JP2004049753A - Three-dimensional image display device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a three-dimensional image display device in which the three-dimensional image of a display object formed on the basis of multi-slice image information collected by an X-ray CT (computed tomography) apparatus, a magnetic resonance diagnostic apparatus, or the like, is rotation displayed at a high speed, further, the three-dimensional image can be easily choose as a two-dimensional one when a desired position is chosen. <P>SOLUTION: This device comprises a multi-slice image inputting section A which inputs slice image information (a), a voxel data structure preparation section B which forms and stores the three-dimensional image of a display object (s) based on the slice image information (a), a view flat surface rotation processing part C which rotates a view flat surface 10 to indicate the three-dimensional image from a given direction for a voxel data structure (b) stored in the voxel data structure preparation section B, a display object scanning processing part D for scanning the display object (s) within the three-dimensional image from the view flat surface 10, and a display portion E for indicating a scanned display object (s). <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a three-dimensional image display device that rotates and displays a three-dimensional image of a display object based on multi-slice image information collected by an X-ray CT (Computed Tomography) device, a magnetic resonance diagnostic device, or the like at a high speed. About. Further, the present invention relates to a three-dimensional image display device which allows a user to easily select a desired position in a three-dimensional image as a two-dimensional image.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, as a medical image, a three-dimensional image is effective for observing a biological tissue having a complicated shape. This is because it is possible to observe complicated shapes of living tissues from various directions. However, it is important to set a direction in which the affected part can be examined well or a direction in which the traveling of blood vessels and the size and extent of the tumor can be understood. To do so, it is important for the doctor to be able to change the direction in which he looks at the three-dimensional image displayed on the personal computer.
[0003]
Some of the difficulties in operating three-dimensional images cannot be compared with those in operating two-dimensional images. For example, in a two-dimensional operation, it is possible to easily observe a human body part (e.g., a face or an organ of an ecological tissue) shown on a display unit of an image such as a CRT by changing its direction on a plane with a mouse, for example, with great freedom. it can. However, in a three-dimensional space, when it is desired to see a desired position, it is not easy to change the face direction as it is. This is because, like a two-dimensional image, in addition to the operation of the X and Y coordinate axes, it is necessary to match the depth with the Z coordinate axis. It is extremely difficult to accurately match the X, Y, and Z axes, and it is not well connected to human intuition, so that the operation takes time. In the operation of changing the face in the desired direction, for example, the coordinates of the X, Y, and Z axes have been input from the keyboard and the direction has been finely adjusted so far. It took a lot of trouble. Still, I often thought I couldn't turn it around. There is also a method of individually moving the X, Y, and Z axes using a mouse or the like to adjust the direction, but this has been a considerable burden on doctors unfamiliar with personal computers.
[0004]
Although it is difficult to operate the three-dimensional image, as shown in FIG. 8C, (1) first, two axes of X and Y can be easily and correctly determined (see S21), but the Z axis can be easily determined. In this case, the position cannot be specified as desired, and it must be repeated by trial and error (see S22), and the burden on the operator is large. {Circle around (2)} Specifically, after the Z-axis is determined, the position is visually checked (see S23), and each time it is clicked, this is repeated several times usually, and finally the Z-axis is determined. It is almost determined at the desired predetermined position (see S24). In spite of such a great effort, the accuracy is not high.
[0005]
A conventionally known three-dimensional medical imaging apparatus that operates as described above will be described. In an X-ray CT apparatus, a magnetic resonance diagnostic apparatus, or the like as the three-dimensional medical image apparatus, as shown in FIGS. 8A and 8B, a display object such as a human body part (a face or an organ of an ecological tissue) or the like is displayed. Based on the collected multi-slice image information (continuous tomographic image) a, the object s forms a voxel data structure b that forms a three-dimensional image by a voxel (Volume Element) method. When an operator such as a doctor wants to view the voxel data structure b from an arbitrary direction and rotates and displays the three-dimensional display voxel data structure b, each voxel c shown in FIG. It performs affine transformation processing d on itself, finds the position of each voxel c after rotation, and places this voxel c in the memory buffer e after this operation (the cumbersome and persistent operation as described above).
[0006]
However, this method requires the same number of memory buffers e as the number of voxels c of the voxel data structure b and an enormous amount of rotation processing time. To briefly explain this structure, for example, in order to change the position of the face, the consistency of the X, Y, and Z axes is adjusted to convert a huge voxel data structure b into an affine transformation d. By adjusting the degree and changing the angle, position, and the like of the face, it is possible to confirm the position of the face, which is appropriate or nearly appropriate when viewed from the display unit of the image such as the CRT.
[0007]
[Problems to be solved by the invention]
However, as described above, in order to rotate and display the voxel data structure b itself to be displayed, a memory buffer e of the same number as the number of voxels c and an enormous amount of rotation processing time are conventionally required. Furthermore, selecting a desired position in a three-dimensional image requires considerable skill and skill. For this reason, even a doctor unfamiliar with a personal computer can easily change the direction of a human body part (face or organ of an ecological tissue) shown on the display part of an image such as a CRT on a plane with a mouse, for example, quite freely. It was an urgent task to be able to observe them. Therefore, the present invention solves such a problem, that is, a voxel data structure can be processed at a high speed even with a relatively small-capacity personal computer, and even an inexperienced doctor can see a desired position. Another object of the present invention is to make the operation as simple as the operation of a two-dimensional image.
[0008]
[Means for Solving the Problems]
The inventor of the present invention has made intensive studies to solve the above-mentioned problems. As a result, the present invention fixes the voxel data structure to be rotated, rotates the view plane itself for projecting and displaying the voxel data structure, and displays the voxel data structure. It is an object of the present invention to provide a three-dimensional image display device which realizes a memory buffer and a rotation processing time independent of a target voxel data size and can be operated as easily as a two-dimensional image.
[0009]
Specifically, the present invention provides a multi-slice image input unit that inputs slice image information, a voxel data structure generation unit that generates and stores a three-dimensional image of a display target based on the slice image information, A view plane rotation processing unit for rotating a view plane to display a three-dimensional image from an arbitrary direction with respect to the voxel data structure stored in the voxel data structure generation unit, and a display object in the three-dimensional image from the view plane By using a three-dimensional image display device including a display object scanning processing unit that scans the object and a display unit that displays the scanned display object, the voxel data structure of the rotation object is fixed, and the voxel By rotating the view plane that projects and displays the data, the rotation processing time can be significantly reduced, and In the original image, when selecting a desired position is one that can be easily selected as a two-dimensional.
[0010]
Further, in the present invention, a trajectory sphere for the voxel data structure is created, and a circumcircle of the trajectory sphere is set as a view plane which is rotated, and the movement on the view plane is performed by a three-dimensional image display device by a screen moving means. Accordingly, when a desired position is selected in a three-dimensional image, the position can be selected as two-dimensional, and the operation can be easily performed.
[0011]
[Action]
The operation of the present invention will be described. First, multi-slice image information is input and converted into a voxel data structure. Thereafter, the view plane is rotated by the view plane rotation means from the display unit without performing affine transformation on the part to be viewed with respect to the display object, and the part to be viewed is determined, and the display object scan is performed. By converting the display object into a two-dimensional image that can be viewed from the view plane by the smoothing, and displaying the display object on the display unit, the three-dimensional image can be rotated and displayed at high speed.
[0012]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a system diagram of the present invention. The three-dimensional image display device of the present invention generates a three-dimensional image of a display object s based on the multi-slice image input unit A for inputting slice image information (continuous tomographic images) a and the slice image information a. A voxel data structure generation unit B to store, a view plane rotation processing unit C that rotates the view plane 10 to display a three-dimensional image from an arbitrary direction, and a display object s in the three-dimensional image from the view plane 10 It comprises a display object scanning processing section D for scanning and a display section E for displaying the scanned display object, and is controlled by a central processing unit (CPU or MPU).
[0013]
The multi-slice image input unit A of the three-dimensional image display device outputs multi-slice image information (continuous tomographic image) a of the subject (continuous tomographic image) a (see FIG. 8A) output from an X-ray CT device, a magnetic resonance diagnostic device, or the like. What you enter. The voxel data structure generation unit B includes a voxel data structure generation unit 1 that generates a voxel data structure b that forms a three-dimensional image by a voxel method based on the slice image information a, and a voxel data structure storage unit 2. It is configured. In particular, since the voxel data structure b is composed of a three-dimensional image, a CPU corresponding to its capacity is required. The multi-slice image input unit A and the voxel data structure generation unit B described above are known technologies.
[0014]
The view plane rotation processing unit C includes a view plane 10, on-screen moving unit 11, and view plane rotation unit 12. As shown in FIGS. 3 to 5, the view plane 10 is a sphere circumscribing the eight vertices of the voxel data structure b (cubic shape), which is a trajectory sphere G, and indicates a plane that matches the device screen output. is there. Specifically, the voxel data structure b in which the multi-slice image information a [see FIG. 7A] is stacked has a cubic or rectangular parallelepiped shape. The sphere is defined as an orbit sphere G. The rotation of the view plane 10 is realized such that the view plane 10 is in direct contact with the orbit sphere G. The on-screen moving means 11 is basically a mouse, but also includes a pen display or a manual finger.
[0015]
The view plane rotation processing section C, which is a feature of the present invention, performs the voxel data structure b in a stationary state without rotating the entire voxel data structure b by the affine transformation processing d as in the prior art. The configuration is such that the view plane 10 circumscribing the orbit sphere G is rotated. The view plane 10 is provided in a display unit E such as a CRT. When the view plane 10 in the display unit E is moved by the on-screen moving means 11, the view plane 10 itself rotates and moves to the voxel. The position to be viewed in the data structure b is projected and displayed on the display unit E. At this time, the voxel data structure b, which has conventionally been rotated, is fixed in the present invention. Such a configuration enables the computer used in the present invention to have a relatively small capacity.
[0016]
Although the view plane 10 is provided in the display unit E as described above, the view plane 10 rotates the circumscribed orbit sphere G, and the state before rotation (movement) to make it easier to understand the operation state. Is the view plane 10a before rotation, and the view plane 10 in the state after rotation is the view plane 10b after rotation. As a result, there are a pre-rotation view plane 10a and a post-rotation view plane 10b as subordinate concepts of the view plane 10. As described above, the view plane rotating means 12 is provided to rotate the view plane 10 appropriately while circumscribing the orbit sphere G in order to rotate from the pre-rotation view plane 10a to the post-rotation view plane 10b.
[0017]
Further, as shown in FIG. 3, on the pre-rotation view plane 10a, when the operator tries to move on the screen on the mobile device 11, firstly, the starting point O ₁ of said screen on the mobile device 11 orbital sphere G In the direction of the normal from the center point Oa. That is, the direction vector in the open starting point O ₁ is formed toward the center point Oa of the track ball G. And basic axes P ₁ of the straight line connecting said start point O ₁ and the center point Oa. When moving on the screen on the mobile device 11, the straight line made in the center Oa of the destination point O ₂ and the track ball G and the destination axis P _2. The mobile moves away shaft P ₂ and the basic axis P ₁ (rotation) should do solid angle θ is determined, the operation processing so that the rotation angle, after the rotation through the view plane rotation processing unit C View The plane 10b is generated.
[0018]
In particular, the essential contents of the present invention will be further described in detail with reference to FIG. The point is that by moving the secondary plane on the display unit E with a mouse or the like, the view plane 10 provided in the display unit E is rotated appropriately while being in contact with the orbital sphere G. This is a structure in which a desired position of the voxel data structure b as a three-dimensional image fixed inside can be seen. First, with respect to the center point Oa of the orbit sphere G, the origin (start point O ₁ ) of the pre-rotation view plane 10a is now oriented in the normal direction as a direction vector. That is, the view plane 10 is configured as a surface on the normal line from the center point Oa of the orbit sphere G. To the axis and basic axis P _1. A point on the pre-rotation view plane 10a and in contact with the orbit sphere G is defined as O ₁ (x, y, z) as a point on the orbit sphere G. Then, it is moved on the pre-rotation view plane 10a via the on-screen moving means 11 such as a mouse. Then, the destination point _{O 2} is on the pre-rotation view plane 10a, the (X, Y, Z). A line connecting the destination point O ₂ (X, Y, Z) and the center Oa of the orbit sphere G is the destination axis P ₂ . By a said destination axis P ₂ and the main axes P _1, the solid angle θ should be moved is determined, a fixed value.
[0019]
Further, if the constituent members are simply described as an example, first, a view plane 10 formed of a styrene foam plate and an orbit ball G also formed of a styrene foam plate will be described. At this time, it is more preferable that both the view plane 10 and the orbit sphere G have a transparent or translucent material. Such track ball G is circumscribed view plane 10 in, and a point O ₁ of the contact point, stabbed first skewer made basic axes P ₁ from the top of the point O _1, then the destination on the view plane 10 When stabbing a second skewer comprising destination axis P ₂ from the point O ₂ places, ∠O ₁ both direction vector is determined by the fact that forms the center Oa of the track ball G, and its first skewer and second skewers The solid angle θ of OaO ₂ is a fixed value.
[0020]
Then, the solid angle θ is changed to a rotation angle θ ₁ on the YZ plane (FIG. 5A), a rotation angle θ ₂ on the XZ plane [FIG. 5B], and a rotation angle θ ₃ on the XY plane [FIG. 5 (C)], assuming that the orbital sphere G is rotated with respect to these planes and has a constant value of a radius R, an arbitrary destination point (X ₀ , Y ₀ , Z ₀ ) in the following case The point location is a position facing the normal direction to the center Oa (origin) of the orbital sphere G. That is, after the confirmation operation (such as clicking), the destination point O ₂ (X, Y, Z) on the post-rotation view plane 10b becomes the origin on the orbit sphere G, and the post-rotation view plane 10b becomes the orbit sphere G It will be on the normal from the upper origin.
[0021]
What is particularly important in the present invention is that when rotating within the range of the solid angle θ, the destination point O ₂ (X, Y, Z) arbitrarily selected on the pre-rotation view plane 10a and the post-rotation view plane 10b Is different from the point O ₂ ′ (X, Y, Z) on the orbiting sphere G, but the direction vector of any normal on the view plane 10 is directed to the center Oa (origin) of the orbiting sphere G. Therefore, on the pre-rotation view plane 10a, the movement destination point O ₂ (X, Y, Z) is, for the operator viewed from the display unit E, a view of O ₂ ′ on the orbit sphere G. Can be the same as That is, the operation on the secondary plane can perform a three-dimensional operation of the orbit sphere G. More specifically, the movement destination point O ₂ (X, Y, Z) is a point in the arithmetic calculation, and the point visible to the operator is substantially the point O ₂ ′ (X, Y) on the orbit sphere G. , Z) is the maximum point.
[0022]
Further, the view plane rotation processing unit C will be described in detail using some mathematical expressions and the like. First, the state in which the view plane 10 was provided for the orbit sphere G of the voxel data structure b was set as an initial state as a three-dimensional space [see the solid line in FIG. 3A and the solid line in FIG. 4]. In this initial state, in FIG. 4, a diagram viewed from the direction A is a YZ plane, a diagram viewed from the direction B is an XZ plane, and a diagram viewed from the direction C is an XY plane. At this time, when the initial screen is generated, the XYZ coordinate values after the movement of the view plane 10 are determined as follows.
Here, the view plane (i, j) = ( _Xc− N / 2 + i, _Yc + M / 2 + j, _Zc− R), and 0 ≦ i ≦ N and 0 ≦ J ≦ M.
[0023]
Then, the rotation angles θ ₁ , θ ₂ , θ ₃ after the rotation of the movement destination axis P ₁ are obtained. For this, the three-dimensional coordinate values of all the view planes 10 are coordinate-transformed by θ ₁ , θ ₂ , and θ ₃ obtained above. In this coordinate conversion mechanism, x, y, and z are XYZ coordinate values before each conversion processing, and X, Y, and Z are coordinate values after each conversion processing. Then
_{X 1 = x, Y 1 =} yCos θ 1 -zSin θ 1, Z 1 = ySin θ 1 + Z 1 Cosθ 1 , and the
_{X 2 = X 1 Cosθ 2 -Z} 1 Sinθ 2, Y 2 = Y 1, a _{Z 2 = X 1 Sinθ 2 +} Z 1 Cosθ 2.
_Furthermore, the _{X = X 2 Cosθ 3 -Y 2} Sinθ 3 Y = X 2 Cosθ 3 -Y 2 Sinθ 3, Z = Z 3.
[0024]
As the address conversion mechanism after the rotation processing, X, Y, and Z are the three-dimensional space coordinate values of the coordinates (i, j) of each view plane 10 obtained by the coordinate conversion mechanism, and the coordinates (i, j) of the view plane 10 ) Does not change. As the structure of the view plane 10, the array (voxel) number is the coordinate value on the view plane 10, and the coordinate value of the voxel in the three-dimensional space is stored in each array.
For example, as an example of mouse dragging, it is assumed that the mouse moves from view plane coordinate values (2, 4), which are coordinate values of α and β systems, to (4, 2). The coordinate values of the α, β system and the coordinate values of the X, Y, Z system do not match. The coordinate values of the α-β coordinates specified by the mouse and the coordinate values after the mouse is moved are automatically converted into addresses in the X, Y, Z coordinate space by the address conversion mechanism.
[0025]
As shown in FIG. 6, the display object scanning processing unit D includes, in the rotated view plane 10 b, a straight line perpendicular to the rotated view plane 10 b from each voxel, that is, the display object scan line 20 and the voxel data structure. The intersection of b with the display object s is obtained. Further, scanning is performed until the data comes out of the voxel data structure b. At this time, scanning is not performed when the display object scanning line 20 from each voxel does not intersect with the display object s of the voxel data structure b.
[0026]
The operation of the present invention will be described based on the flowchart shown in FIG. First, multi-slice image information a is input (see S1), a voxel data structure b is generated (see S2), and the voxel data structure b is completed (see S3). Then, a desired position (a desired position) for the display object s in the voxel data structure b is selected and started (see S4). The display unit E moves to a desired position on the pre-rotation view plane 10a by the on-screen moving unit 11 (see S5). After confirming the position, a confirmation operation (click, etc.) is performed (see S6). Immediately thereafter, the pre-rotation view plane 10a is rotated by the view plane rotation means 12 by a predetermined calculation process, and the position after movement is determined as the post-rotation view plane 10b (see S7). Then, the display object is scanned to convert the display object s into a two-dimensional image visible from the view plane 10 (see S8), and the display object s is displayed on the display unit E (see S9). Actually, the processes from the position confirmation (S6) to the display on the display unit E (S9) are instantaneously performed by computer processing. That is, the conversion of the three-dimensional image to the desired position can be rotated and displayed at high speed.
[0027]
Further, according to the present invention described in detail above, when the X, Y, and Z axis sizes of the target voxel data are fixed to N and the rotation time calculation amount is represented by O-notation as an asymptotic evaluation, the target voxel data is obtained. The method of affine transformation of the data itself is O (N ³ ). However, in the present invention, even when there is a large difference in the voxel size in each axis direction of the voxel data, the voxel size obtained by squaring the diagonal line of the worst voxel data is sufficient, so the following equation ((3N ² ) ^1/2 ) ² = Since 3N ² holds, the result is O (N ² ).
[0028]
To prove that the above theoretical values are correct, a description will be given with reference to the drawings. The graph of FIG. 7 shows a logarithmic graph for comparing the rotation processing time between the apparatus according to the present invention and the conventional apparatus, wherein the horizontal axis represents the number N of voxels on one side of the voxel data structure, and the vertical axis represents the rotation processing time T. . The object data was prepared with a cubic structure and a total number of voxels of 1,000 to 27,000,000 at intervals of 1,000 voxels, and the rotation processing time of the apparatus according to the embodiment of the present invention and the rotation processing time of the conventional apparatus were calculated for the voxel data. However, a line graph 30 of the rotation processing time in the example apparatus of the present invention and a line graph 31 of the rotation processing time in the conventional apparatus were obtained. As described above, the present invention can provide a memory buffer independent of the voxel data size to be displayed and a high-speed three-dimensional image display device that realizes a rotation processing time.
[0029]
【The invention's effect】
In the present invention, a relatively large-capacity voxel data structure b to be rotated is fixed, and the view plane 10 itself for projecting and displaying the voxel data structure b is rotated to relatively reduce the CPU capacity. There is also an advantage that the voxel data structure can be processed at high speed. Furthermore, when a desired position is selected in a three-dimensional image, a certain level of skill is required. In the present invention, the three-dimensional image can be grasped as a two-dimensional image on the view plane 10. Therefore, there is an advantage that the selection of a desired position can be greatly simplified. This makes it very easy for an unskilled physician to see the desired position, as with manipulating a two-dimensional image.
[Brief description of the drawings]
Figure 1 is a flowchart of the block diagram of the system showing the overall invention the present invention; FIG 3 (A) is a perspective view showing a main part (B) of the present invention to Q ₁ (A) - Q ₁ arrow sectional view (C) is a perspective view showing an operating state of the view plane rotation processing unit comprising a main portion of Q ₂ -Q ₂ cross-sectional view taken along [4] the present invention Figure 5 of (a) ( A) is a YZ plane view of FIG. 4 viewed from direction A. FIG. 4B is an XZ plane view of FIG. 4 viewed from direction B. FIG. 4C is an XY plane view of FIG. 4 viewed from direction C. FIG. 6 is a state diagram showing an operation state of a display object scanning unit according to the present invention; FIG. (A) is a simplified diagram of generating a voxel structure from multi-slice image information of the prior art. (B) is a process of voxel data rotation processing of the prior art. (C) is a flow chart DESCRIPTION OF REFERENCE NUMERALS conventional operating conditions
A multi-slice image input unit B voxel data structure generation unit C view plane rotation processing unit D display object scanning processing unit E display unit G orbit sphere 10 view plane

Claims (4)

A multi-slice image input unit for inputting slice image information, a voxel data structure generation unit for generating and storing a three-dimensional image of the display object based on the slice image information, and a voxel data structure generation unit A view plane rotation processing unit that rotates a view plane to display a three-dimensional image from an arbitrary direction with respect to the voxel data structure, and a display object scanning unit that scans a display object in the three-dimensional image from the view plane A three-dimensional image display device, comprising: a processing unit; and a display unit that displays a scanned display target. 3. The tertiary tertiary system according to claim 1, wherein a trajectory sphere for the voxel data structure is created, and a circumscribed circle of the trajectory sphere is set as a view plane in which the circumcircle is rotated, and the movement in the view plane is performed by an on-screen moving means. Original image display device. 3. The three-dimensional image display device according to claim 2, wherein the orbit sphere is inscribed at each vertex of the substantially cubic voxel data structure. 3. The three-dimensional image display device according to claim 2, wherein the on-screen moving means is operated by a mouse.

Images