Image processing apparatus, image capturing apparatus, program, and image processing method
The image processing device reconstructs three-dimensional organ shapes from two orthogonal projection data, addressing the limitations of SPECT and PET by minimizing patient confinement and processing time, and lowering costs.
Patent Information
- Application Number
- JP2024032383
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-03-04
- Publication Date
- 2025-09-17
AI Technical Summary
Existing nuclear medicine tests like SPECT and PET require prolonged patient confinement and extensive image processing to reconstruct three-dimensional organ shapes, and AI-based methods are costly due to the need for large training datasets.
A method to reconstruct three-dimensional shapes from a small amount of projection data using an image processing device that captures images from two orthogonal directions, identifies key points and diameters, and generates curves to overlay cross-sectional shapes, optionally with preprocessing to handle abnormal pixel values.
This approach allows for rapid reconstruction of high-resolution three-dimensional organ shapes with reduced patient confinement and processing time, while reducing development costs.
Smart Images
Figure 2025134457000001_ABST
Abstract
Description
[Technical Field]
[0001] This invention relates to a method for restoring the three-dimensional shape of an object such as an organ from projection data of a medical image. [Background technology]
[0002] Understanding the shape of a patient's organs is useful in clinical practice. For example, in liver disease, the shape and function of the liver change dynamically (over time) with the condition of the disease, so it is known that evaluation of liver function using three-dimensional images is desirable.
[0003] It is known that nuclear medicine examinations such as SPECT and PET can reconstruct three-dimensional images from the captured projection image data. Cross-sectional images are reconstructed from the two-dimensional projection data captured in SPECT and PET examinations, and three-dimensional images are reconstructed using the cross-sectional images. This makes it possible to reconstruct the three-dimensional shape of the patient's organs. [Prior art documents] [Non-patent literature]
[0004] [Non-Patent Document 1] Yu Sun, et al. “Coordinate-based Internal Learning for Imaging Inverse Problems”, IEEE Transactions on Computational Imaging. Vol.7, 2021 [Non-patent document 2] Liyue Shen, et al. "Implicit Neural Representation Learning with Prior Embedding for Sparsely Sampled Image reconstruction", IEEE Transactions on Neural Networks and Learning Systems, 2022 [Non-patent document 3] Kunal Gupta, et al. "Neural Computed Tomography" [online] January 17, 2022 Cornell University arXiv Internet<https: / / doi.org / 10.48550 / arXiv.2201.06574> Summary of the Invention [Problem to be solved by the invention]
[0005] However, in nuclear medicine tests such as SPECT and PET, radiation emitted by a patient who has been administered a radioactive drug is collected as projection data using a detector that rotates around the body, which not only requires the patient to be restrained for a long time, but also requires time for image processing.
[0006] Recently, methods for restoring three-dimensional shapes from two-dimensional images (planar images) using AI have been reported (Non-Patent Documents 1 to 3). However, since methods using AI require a large amount of training data, the development costs are very high, and a low-cost three-dimensional shape restoration process is needed.
[0007] In view of the above background, an object of the present invention is to provide a technique for reconstructing the three-dimensional shape of an object from a small amount of projection data. [Means for solving the problem]
[0008] (Configuration 1) The image processing device of configuration 1 includes an input unit that inputs first projection data and second projection data of an object photographed from different directions, and a calculation unit that reconstructs a three-dimensional shape of the object based on the first projection data and the second projection data, wherein the calculation unit identifies a first diameter and its endpoints of the object at a predetermined height imaged in the first projection data, and a second diameter and its endpoints of the object at the predetermined height imaged in the second projection data, identifies a first point in the first diameter with the highest pixel value, and a second point in the second diameter with the highest pixel value, intersects the first diameter and the second diameter so that the second diameter passes through the first point and the second point on a virtual plane obtained by slicing the object at the predetermined height, generates a curve that passes through the endpoints of the first diameter and the second diameter, changes the predetermined height to generate a curve for each height, and overlays the generated curves in the height direction to reconstruct the three-dimensional shape of the object.
[0009] (Configuration 2) In the image processing device of configuration 1, the calculation unit may mask pixels having pixel values higher than a predetermined threshold in the image of the object captured in the first projection data and the image of the object captured in the second projection data as a preprocessing step for restoring the three-dimensional shape of the object.
[0010] (Configuration 3) In the image processing device of configuration 1, the calculation unit may smooth the image of the object reflected in the first projection data and the image of the object reflected in the second projection data as a preprocessing step for restoring the three-dimensional shape of the object.
[0011] (Configuration 4) In the image processing device of configuration 1, the calculation unit may erase an image of a specified area from the image of the object reflected in the first projection data and the image of the object reflected in the second projection data as a preprocessing step for restoring the three-dimensional shape of the object, and restore the image of the erased area using images of its surroundings.
[0012] (Configuration 5) In the image processing device of any of configurations 1 to 4, the calculation unit may change the predetermined height and identify the first point and the second point for each height, correct the position of the first point so that the first points are smoothly connected in the height direction, correct the position of the second point so that the second points are smoothly connected in the height direction, and intersect the first diameter and the second diameter using the corrected first point and the corrected second point.
[0013] (Configuration 6) In the image processing device of any one of configurations 1 to 5, the first projection data and the second projection data may be projection data of the object acquired by photographing from two orthogonal directions.
[0014] (Configuration 7) An image capturing device of configuration 7 includes an image capturing unit that captures images of an object from different directions to acquire first projection data and second projection data, and a calculation unit that reconstructs a three-dimensional shape of the object based on the first projection data and the second projection data, wherein the calculation unit identifies a first diameter and its endpoints of the object at a predetermined height that are captured in the first projection data, and a second diameter and its endpoints of the object at the predetermined height that are captured in the second projection data, identifies a first point in the first diameter that has the highest pixel value, and a second point in the second diameter that has the highest pixel value, intersects the first diameter and the second diameter so that the second diameter passes through the first point and the second point on a virtual plane that slices the object at the predetermined height, generates a curve that passes through the endpoints of the first diameter and the second diameter, changes the predetermined height to generate a curve for each height, and overlays the generated curves in the height direction to reconstruct the three-dimensional shape of the object.
[0015] (Configuration 8) A program of configuration 8 is a program for reconstructing a three-dimensional shape of an object from first projection data and second projection data of the object photographed from different directions, and causes a computer to execute the following steps: inputting the first projection data and the second projection data; specifying a first diameter and its endpoints of the object at a predetermined height imaged in the first projection data, and a second diameter and its endpoints of the object at the predetermined height imaged in the second projection data; specifying a first point in the first diameter with the highest pixel value and a second point in the second diameter with the highest pixel value; intersecting the first diameter and the second diameter on an imaginary plane obtained by slicing the object at the predetermined height so that the second diameter passes through the first point and the second point, and generating a curve passing through the endpoints of the first diameter and the second diameter; and changing the predetermined height to generate a curve for each height, and overlaying the generated curves in the height direction to reconstruct the three-dimensional shape of the object.
[0016] (Configuration 9) An image processing method of configuration 9 is an image processing method for restoring a three-dimensional shape of an object from first projection data and second projection data of the object photographed from different directions by an image processing device, the image processing device comprising the steps of: inputting the first projection data and the second projection data; specifying a first diameter and its end points of the object at a predetermined height imaged in the first projection data; and specifying a second diameter and its end points of the object at the predetermined height imaged in the second projection data; and specifying a radius of the object at the first diameter and its end points of the object at the predetermined height imaged in the second projection data. The method comprises the steps of: identifying a first point with the highest pixel value and a second point within the second diameter with the highest pixel value; intersecting the first diameter and the second diameter in a virtual plane obtained by slicing the object at the predetermined height, so that the second diameter passes through the first point and the second point, and generating a curve passing through an end point of the first diameter and an end point of the second diameter; and varying the predetermined height, generating a curve for each height, and overlaying the generated curves in the height direction to reconstruct the three-dimensional shape of the object. [Effects of the Invention]
[0017] According to the image processing device of the present invention, the three-dimensional shape of an object can be reconstructed from the first projection data and the second projection data. [Brief explanation of the drawings]
[0018] [Figure 1] 1 is a diagram illustrating a configuration of an image capturing device according to an embodiment. [Figure 2] FIG. 10 is a diagram showing an outline of a process for reconstructing a three-dimensional shape from two pieces of projection data. [Figure 3] FIG. 10 is a diagram showing a process for generating a cross-sectional shape when an object is sliced by height. [Figure 4] 10 is a flowchart showing the flow of image processing by the image processing device. [Figure 5] FIG. 10 is a diagram illustrating a process performed by a calculation unit in an image processing device according to a second embodiment. [Figure 6] 10 is a flowchart showing the flow of image processing by the image processing device according to the second embodiment. [Figure 7] FIG. 10 is a diagram showing the results of a comparison between the volume of the three-dimensional shape of the liver reconstructed from two projection data using the image processing method of this embodiment and the volume of the three-dimensional shape of the liver reconstructed from a SPECT image. [Figure 8] FIG. 10 is a diagram showing an example of a three-dimensional shape of a liver reconstructed from two pieces of projection data. DETAILED DESCRIPTION OF THE INVENTION
[0019] The image processing device of this embodiment will be described below with reference to the drawings. Note that the following description is merely an example of a preferred embodiment and is not intended to limit the scope of the invention as defined in the claims.
[0020] 1 is a diagram showing the configuration of an image capturing device 1 according to an embodiment. The image capturing device 1 includes an image capturing device 10 that captures an image of a subject, and an image processing device 20 that performs image processing of projection data obtained by the image capturing device 10.
[0021] The imaging device 10 is a device that captures medical images of a patient, and is a device that performs imaging using SPECT, PET, CT, X-rays, etc. When imaging using SPECT, PET, CT, etc., a camera usually moves around the patient to capture multiple images from all directions. The image processing device 20 of this embodiment can reconstruct the three-dimensional shape of an object from two pieces of projection data captured from different directions, so it is sufficient to perform imaging from two directions. In the following description, the two pieces of projection data will be referred to as "first projection data" and "second projection data."
[0022] Preferably, the imaging device 10 is a nuclear medicine imaging device used in nuclear medicine examinations, and the first projection data and the second projection data are two-dimensional nuclear medicine data obtained by detecting radiation emitted from a radiopharmaceutical administered to the body of a subject. In other words, the first projection data and the second projection data may be planar nuclear medicine images captured from different directions.
[0023] The image processing device 20 includes an input unit 21, a calculation unit 22, an output unit 23, and a storage unit 24. The input unit 21 has a function of accepting input of first projection data and second projection data captured by the imaging device 10. In this embodiment, the first projection data and second projection data are two pieces of projection data captured from orthogonal directions. The calculation unit 22 performs calculation processing to reconstruct the three-dimensional shape of the object based on the first projection data and the second projection data. The output unit 23 has a function of outputting the reconstructed three-dimensional shape. The storage unit 24 has a function of storing the input projection data, data on the reconstructed three-dimensional shape, etc.
[0024] Here, we will explain the object whose three-dimensional shape is to be restored by the image processing device 20. The object whose three-dimensional shape is to be restored is a part on which the user wishes to focus, and one example is an organ. The user may set a region of interest in the part on which the user wishes to focus, and the region of interest may be used as the object to be restored.
[0025] For example, in the case of liver receptor scintigraphy, the area of interest is the area where asialoglycoprotein labeled with a radioisotope (e.g., Tc-99m) specifically accumulates, and the target is the liver. In the case of dopamine transporter scintigraphy, isoflurane ( 123 The striatum (dopamine transporter), where radioactive tracers with affinity for dopamine transporters such as I) specifically accumulate, is the area of interest, and the target is the striatum. However, the target is not necessarily limited to organs. In the case of liver receptor scintigraphy, the contour of the entire body can be made the target by setting a threshold on the projection data to suppress the upper end of the count. In the case of dopamine transporter scintigraphy, the contour of the entire head can be made the target by setting a threshold on the projection data to suppress the upper end of the count.
[0026] [Restore process] FIG. 2 is a diagram showing an overview of the process of reconstructing a three-dimensional shape from two pieces of projection data. In FIG. 2, projection data from liver receptor scintigraphy is used as an example, and the object is the liver. The two images in the upper row are the first projection data and the second projection data, respectively. When these two pieces of projection data are drawn in three-dimensional space, the image shown in the lower row is obtained, and the shape of the object that formed the basis of the two pieces of projection data can be reconstructed. This is an overview of the process performed by the image processing device of this embodiment. The image processing device of this embodiment appropriately reconstructs the three-dimensional shape using not only the contours of the object in the two two-dimensional images, but also pixel values. Next, the reconstruction process by the calculation unit 22 will be described with reference to FIG. 3.
[0027] The calculation unit 22 generates cross-sectional shapes by slicing the object on a plurality of virtual planes at different heights, and reconstructs the entire object by superimposing the cross-sectional shapes at each height. Here, height refers to a direction perpendicular to the plane formed by the two directions in which the first projection data and the second projection data were captured. For example, if the plane formed by the directions in which the first projection data and the second projection data were captured is a horizontal plane (i.e., the images were captured from two horizontal directions), height is the vertical direction.
[0028] In the image in Figure 3, the height is specified by marking the vertical axis with 0, 20, 40, 60, and 80. For example, by reconstructing a cross-sectional shape sliced at height 1, a cross-sectional shape sliced at height 2, ..., a cross-sectional shape sliced at height 99, and stacking these, the three-dimensional shape of the target liver is reconstructed. The height used to generate the cross-sectional shapes shown here does not have to be in increments of 1. Reconstruction can be more accurate by dividing the height into smaller increments.
[0029] FIG. 3 focuses on a height of 50 and shows the process of generating a cross-sectional shape when an object is sliced at a height of 50. When the first projection data is sliced at a height of 50, a straight line L1 with points a and b as its endpoints is obtained. The straight line L1 is the diameter of the object at a height of 50 as viewed from the shooting direction of the first projection data. When the second projection data is sliced at a height of 50, a straight line L2 with points c and d as its endpoints is obtained. The straight line L2 is the diameter of the object at a height of 50 as viewed from the shooting direction of the second projection data.
[0030] Since the shooting directions of the first projection data and the second projection data are orthogonal, the lines L1 and L2 are also orthogonal. The point of intersection when the lines L1 and L2 are orthogonal is determined as follows. The point on line L1 with the highest pixel value is defined as point m. The highest pixel value means that it is the thickest part of the object when viewed from the orthogonal direction. Therefore, line L2 intersects with line L1 at point m. Similarly, if the point on line L2 with the highest pixel value is defined as point n, line L1 intersects with line L2 at point n. From the above, the intersection P of lines L1 and L2 can be determined.
[0031] By the above process, when focusing on the position at height 50, it is possible to draw perpendicular lines L1 and L2, as shown in the middle diagram of Figure 3. Next, as shown in the bottom diagram of Figure 3, a spline curve SP is drawn that passes through endpoints a and b of line L1 and endpoints c and d of line L2. This makes it possible to generate a cross-sectional shape of the object sliced on a virtual plane at height 50. By using similar processing, cross-sectional shapes sliced at multiple heights can be generated, and the three-dimensional shape of the object can be reconstructed by stacking the generated cross-sectional shapes.
[0032] [Preprocessing] The calculation unit 22 performs preprocessing on the first projection data and the second projection data before performing the restoration process. For example, if there is a location in the object where the pixel value is high due to the presence of abnormal accumulation, etc., the intersection point of the lines L1 and L2 may be incorrect. In other words, it may be determined that the lines L1 and L2 intersect at the location where the pixel value is high due to abnormal accumulation, resulting in a cross-sectional shape that is different from the actual shape.
[0033] The purpose of preprocessing is to suppress the influence of such abnormal values and enable appropriate reconstruction of the three-dimensional shape. However, preprocessing is optional and not essential. As will be described later, the image processing device of this embodiment uses projected data (integral data) as the calculation target, and therefore is able to perform calculations even if the pixel values inside the object are somewhat non-uniform.
[0034] The pre-processing performed by the calculation unit 22 will now be described. (1) Threshold method The calculation unit 22 masks pixels having pixel values higher than a predetermined threshold in the image of the object captured in the first projection data and the image of the object captured in the second projection data. The threshold may be set in advance depending on the type of image to be captured, or may be determined based on a histogram of pixel values of the captured image, etc.
[0035] (2) Smoothing method The calculation unit 22 smoothes the image of the object captured in the first projection data and the image of the object captured in the second projection data. Image smoothing is a process of averaging the pixel value of a certain point using the pixel values of its surrounding points, making it possible to make the change in pixel value continuous.
[0036] (3) Image restoration Image restoration here refers to a technique for restoring a missing image region using surrounding images. This technique is also called image inpainting. Various image restoration algorithms have been proposed. For example, a method for performing restoration by regularization or modeling, assuming smoothness and periodicity of neighboring pixels, or a method for pasting textures synthesized from non-missing regions, has been proposed. Recently, various techniques, such as a method using deep learning, have also been proposed. The calculation unit 22 erases the image in the specified region. The region to be erased may be specified based on user input, or a region having a pixel value equal to or greater than a predetermined threshold may be specified using the calculation unit 22. The calculation unit 22 restores the image in the erased region using the image restoration technique described above.
[0037] 4 is a flowchart showing the flow of image processing by the image processing device 20. The image processing device 20 receives first projection data and second projection data captured by an imaging device (S10). The image processing device 20 performs preprocessing on the input first projection data and second projection data (S11). The preprocessing may be performed using any of the following methods: threshold method, smoothing method, and image inpainting. Furthermore, the preprocessing is optional, and may not be performed depending on the input image.
[0038] The image processing device 20 identifies a diameter (i.e., straight line L1) and end points of the object in a virtual plane obtained by slicing the object reflected in the first projection data at a predetermined height, and also identifies a diameter (i.e., straight line L2) and end points of the object in a virtual plane obtained by slicing the object reflected in the second projection data at the same height (S12). The image processing device 20 identifies pixel m with the maximum pixel value among the pixels on straight line L1, and also identifies pixel n with the maximum pixel value among the pixels on straight line L2 (S13).
[0039] The image processing device 20 intersects the lines L1 and L2 at pixel m and pixel n as the intersection points, and generates a spline curve that passes through the end points a and b of the line L1 and the end points c and d of the line L2 (S14). The area enclosed by this spline curve becomes the cross-sectional shape of the object at a predetermined height. The image processing device 20 determines whether the process of restoring the cross-sectional shapes for all heights (S12 to S15) has been completed (S15), and if there are heights for which the process has not yet been completed (NO in S15), it repeatedly performs the process of determining the cross-sectional shape for each height (S12 to S15).
[0040] When the process of restoring the cross-sectional shapes for all heights is completed (YES in S15), the image processing device 20 stacks the spline curves (i.e., cross-sectional shapes) obtained at each height in the height direction to restore the three-dimensional shape of the object (S16).The image processing device 20 outputs an image of the restored three-dimensional shape (S17).
[0041] The configuration of the image processing device 20 of this embodiment has been described above, but an example of the hardware of the image processing device 20 is a computer equipped with a CPU, RAM, ROM, hard disk, display, keyboard, mouse, communication interface, etc. The image processing device 20 is realized by storing a program having modules that realize each of the above functions in RAM or ROM and executing the program by the CPU. Such programs are also included in the scope of the present invention.
[0042] (Effects of this embodiment) Previously, to reconstruct the three-dimensional shape of an organ or other object, it was necessary to obtain cross-sectional images, and in the case of SPECT, to obtain these images, radiation emitted from the patient after administration of a radiopharmaceutical was collected by a detector that rotates around the body, and then image processing was required, such as image reconstruction and setting of the region of interest. This processing not only required a long time for the patient to be confined to the data collection, but also increased the image processing time.
[0043] The image processing device 20 and image capturing device 1 of this embodiment can reconstruct the three-dimensional shape of an object based on first projection data and second projection data obtained by capturing images from orthogonal directions, thereby shortening the time the patient is confined and reducing the burden of image processing.
[0044] Furthermore, with the image processing device 20 of this embodiment, projection data from two directions is sufficient, and the resolution of the projection data in two directions can be increased compared to conventional methods that require projection data from all directions. In conventional methods, attempting to obtain high-resolution projection data from all directions would require the patient to spend even longer time in occupying the body, making it difficult to increase the resolution of each individual image. According to this embodiment, the resolution of the projection data can be increased, allowing a high-resolution three-dimensional shape to be reconstructed.
[0045] (Second embodiment) Next, an image processing device according to a second embodiment will be described. The basic configuration of the image processing device according to the second embodiment is the same as that of the image processing device 20 according to the first embodiment. The image processing device according to the second embodiment may be combined with an imaging device to form an image capturing device. The image processing device according to the second embodiment is designed so that the intersections of the straight lines L1 and L2 obtained at each height are smoothly continuous.
[0046] Fig. 5 is a diagram illustrating the processing performed by the calculation unit in the image processing device of the second embodiment. Fig. 5 shows examples of the first projection data and the second projection data, and the points seen inside the object indicate the pixels with the highest pixel values at each height. In the first projection data, they correspond to the "first points" at each height, and in the second projection data, they correspond to the "second points" at each height.
[0047] The calculation unit draws a curve B1 that smoothly connects multiple first points of the first projection data, and corrects the position of the first points so that they are on the curve B1. The calculation unit also draws a curve B2 that smoothly connects multiple second points of the second projection data, and corrects the position of the second points so that they are on the curve B2. The corrected first and second points may not be the points with the highest pixel values at a given height, but the corrected first and second points are used when calculating the cross-sectional shape of the object sliced for each height. The process in which the calculation unit uses the corrected first and second points to calculate a spline curve representing the cross-sectional shape for each height is the same as that of the image processing device of the first embodiment (see FIG. 3).
[0048] FIG. 6 is a flowchart showing the flow of image processing by the image processing device of the second embodiment. The image processing device inputs first projection data and second projection data captured by an imaging device (S20). The image processing device performs preprocessing on the input first projection data and second projection data (S21). The preprocessing may be performed using any of the following methods: threshold method, smoothing method, and image inpainting. Furthermore, preprocessing is optional, and may not be performed depending on the input image.
[0049] The image processing device identifies a diameter (i.e., straight line L1) and end points of the object at a predetermined height in a virtual plane obtained by slicing the object reflected in the first projection data at the predetermined height, and also identifies a diameter (i.e., straight line L2) and end points of the object at the predetermined height in a virtual plane obtained by slicing the object reflected in the second projection data at the same height (S22). Next, the image processing device identifies pixel m with the maximum pixel value among the pixels on straight line L1, and identifies pixel n with the maximum pixel value among the pixels on straight line L2 (S23).
[0050] The image processing device determines whether the process of identifying the radius, end point, and maximum pixel position for all heights (S22, S23) has been completed (S24). If there are heights for which the process has not yet been completed (NO in S24), the process of identifying the radius, end point, and maximum pixel position for each height (S22, S23) is repeated. If the process of identifying the radius, end point, and maximum pixel position for all heights has been completed (YES in S24), the image processing device corrects the maximum pixel position for each height in the first projection data and the second projection data (S25). The process of correcting the maximum pixel position is a process of correcting the position of the maximum pixel so that it is smoothly connected in the height direction. For convenience of explanation, the position of the maximum pixel after correction for the first projection data will be referred to as the "first point," and the position of the maximum pixel after correction for the second projection data will be referred to as the "second point."
[0051] Next, the image processing device intersects the straight lines L1 and L2 so that the corrected first and second points intersect, and generates a spline curve that passes through the end points a and b of the straight line L1 and the end points c and d of the straight line L2 (S26). The area enclosed by this spline curve becomes the cross-sectional shape of the object at a predetermined height. The image processing device stacks the spline curves (i.e., cross-sectional shapes) obtained at each height in the height direction to reconstruct the three-dimensional shape of the object (S27). The image processing device outputs an image of the reconstructed three-dimensional shape (S28).
[0052] The image processing device of the second embodiment has been described above. The image processing device of the second embodiment has the same effect as the image processing device of the first embodiment, that is, it can reconstruct the three-dimensional shape of an object from two frames of projection data.
[0053] Furthermore, the image processing device of the second embodiment corrects the positions of the maximum pixels at each height so that they are smoothly connected. Since the positions where the lines of the maximum diameters of the cross sections of an object at each height intersect are usually considered to be continuous, correcting the positions of the maximum pixels at each height so that they are smoothly connected allows the three-dimensional shape of the object to be properly restored.
[0054] Although the image processing device and image processing method of the present invention have been described in detail above by way of embodiments, the present invention is not limited to the above-described embodiments.
[0055] In the above-described embodiment, an example was given in which first projection data and second projection data were obtained by capturing images from two orthogonal directions, but the first projection data and second projection data do not necessarily have to be data obtained from two orthogonal directions. Even when the capturing directions of the first projection data and the second projection data form a predetermined angle other than a right angle, it is possible to reconstruct the three-dimensional shape of the object using a method similar to that described above. However, the three-dimensional shape can be most appropriately reconstructed when the capturing directions of the first projection data and the second projection data are orthogonal.
[0056] Although the image processing device and image processing method of this embodiment are applied to liver receptor scintigraphy in the above example, the following medical images are also considered to be applicable to the present invention.
[0057] Dopamine transporter scintigraphy As an index of the degree of striatal accumulation, analysis is performed using the ratio of striatal accumulation to nonspecific accumulation (SBR: Specific Binding Ratio). This requires information on striatal volume, but striatal volume has not previously been measured for individual patients. Using this method, it is possible to determine the patient-specific striatal volume, thereby enabling accurate SBR calculations.
[0058] Attenuation correction for head regions (cerebral blood flow and dopamine transporter scintigraphy) To perform attenuation correction, linear attenuation coefficient distribution information for the target region is required. Conventionally, linear attenuation coefficient distribution information has been created manually or using an external radiation source or X-ray CT. When created manually, there is variation between operators, and methods using an external radiation source or X-ray CT involve additional radiation exposure. Therefore, by creating a three-dimensional shape of the head using this method, linear attenuation coefficient distribution information for the target region can be obtained without variations between operators or additional radiation exposure.
[0059] Myocardial perfusion scintigraphy (removal of extracardiac accumulation) Drug accumulation near the myocardium can sometimes interfere with image interpretation. This method can identify extracardiac accumulation from the myocardium (in this case, the area outside the three-dimensional shape is extracardiac accumulation) or from the three-dimensional shape of extracardiac accumulation. This makes it possible to remove extracardiac accumulation from cross-sectional images and create images from which extracardiac accumulation has been removed by limiting the image reconstruction range.
[0060] Thyroid scan Thyroid weight may be obtained and the 131 In internal radiotherapy, the major axis and area of both the left and right lobes are measured from the scintigram, and the weight is calculated using the Allen-Goodwin method or the Okubo method. However, these are only approximate calculations. By using this method, the three-dimensional shape of the thyroid gland can be obtained and a general thyroid density can be applied, allowing for a more accurate calculation of thyroid weight.
[0061] Renal scintigraphy During quantitative analysis, kidney depth correction (attenuation correction) is required. Approximation formulas such as the Tonnesen, Ito (Kazu), and Ito (Tsuna) methods have been used to estimate this depth. This method allows for more accurate estimation of kidney depth by obtaining the three-dimensional shape of the kidney and body contour.
[0062] Lung perfusion scan Before resection, the residual function after surgery can be estimated from the three-dimensional shape of the blood flow distribution.
[0063] In any of the above applications, the calculation method described in the embodiment can be used. However, when extracting a striatum, the camera arrangement may be narrow-angled rather than perpendicular. This allows the two striatum to be detected separately (i.e., separately). [Example]
[0064] An example in which image processing was performed using the image processing method of this embodiment will be described below. 99m After administering Tc-GSA, the three-dimensional shape and volume of the liver are estimated from planar images taken from two orthogonal directions.
[0065] (Experimental Method) The volume of the three-dimensional shape of the liver reconstructed from two frames of projection data using the image processing method of this embodiment is compared with the volume of the three-dimensional shape of the liver reconstructed from a SPECT image.
[0066] (subject) Data collection period: November 2014 to October 2019 Number of subjects: 9 (8 males, 1 female) Age: 61.63±15.73 years Weight: 69.73 ± 15.84 kg
[0067] (imaging equipment) Equipment: Bright View X with XCT (Philips) Radioisotopes: 99m Tc 185MBq Collimator: Low Energy High Resolution (LEHR)
[0068] (Plane image: this example) Zoom: 1.46 x (40.9) cm Matrix: 256 x 256 Time: 120 seconds
[0069] (SPECT image: comparison example) Zoom: 1.46 x (40.9) cm Matrix: 128x128 Number of steps: 64 Rotation: 360° (180° x 2) Time: 30 seconds / angle Image reconstruction: the Astonish method AC (attenuation correction): CTAC method SC (Scatter Correction): Effective source scatter estimation (ESSE) method CT: 120kV, 30mA
[0070] (Experimental results) The experimental results are shown in Figure 7. The horizontal axis represents the volume of the three-dimensional shape of the liver reconstructed using the method of this embodiment, and the vertical axis represents the volume of the three-dimensional shape of the liver reconstructed from the SPECT image. As shown in Figure 7, a linear correlation was observed between the volume of the three-dimensional shape reconstructed from two frames of projection data and the volume of the three-dimensional shape reconstructed from the SPECT image.
[0071] Figure 8 shows an example of the three-dimensional shape of a liver reconstructed from two pieces of projection data. Figures 8(a) to 8(f) are images of the same liver viewed from different directions. It was visually confirmed that the appearance of the reconstructed object was close to the shape of the liver reconstructed from the SPECT image.
[0072] From the above, it was confirmed that the image processing method of this embodiment can restore a three-dimensional shape from projection data. [Explanation of symbols]
[0073] 1. Imaging device 10 Imaging equipment 20 Image processing device 21 Input section 22 Arithmetic section 23 Output section 24 Memory section
Claims
1. an input unit for inputting first projection data and second projection data of an object photographed from different directions; a calculation unit that reconstructs a three-dimensional shape of the object based on the first projection data and the second projection data, The calculation unit Identifying a first diameter and an end point of the object at a predetermined height imaged in the first projection data, and a second diameter and an end point of the object at the predetermined height imaged in the second projection data; identifying a first point within the first radius having the highest pixel value and a second point within the second radius having the highest pixel value; intersecting the first diameter and the second diameter so that the second diameter passes through the first point and the second point on an imaginary plane obtained by slicing the object at the predetermined height, and generating a curve passing through an end point of the first diameter and an end point of the second diameter; An image processing device that generates a curve for each height by changing the predetermined height, and overlays the generated curves in the height direction to restore the three-dimensional shape of the object.
2. 2. The image processing device according to claim 1, wherein the calculation unit masks pixels in the image of the object captured in the first projection data and the image of the object captured in the second projection data that have pixel values higher than a predetermined threshold as a preprocessing step for restoring the three-dimensional shape of the object.
3. The image processing device according to claim 1 , wherein the calculation unit smoothes the image of the object captured in the first projection data and the image of the object captured in the second projection data as a preprocessing step for restoring the three-dimensional shape of the object.
4. 2. The image processing device according to claim 1, wherein, as a preprocessing step for restoring the three-dimensional shape of the object, the calculation unit erases an image of a specified area from the image of the object reflected in the first projection data and the image of the object reflected in the second projection data, and restores the image of the erased area using images of its surroundings.
5. The image processing device according to any one of claims 1 to 4, wherein the calculation unit changes the specified height and identifies the first point and the second point for each height, corrects the position of the first point so that the first points are smoothly connected in the height direction, corrects the position of the second point so that the second points are smoothly connected in the height direction, and intersects the first diameter and the second diameter using the corrected first point and the corrected second point.
6. 5. The image processing apparatus according to claim 1, wherein the first projection data and the second projection data are projection data of the object acquired by photographing from two orthogonal directions.
7. an imaging unit that images an object from different directions to acquire first projection data and second projection data; a calculation unit that reconstructs a three-dimensional shape of the object based on the first projection data and the second projection data, The calculation unit Identifying a first diameter and an end point of the object at a predetermined height imaged in the first projection data, and a second diameter and an end point of the object at the predetermined height imaged in the second projection data; identifying a first point within the first radius having the highest pixel value and a second point within the second radius having the highest pixel value; intersecting the first diameter and the second diameter so that the second diameter passes through the first point and the second point on an imaginary plane obtained by slicing the object at the predetermined height, and generating a curve passing through an end point of the first diameter and an end point of the second diameter; An image capturing device that generates a curve for each height by changing the predetermined height, and reconstructs the three-dimensional shape of the object by overlapping the generated curves in the height direction.
8. A program for reconstructing a three-dimensional shape of an object from first projection data and second projection data of the object photographed from different directions, the program comprising: inputting the first projection data and the second projection data; specifying a first diameter and an end point of the object at a predetermined height imaged in the first projection data, and a second diameter and an end point of the object at the predetermined height imaged in the second projection data; identifying a first point within the first radius having the highest pixel value and a second point within the second radius having the highest pixel value; intersecting the first diameter and the second diameter in a virtual plane obtained by slicing the object at the predetermined height so that the second diameter passes through the first point and the second point, and generating a curve passing through an end point of the first diameter and an end point of the second diameter; a step of generating a curve for each height by changing the predetermined height, and overlaying the generated curves in the height direction to restore the three-dimensional shape of the object; A program that executes the following.
9. An image processing method for reconstructing a three-dimensional shape of an object from first projection data and second projection data of the object photographed from different directions by an image processing device, comprising: the image processing device inputting the first projection data and the second projection data; a step in which the image processing device specifies a first diameter and an end point of the object at a predetermined height imaged in the first projection data, and a second diameter and an end point of the object at the predetermined height imaged in the second projection data; the image processing device identifying a first point within the first diameter having the highest pixel value and a second point within the second diameter having the highest pixel value; the image processing device intersects the first diameter and the second diameter on a virtual plane obtained by slicing the object at the predetermined height so that the second diameter passes through the first point and the second point, and generates a curve passing through an end point of the first diameter and an end point of the second diameter; the image processing device generates a curve for each height by changing the predetermined height, and restores the three-dimensional shape of the object by superimposing the generated curves in the height direction; An image processing method comprising: