image data is displayed on the image display means 40, and the state where the volume data is projected can be confirmed. When the projection processing at one viewpoint angle is completed, the voxel array visualization device determines whether the processing has been completed for all viewpoint angles. Specifically, it is determined whether or not F indicating the frame number has reached the designated number of frames Fm. If F is less than Fm as a result of the determination, the value of F is incremented by one, and then the process returns to S1 to process the next frame.

  The preferred embodiment of the present invention has been described above. However, the present invention is not limited to the above embodiment, and various modifications can be made. For example, in the above-described embodiment, the shading luminance calculation unit 33 calculates the shading luminance L (x ′, y ′) and multiplies each pixel value B (x ′, y ′) of the luminance frame by this. However, it is not always necessary to multiply the shading luminance L (x ′, y ′). When shading brightness is multiplied, shadows are added, making it possible to create a more stereoscopic projection image. However, the voxel array visualization device of the present invention can quickly project projection data from a voxel array even if shadows are not added. Has the effect of creating.

  Further, in the above embodiment, by sequentially performing the process of projecting while changing the viewpoint angle, it is possible to obtain an image as if the voxel array arranged at a certain fixed position is viewed while moving around. . In this case, if the projection processing of each frame is slowed down, the moving motion becomes dull and the high-speed processing according to the present invention is effective. Even when the display is performed with only one designated viewpoint angle without changing the viewpoint angle, the effect of creating projection data from the voxel array at a higher speed than the conventional projection processing can be obtained.

(3. When processing the entire time-series data)
In the above embodiment, the case where the voxel array when the time T is specified is set as the processing target has been described, but it is also possible to set all time-series voxel array sets as shown in FIG. 2 as the processing target. In this case, the voxel array at time T = 1 is projected and visualized as frame 1, and the voxel array at time T = 2 is projected and visualized as frame 2, so that time T = Fm (frame FIG. In the example of 2, the processing up to the projection and visualization of the 61) voxel array is performed. The viewpoint angle may be changed as in the above embodiment, or may be fixed.

  When all voxel array sets are to be processed, the basic processing flow is the same as that shown in the flowchart of FIG. 3, but whether or not the entire processing is to be terminated after the gradation conversion processing in S6. The determination is made based on whether the processing of the voxel array at all times T has been completed.

It is a block diagram of the voxel arrangement | sequence visualization apparatus which concerns on this invention. It is a figure which shows typically the structure of the volume data used by this embodiment. It is a flowchart which shows the outline | summary of the processing operation of a voxel arrangement | sequence visualization apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Voxel arrangement | sequence storage means 20 ... Voxel conversion process part 21 ... Space arrangement means 22 ... View angle calculation means 23 ... Voxel arrangement rotation means 30 ... Pixel conversion process part 31. ..Projection processing means 32 ... shadow processing means 33 ... gradation conversion means 40 ... image display means 50 ... parameter setting means

Claims (7)

A visualization device for creating projection image data obtained by projecting a voxel array on a two-dimensional plane according to a specified viewpoint angle with respect to a voxel array in which voxel values are defined in a three-dimensional coordinate system,
A luminance frame for accumulating luminance values in a two-dimensional coordinate system;
A Z buffer having the same pixel size as the luminance frame and storing a depth Z value;
Voxel array rotation means for executing rotation processing on a three-dimensional coordinate system at a specified viewing angle with respect to the voxel array;
For each voxel value V (x, y, z) corresponding to the (x, y, z) coordinates after execution of the rotation process, only when the z value has a value closer to the viewpoint than the value recorded in the Z buffer. While updating the Z buffer with the z value, a process of reflecting the voxel value corresponding to each z value to the corresponding pixel (x ′, y ′) of the luminance frame is performed, and the luminance of each pixel of the luminance frame Projection processing means for calculating a value V (x ′, y ′);
Gradation conversion means for performing a predetermined calculation in order to match the luminance value of each pixel of the luminance frame calculated by the projection processing means with the output gradation of the display means;
A voxel array visualization device characterized by comprising: In claim 1,
For the luminance value of each pixel of the luminance frame obtained by the projection processing means, a normal vector is calculated based on the value of the Z buffer of the pixel and surrounding pixels, and the inner product with a preset parallel light source vector Further comprising a shadow processing means for multiplying the shadow luminance value of the pixel calculated as described above, and the gradation converting means performs processing on the luminance value of each pixel of the luminance frame processed by the shadow processing means. A voxel array visualization device characterized by being. In claim 1 or claim 2,
When the projection processing unit reflects the voxel value corresponding to each z value to the corresponding pixel (x, y) on the luminance frame, the projection processing unit is based on the distance between the Z buffer value and the z value at that time. A voxel array visualization device characterized by determining a value to be reflected. In any one of Claims 1-3,
The viewpoint angle is specified by setting three kinds of angles, a rotation angle on the XY plane, a rotation angle on the XZ plane, and a rotation angle on the YZ plane in a three-dimensional coordinate system, and the voxel array rotation means is A voxel array visualization device that sequentially performs rotation on the three planes according to a set order. In any one of Claims 1-4,
A view angle calculation means for continuously specifying the view angle in time series and designating a plurality of the view angles;
The voxel array visualization device, wherein the voxel array rotation means executes processing for a specified viewpoint angle. In any one of Claims 1-5,
A plurality of voxel arrays are sequentially extracted within a specified time range from a voxel array set in which a plurality of voxel arrays are defined in time series in advance, and processing is performed on each of the extracted voxel arrays. A voxel array visualization device that executes and sequentially calculates a luminance value for each voxel array.   The program for functioning a computer as a voxel arrangement | sequence visualization apparatus in any one of Claims 1-6.

Images