Medical image processing device, treatment system, medical image processing method, program, and storage medium

Abstract

A medical image processing device according to an embodiment comprises: a first image acquisition unit that acquires a first image captured in a first stage; a second image acquisition unit that acquires a second image captured in a second stage; a 3D-3D positioning execution unit that aligns the position of the second image with the position of the first image; a region information acquisition unit that acquires region information corresponding to a region of the first image; a region estimation unit that estimates a region corresponding to the region information in the second image; a tracking model acquisition unit that acquires a tracking model; a tracking unit that tracks an object in the second image using the tracking model; and an error calculation unit that calculates an error between a tracking position, which is the position of the object in the second image tracked by the tracking unit, and the position of the object in the second image.

Description

Medical imaging processing device, treatment system, medical imaging processing method, program, and recording medium

[0001] Embodiments of the present invention relate to a medical imaging processing device, a treatment system, a medical imaging processing method, a program, and a recording medium. This application claims priority based on Japanese Patent Application No. 2024-193973 filed in Japan on November 5, 2024, and incorporates its content herein by reference.

[0002] Radiation therapy is a treatment method that destroys a lesion in a patient's body by irradiating the lesion with radiation. At this time, the radiation needs to be accurately irradiated to the position of the lesion. This is because if the normal tissue in the patient's body is irradiated with radiation, it may affect that normal tissue. Therefore, when performing radiation therapy, the position of the lesion in the patient's body is three-dimensionally grasped at the treatment planning stage. To perform this grasping, a three-dimensional fluoroscopic image of the patient is taken. A CT image obtained by computed tomography (CT) is an example of a three-dimensional fluoroscopic image. Based on the grasped position of the lesion, the direction of radiation irradiation and the intensity of the irradiated radiation are planned so as to reduce the irradiation to normal tissue. Then, in the treatment stage where the treatment is actually performed, the patient's position is adjusted to the position of the patient at the treatment planning stage, and the radiation is irradiated to the lesion according to the planned irradiation direction and irradiation intensity.

[0003] In the patient positioning in the treatment stage, a three-dimensional fluoroscopic image is virtually placed in the treatment room, and the position of the mobile bed on which the patient is actually lying in the treatment room is adjusted so that it coincides with the position of the three-dimensional fluoroscopic image.

[0004] Japanese Patent Publication No. 2018-507073

[0005] In radiation therapy, it is important to address interfractional changes. Interfractional change is a phenomenon in which the position of an object such as a tumor in a patient's body changes between the treatment planning stage and the treatment stage. That is, it is required to align the irradiation position of the radiation specified in the treatment planning stage with the position of the object in the treatment stage.

[0006] This invention has been made in consideration of these circumstances and aims to provide a medical image processing device, a treatment system, a medical image processing method, a program, and a recording medium that can deal with interferenceal change.

[0007] The medical image processing apparatus of the embodiment includes: a first image acquisition unit that acquires a first image, which is a three-dimensional fluoroscopic image of a patient taken in a first stage; a second image acquisition unit that acquires a second image, which is a three-dimensional fluoroscopic image of the patient taken in a second stage later than the first stage; a 3D-3D positioning execution unit that aligns the position of the second image with the position of the first image based on the pixel values ​​of the first and second images; a region information acquisition unit that acquires region information corresponding to the region of the first image; a region estimation unit that estimates the region in the second image corresponding to the region information; a tracking model acquisition unit that acquires a tracking model created based on the first image; a tracking unit that tracks an object in the second image using the tracking model; and an error calculation unit that calculates the error between the tracking position, which is the position of the object in the second image tracked by the tracking unit, and the position of the object in the second image.

[0008] According to embodiments of the present invention, it is possible to provide a medical image processing apparatus, a treatment system, a medical image processing method, a program, and a recording medium capable of dealing with interferenceal change.

[0009] A block diagram showing the schematic configuration of a treatment system equipped with the medical image processing device of the embodiment. A block diagram showing the schematic configuration of the treatment system equipped with the medical image processing device of the embodiment, viewed from a different angle than Figure 1. A block diagram mainly showing the schematic configuration of the medical image processing device of the embodiment. A diagram estimating the region in the second image based on the first image. A diagram explaining the difference in tumor position according to respiratory waveform in 4D-CT and DRR. A flowchart showing an example of the processing flow performed by the medical image processing device of the embodiment.

[0010] The medical image processing apparatus, treatment system, medical image processing method, and program of the embodiment will be described below with reference to the drawings.

[0011] [Overall Configuration] Figure 1 is a block diagram showing the schematic configuration of a treatment system 1 equipped with a medical image processing device 100 of an embodiment. The treatment system 1 includes, for example, a treatment device 10, a medical image processing device 100, and a display device 200. The treatment device 10 includes, for example, a patient bed 12, a patient bed control unit 14, a treatment beam irradiation gate 18 (irradiation unit), and a first imaging device D1. In Figure 1, a computed tomography (CT) device 16 (hereinafter referred to as "CT imaging device 16") is shown as an example of the first imaging device D1.

[0012] The treatment table 12 is a movable treatment table that fixes a patient P (subject) receiving radiation therapy in a lying position, for example, using a restraint device. The treatment table 12 moves with the patient P fixed inside the annular CT scanner 16 having an opening, according to the control from the treatment table control unit 14. The treatment table control unit 14 controls the translational mechanism and rotational mechanism provided on the treatment table 12 in order to position the patient P fixed to the treatment table 12 according to the movement amount signal output by the medical image processing device 100. The translational mechanism can drive the treatment table 12 in three axial directions (X axis, Y axis, Z axis). The rotational mechanism can rotate the treatment table 12 around three axes. In other words, the treatment table control unit 14 moves the treatment table 12 with six degrees of freedom by controlling, for example, the translational mechanism and rotational mechanism of the treatment table 12. The bed control unit 14 has six degrees of freedom to control the bed 12, but may have fewer than six degrees of freedom (for example, four degrees of freedom) or more than six degrees of freedom (for example, eight degrees of freedom). The bed 12 is installed so as to be movable to both positions when the position where imaging by the CT scanner 16 is performed and the position where the treatment beam B is irradiated by the treatment beam irradiation gate 18 are different.

[0013] The CT scanner 16 performs three-dimensional computed tomography. The CT scanner 16 has multiple radiation sources arranged inside the annular (gantry) opening. Each of the multiple radiation sources emits radiation to visualize the inside of the patient P's body. In other words, the CT scanner 16 emits radiation from multiple locations around the patient P. In the CT scanner 16, the radiation emitted from each of the multiple radiation sources is, for example, X-rays. The CT scanner 16 has multiple radiation detectors arranged inside the annular opening. Each of the multiple radiation detectors detects radiation emitted from the corresponding radiation source and that has passed through the patient P's body. The CT scanner 16 generates a CT image of the inside of the patient P's body based on the magnitude of the radiation energy detected by each of the multiple radiation detectors. The CT image of patient P generated by the CT scanner 16 is a three-dimensional digital image. The three-dimensional digital image represents the magnitude of the degree of radiation attenuation at each of the multiple locations inside the body as a digital value. The CT scanner 16 outputs the generated CT image to the medical image processing device 100. The imaging of the inside of the patient P's body in the CT scanner 16, that is, the generation of a CT image based on the irradiation of radiation from each of the multiple radiation sources and the radiation detected by each of the multiple radiation detectors, is controlled, for example, by an imaging control unit (not shown).

[0014] The treatment beam irradiation gate 18 irradiates the patient P with radiation as treatment beam B to destroy the tumor (lesion), which is the target area for treatment, located within the patient P's body. Treatment beam B can be, for example, X-rays, gamma rays, electron beams, proton beams, neutron beams, or heavy ion beams. Treatment beam B is irradiated linearly from the treatment beam irradiation gate 18 to the patient P (more specifically, the tumor inside the patient P's body). The irradiation of treatment beam B at the treatment beam irradiation gate 18 is controlled, for example, by a treatment beam irradiation control unit (not shown). In the illustrated treatment system 1, the treatment beam irradiation gate 18 is an example of an "irradiation unit".

[0015] In radiation therapy, the irradiation direction and intensity of the treatment beam B are planned by simulating the patient P being placed on the treatment table 12 in the treatment room. This stage of planning the treatment is called the "treatment planning stage." The treatment planning stage is an example of the "first stage." After the first stage, the second stage is performed. Here, the "second stage" is a concept that includes not only the period during which the treatment beam B is actually irradiated to the patient P, but also the period immediately before and after that.

[0016] In the treatment planning stage, specifically, a physician identifies the areas to be irradiated from a CT image (an example of the "first image"), or such processing is performed automatically. For this reason, the CT image in the treatment planning stage is associated with information related to the treatment plan (hereinafter referred to as "treatment planning information"), such as parameters representing the angle of the treatment table 12 in the treatment room and the patient's position (lying on their back or stomach). This is also true for CT images taken immediately before radiation therapy and CT images taken during previous radiation therapy. In other words, the CT image taken inside the patient P's body by the CT scanner 16 is associated with treatment planning information, such as parameters representing the angle of the treatment table 12 and the patient's position at the time of imaging. The treatment planning information may also include information regarding the direction and intensity of irradiation when the treatment beam B is irradiated onto the patient P. The treatment planning information may also include information such as the location of the tumor and the location of risk organs that should be avoided from radiation exposure. The treatment planning information may include parameters assigned to each pixel of the CT image for coloring the location of the tumor, the location of risk organs, etc. Treatment plan information may be assigned in association with a predetermined region of the first image. Among the treatment plan information, the information assigned in association with a predetermined region of the first image is referred to as "region information." For example, if the region information is the location of a tumor or a risk organ, the region information is assigned in association with pixels, etc., of the region in the first image that corresponds to the tumor or risk organ.

[0017] Furthermore, the region information assigned to the first image may include a so-called gate window. The gate window is used to control the irradiation position and timing of the treatment beam B. More specifically, by turning on the output of the treatment beam irradiation gate 18 only during the period when a specific target is within the range of the gate window, it becomes possible to accurately irradiate tumors, etc., with the treatment beam B. By using a gate window, for example, if the position of a tumor, etc., moves with the patient P's breathing, it becomes possible to irradiate only the period when the tumor is in a suitable position with the treatment beam B.

[0018] Figure 1 shows a CT scanner 16 as an example of the first imaging device D1. However, the first imaging device D1 can be any device that generates three-dimensional images of the inside of the patient P's body, such as a cone-beam (CB) CT scanner, a magnetic resonance imaging (MRI) scanner, or an ultrasound diagnostic device.

[0019] In the second stage, the medical image processing device 100 outputs a movement amount signal to the bed control unit 14 to move the bed 12 in order to position the patient P in the same position as in the treatment planning stage. In other words, the medical image processing device 100 outputs a movement amount signal to the bed control unit 14 to move the patient P so that the treatment beam B can be appropriately irradiated to the tumor or tissue to be treated in radiotherapy. The movement amount signal is determined based on the first displacement amount and the second displacement amount, etc., which will be described later.

[0020] In the second stage, the display device 200 displays images to present various information in the treatment system 1 to the radiation therapist (such as a doctor) using the treatment system 1, including information obtained during the patient P positioning process in the medical image processing device 100. The display device 200 displays various images, such as CT images and X-ray fluoroscopy images output by the medical image processing device 100, or images on which various information is superimposed. Here, various information includes, for example, patient information (age, sex, height, weight, etc.), image acquisition conditions (acquisition site, presence or absence of contrast agent, tube voltage, tube current, etc.), date and time of acquisition, or patient position (head supine, feet prone, etc.). The display device 200 includes, for example, a device such as a liquid crystal display (LCD). The radiation therapist can obtain information for performing radiation therapy using the treatment system 1 by visually confirming the images displayed on the display device 200. The treatment system 1 may be configured to include a user interface, such as an operating unit (not shown), which is operated by the person performing the radiation therapy, and to allow manual operation of various functions performed by the treatment system 1.

[0021] Figure 2 is a schematic diagram of the treatment system 1, viewed from a different angle than Figure 1. Figure 2 is a block diagram showing the schematic configuration of the treatment system. As shown in Figure 2, the treatment system 1 includes a second imaging device D2. The second imaging device D2 has, for example, two radiation sources 20 (radiation source 20-1 and radiation source 20-2) and two radiation detectors 30 (radiation detector 30-1 and radiation detector 30-2). However, the second imaging device D2 may be, for example, an ultrasound diagnostic device.

[0022] Radiation source 20-1 irradiates patient P with radiation r-1 at a predetermined angle for fluoroscopy of the patient's body. Radiation source 20-2 irradiates patient P with radiation r-2 at a predetermined angle different from that of radiation source 20-1 for fluoroscopy of the patient's body. Radiation r-1 and radiation r-2 are, for example, X-rays. Figure 2 shows a case where X-ray imaging is performed from two directions on patient P, who is fixed on a bed 12. Note that in Figure 2, the control unit that controls the irradiation of radiation r-1 and r-2 by radiation source 20 is omitted. In the following description, radiation r-1 and r-2 may be simply referred to as radiation r.

[0023] Radiation detector 30-1 detects radiation r-1 that has been irradiated from radiation source 20-1 and passed through the body of patient P, and generates an X-ray fluoroscopic image of the inside of patient P corresponding to the energy magnitude of the detected radiation r-1. Radiation detector 30-2 detects radiation r-2 that has been irradiated from radiation source 20-2 and passed through the body of patient P, and generates an X-ray fluoroscopic image of the inside of patient P corresponding to the energy magnitude of the detected radiation r-2. In radiation detector 30, multiple X-ray detectors are arranged in a two-dimensional array. Radiation detector 30 generates a digital image as an X-ray fluoroscopic image, which represents the energy magnitude of the radiation r that has reached each of the multiple X-ray detectors as a digital value. This X-ray fluoroscopic image is an example of a "third image". The third image is a two-dimensional fluoroscopic image of patient P taken immediately before radiation therapy (i.e., in the second stage). The third image may also be an echo image generated by an ultrasound diagnostic device.

[0024] The radiation detector 30 is, for example, a flat panel detector (FPD), an image intensifier, or a color image intensifier. The following describes the case where each of the multiple radiation detectors 30 is an FPD. Each radiation detector 30 (FPD) outputs the generated X-ray fluoroscopic image to the medical image processing device 100. Note that in Figure 2, the control unit that controls the generation of X-ray fluoroscopic images by the radiation detectors 30 is omitted from the diagram.

[0025] The specific configuration of the second imaging device D2 may be changed. For example, the second imaging device D2 may be equipped with three or more sets of radiation sources 20 and radiation detectors 30. Alternatively, the second imaging device D2 may be equipped with only one set of radiation sources 20 and radiation detectors 30. Hereinafter, the combination of radiation sources 20 and radiation detectors 30 may be referred to as an "X-ray imaging device".

[0026] The various components shown in Figures 1 and 2 may be connected to each other by wires, or they may be connected wirelessly, for example, by a LAN (Local Area Network) or a WAN (Wide Area Network).

[0027] [Medical Image Processing Device] The medical image processing device 100 of the embodiment will be described below. Figure 3 is a block diagram mainly showing the schematic configuration of the medical image processing device 100 of the embodiment. The medical image processing device 100 includes, for example, a first image acquisition unit 110, a second image acquisition unit 120, a 3D-3D positioning execution unit 130, a region information acquisition unit 140, a region estimation unit 150, a tracking model acquisition unit 160, a tracking unit 170, an error calculation unit 180, and a model correction unit 190.

[0028] Some or all of the components of the medical image processing device 100 are realized, for example, by a hardware processor such as a CPU (Central Processing Unit) executing a program (software). Some or all of these components may be realized by hardware (including circuitry) such as an LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or GPU (Graphics Processing Unit), or by the cooperation of software and hardware. Some or all of the functions of these components may be realized by a dedicated LSI. The program may be stored in advance in a storage device (a storage device equipped with a non-transient recording medium) such as ROM (Read Only Memory), RAM (Random Access Memory), HDD (Hard Disk Drive), or flash memory provided by the medical image processing device 100. The program may also be stored in a removable recording medium (a non-transient recording medium) such as a DVD or CD-ROM, and installed in the HDD or flash memory of the medical image processing device 100 when the recording medium is inserted into the drive device provided by the medical image processing device 100. The program may also be downloaded from another computer device via a network and installed in the HDD or flash memory of the medical image processing device 100.

[0029] The first image acquisition unit 110 acquires a first image of patient P during the treatment planning stage (first stage). The first image is a three-dimensional CT image representing the three-dimensional shape of the patient P's body, for example, taken by a CT scanner 16 during the treatment planning stage. The first image is used to determine the direction (path including inclination and distance) and intensity of the treatment beam B irradiated onto patient P during radiation therapy. The first image may also be an MRI image taken by an MRI scanner during the treatment planning stage. The first image may be accompanied by the aforementioned region information. The first image acquisition unit 110 may acquire multiple still images taken at different timings within the respiratory cycle of the same patient P as the first image. The first image may also be a video, in which time axis information is added to a three-dimensional fluoroscopic image. For example, the first image may be a 3D-CT image.

[0030] The second image acquisition unit 120 acquires a second image of patient P immediately before the start of radiation therapy (i.e., the second stage). The second image is a three-dimensional CT image representing the three-dimensional shape of the inside of patient P's body, taken, for example, by a CT scanner 16, in order to adjust the position of patient P when irradiating with the treatment beam B during radiation therapy (i.e., positioning). In other words, the second image is an image taken by the CT scanner 16 immediately before irradiating with the treatment beam B from the treatment beam irradiation gate 18. In this case, the time at which the first image is acquired and the time at which the second image is acquired are different, but the method of acquiring the first image is the same as the method of acquiring the second image. The second image may also be an MRI image taken by an MRI scanner in the second stage. The second image acquisition unit 120 may acquire multiple still images taken at different timings within the respiratory cycle of the same patient P as the second image. The second image may also be a video, in which time axis information is added to a three-dimensional fluoroscopic image. For example, the second image may be a 3D-CT image. The second image may have the region information assigned to the first image copied and applied to it.

[0031] The medical image processing device 100 may generate a two-dimensional DRR image based on a three-dimensional second image. A DRR (Digitally Reconstructed Radiograph) image is a digitally reconstructed X-ray photograph obtained by virtually reconstructing an X-ray fluoroscopic image from a three-dimensional image (for example, a CT image).

[0032] The 3D-3D positioning execution unit 130 performs 3D-3D positioning processing to align the position of patient P when performing radiation therapy, based on the first image acquired by the first image acquisition unit 110 and the second image acquired by the second image acquisition unit 120. More specifically, for example, the medical image processing device 100 calculates the amount of three-dimensional displacement (hereinafter sometimes referred to as the "first displacement amount") between the first image acquired by the first image acquisition unit 110 and the second image acquired by the second image acquisition unit 120. The medical image processing device 100 aligns the position between the first image and the second image by correcting the second image by the calculated first displacement amount. The 3D-3D positioning execution unit 130 may also compare the first image with one or more second images, select a pair of first and second images with a high similarity based on pixel values, and perform positioning with the selected pair. Here, "two or more second images" refers, for example, to multiple CT images taken at different times within the respiratory cycle. The location of the tumor may change depending on the respiratory cycle, and by using a pair of first and second images with a high degree of similarity, it is possible to reduce the deviation due to the respiratory cycle during 3D-3D positioning.

[0033] In the second stage, the medical image processing device 100 may output a movement amount signal to the bed control unit 14 to move the bed 12 on which the patient is placed and fixed by a first displacement amount, and the bed control unit 14 may move the bed 12 by the first displacement amount. Also, if the CT scanning device 16 and the treatment beam irradiation gate 18 are installed at separate locations, the medical image processing device 100 may output a movement amount signal to the bed control unit 14 to move the bed 12 by the distance between the CT scanning position and the irradiation position plus the first displacement amount. The bed control unit 14 may move the bed 12 by the distance between the CT scanning position and the irradiation position plus the first displacement amount.

[0034] The region information acquisition unit 140 acquires region information corresponding to a predetermined region of the first image. In the example shown in Figure 4(a), "region A" is set in the chest area of ​​the first image. The aforementioned region information is associated with this region A. A target T is located within the range of region A. Target T is, for example, a tumor. The region information may be stored together with the first image on a recording medium provided by the medical image processing device 100. In this case, the region information acquisition unit 140 reads the region information from the recording medium.

[0035] The region estimation unit 150 estimates the region in the second image that corresponds to the region information, based on the first image and the region information. In the example shown in Figure 4(a), the region information assigned to the first image corresponds to region A of the patient P's chest. Therefore, as shown in Figure 4(b), the region estimation unit 150 estimates the region A' of the chest that should be associated with the region information in the second image. The estimation method is not limited. For example, the position of region A' may be estimated based on the positional relationship (vector L) between the diaphragm boundary Q and region A in the first image and the diaphragm boundary Q in the second image. In the example shown in Figures 4(a) and 4(b), the position of the diaphragm boundary Q at the time of exhalation, i.e., when the diaphragm is at its lowest point in both images, is shifted between the first and second images. For this reason, it is inferred that the position of the target T is also shifted between the first and second images. This is an example of interferenceal change. Therefore, region A' in the second image is estimated to be in a shifted position relative to region A in the first image. Region information (for example, the direction and intensity of irradiation when irradiating patient P with treatment beam B) may be applied to the estimated region A' in the second image, and treatment may be performed. Alternatively, a three-dimensional motion vector between the first and second images may be obtained using DIR (Deformable Image Registration), and the tumor position in the second image may be determined from the amount of displacement corresponding to the tumor in the first image.

[0036] The tracking model acquisition unit 160 acquires a tracking model created based on the first image. Here, "tracking model" is a model for tracking the position of an object in multiple still or moving images. "Object" is, for example, a part of the human body referenced in radiation therapy, such as a tumor or a marker. The object may be the same as or different from the target T. The tracking model may be a model obtained by machine learning using the first image, which describes the positional relationship between the object and a landmark (e.g., the boundary of the diaphragm, a specific bone, etc.) that serves as a reference for estimating the object's position. Alternatively, the tracking model may be a model obtained by machine learning using the first image, which estimates the object's position from the image pattern (e.g., the distribution of pixel values ​​for each pixel) in the still or moving image.

[0037] The tracking model may be a pre-trained model that has been trained using AI (Artificial Intelligence) functions. Alternatively, the tracking model may be a model that tracks the position of an object using deep learning, a type of machine learning. Examples of classifiers used in the tracking model include Random Forest, Decision Tree, Support Vector Machine (SVM), K-Nearest Neighbor Algorithm (KNN), and Logistic Regression.

[0038] The tracking model may be stored, for example, along with the first image and region information, on a recording medium provided by the medical image processing device 100. For example, the tracking model is created during the treatment planning stage. In this case, the tracking model acquisition unit 160 acquires the tracking model by reading it from the recording medium.

[0039] The tracking unit 170 tracks the position of an object in the second image using the tracking model acquired by the tracking model acquisition unit 160. In this specification, the position of an object in the second image obtained by the tracking unit 170 is referred to as the "tracked position". For example, if the second image consists of multiple still images, the position of the object in each still image is obtained as the "tracked position". Alternatively, if the second image is a video, the position of the object in the video is obtained as the "tracked position" associated with the time axis.

[0040] A specific example of tracking by the tracking unit 170 will be explained using Figure 4. Figure 4(a) is an example of a first image. The symbol P indicates the boundary of the diaphragm as an example of a "landmark for estimating the position of an object". The symbol T indicates the position of the tumor as an example of an "object". That is, in the example shown in Figure 4, the target and the object are the same tumor. In Figure 4(a), the positional relationship between the boundary Q of the diaphragm and the target T is shown, for example, by the distance and direction of vector L. Figure 4(b) is an example of a second image. Comparing Figures 4(a) and (b), the position of the boundary Q of the diaphragm is different. For example, if the diaphragm and the target T move synchronously, in Figure 4(b), the position of the target T in the second image can be tracked using the position of the boundary Q of the diaphragm and vector L. Note that the method for estimating region A' and the method for tracking the position of the target T in the second image may be the same or different. For example, region A' may be estimated using vector L, and the position of target T may be tracked using the image pattern in the second image (e.g., the distribution of pixel values ​​for each pixel).

[0041] Using Figure 5, an example of the tracking unit 170 tracking the position of different objects according to the respiratory cycle will be explained. As shown in Figure 5(a), a patient's respiration can be represented by a respiratory waveform with time on the horizontal axis and amplitude on the vertical axis. For example, if a tumor (an example of an "object") is located around the chest, the position of the tumor changes according to the phase in the respiratory waveform. That is, the position of the tumor in the 4D-CT (an example of a "second image") shown in Figure 5(b) changes with time. By using a tracking model, the tracking unit 170 can track the position of the tumor for each phase in the respiratory waveform. Furthermore, if a DRR is generated using 4D-CT, as shown in Figure 5(c), the trajectory of the tumor's position in the DRR, which consists of multiple still images, can be obtained. This trajectory may be used to correct the tracking model by the model correction unit 190.

[0042] The error calculation unit 180 calculates the error between the tracking position determined by the tracking unit 170 and the position of the object in the second image detected by a method other than the tracking model. Here, "a method other than the tracking model" may be, for example, the radiation therapist visually inspecting the second image. Alternatively, "a method other than the tracking model" may be image analysis of the DRR generated based on the second image. Alternatively, a three-dimensional motion vector between the first and second images may be obtained using DIR (Deformable Image Registration), and the tumor position in the second image may be determined from the amount of displacement corresponding to the tumor in the first image.

[0043] The model correction unit 190 corrects the tracking model based on the error calculated by the error calculation unit 180. Correction means, for example, correcting the parameters of the model so that L' of the second image is obtained instead of L of the first image shown in FIG. 4. Or, the parameters of the model are corrected so as to estimate the position of the object from the image pattern of the tumor in the second image (for example, the distribution status of pixel values for each pixel). Or, in the case of correcting an AI model or a machine learning model, the second image is included in the learning data for re-learning or fine-tuning. The corrected tracking model may be stored in a recording medium provided in the medical image processing apparatus 100 and used for tracking by the tracking unit 170 after the next time. By correcting the tracking model in this way, it becomes possible to track more accurately the position of the object in the second image, that is, in a plurality of still images or moving images. By these processes, it is possible to cope with a so-called interfractional change in which the position of the object changes between the first image and the second image. Also, in the second image, for example, it becomes possible to accurately grasp the position of the object that varies depending on the respiratory cycle and irradiate the treatment beam B more accurately to the target.

[0044] Next, referring to FIG. 6, the flow of processing executed by the medical image processing apparatus 100 will be described. FIG. 6 is a flowchart showing an example of the flow of processing executed by the medical image processing apparatus 100.

[0045] First, the first image acquisition unit 110 acquires a three-dimensional fluoroscopic image (e.g., a CT image) that has been taken in advance during the treatment planning stage as the first image (step S100). Next, the medical image processing device 100 moves the bed 12 on which the patient P is fixed to a position where it can be photographed by the first imaging device D1 (e.g., a CT scanner 16) (step S102). Specifically, the medical image processing device 100 outputs a movement amount signal to the bed control unit 14. The bed control unit 14 controls the translation mechanism and the like provided on the bed 12. Next, the first imaging device D1 takes a three-dimensional fluoroscopic image (e.g., a 4D-CT image) of the patient P (step S104). Next, the second image acquisition unit 120 acquires the three-dimensional fluoroscopic image (e.g., a 4D-CT image or multiple CT images) taken in step S104 as the second image (step S106).

[0046] Next, the 3D-3D positioning execution unit 130 performs 3D-3D positioning processing based on the first image acquired by the first image acquisition unit 110 and the second image acquired by the second image acquisition unit 120, and calculates the first displacement amount (step S108). Next, the region information acquisition unit 140 acquires region information associated with the first image (step S110). Next, the region estimation unit 150 estimates the position of the region on the second image that corresponds to the region associated with the region information in the first image (step S112). Next, the tracking unit 170 tracks the position of the object (e.g., tumor T) in the second image using the tracking model acquired by the tracking model acquisition unit 160 (step S114). Next, the error calculation unit 180 calculates the error between the tracking position, which is the position of the object in the second image tracked by the tracking unit 170, and the position of the object in the second image (step S116). Next, the model correction unit 190 corrects the tracking model based on the error calculated by the error calculation unit 180 (step S118).

[0047] As described above, the medical image processing apparatus 100 of the embodiment includes a first image acquisition unit 110 that acquires a first image, which is a three-dimensional fluoroscopic image of a patient taken in the first stage; a second image acquisition unit 120 that acquires a second image, which is a three-dimensional fluoroscopic image of the patient taken in a second stage after the first stage; a 3D-3D positioning execution unit 130 that aligns the position of the second image with the position of the first image based on the pixel values of the first image and the second image; a region information acquisition unit 140 that acquires region information corresponding to the region of the first image; a region estimation unit 150 that estimates a region corresponding to the region information in the second image; a tracking model acquisition unit 160 that acquires a tracking model created based on the first image; a tracking unit 170 that tracks an object in the second image using the tracking model; and an error calculation unit 180 that calculates an error between a tracking position, which is the position of the object in the second image tracked by the tracking unit, and the position of the object in the second image. With this configuration, the medical image processing apparatus 100 can accurately track the position of the object in the second image and can cope with inter-fractional changes.

[0048] The medical image processing apparatus 100 may include a model correction unit 190 that corrects the tracking model based on the second image and the error calculated by the error calculation unit 180. With this configuration, the tracking accuracy by the tracking model can be further improved.

[0049] The region A' estimated by the region estimation unit 150 may be the range of the tumor in the second image.

[0050] The region A' estimated by the region estimation unit 150 may be the range of the marker in the second image.

[0051] The tracking model used by the tracking unit 170 may be a model obtained by machine learning using the image pattern of the tumor in the first image.

[0052] The tracking model used by the tracking unit 170 may be a model obtained by machine learning the relationship between the position of the tumor and the position of the diaphragm in the first image.

[0053] The error calculated by the error calculation unit 180 may be the three-dimensional Euclidean distance between the centroid position of the object in the second image and the tracking position by the tracking unit 170.

[0054] The error calculated by the error calculation unit 180 may be the distance between the trajectory obtained from the tumor position of the first image, which is two or more still images or a video, and the tracking position determined by the tracking unit 170.

[0055] The model correction unit 190 may correct the tracking model based on the difference between the image pattern of the tumor in the first image and the image pattern of the tumor in the second image.

[0056] The model correction unit 190 may correct the tracking model based on the relationship between the tumor's position in the second image and the diaphragm's boundary position.

[0057] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents.

[0058] 1...Treatment system 10...Treatment device 12...Treatment table 14...Treatment table control unit 100...Medical image processing device 110...First image acquisition unit 120...Second image acquisition unit 130...3D-3D positioning execution unit 140...Region information acquisition unit 150...Region estimation unit 160...Tracking model acquisition unit 170...Tracking unit 180...Error calculation unit 190...Model correction unit

Claims

1. A medical image processing apparatus comprising: a first image acquisition unit that acquires a first image, which is a three-dimensional fluoroscopic image of a patient taken in a first stage; a second image acquisition unit that acquires a second image, which is a three-dimensional fluoroscopic image of the patient taken in a second stage that is later than the first stage; a 3D-3D positioning execution unit that aligns the position of the second image with the position of the first image based on the pixel values ​​of the first and second images; a region information acquisition unit that acquires region information corresponding to a region of the first image; a region estimation unit that estimates a region in the second image corresponding to the region information; a tracking model acquisition unit that acquires a tracking model created based on the first image; a tracking unit that tracks an object in the second image using the tracking model; and an error calculation unit that calculates the error between a tracking position, which is the position of the object in the second image tracked by the tracking unit, and the position of the object in the second image.

2. The medical image processing apparatus according to claim 1, further comprising a model correction unit for correcting the tracking model based on the second image and the error.

3. The medical image processing apparatus according to claim 1, wherein the region estimated by the region estimation unit is the extent of the tumor in the second image.

4. The region estimated by the region estimation unit is the range of the marker in the second image, as described in claim 1.

5. The medical image processing apparatus according to claim 1, wherein the tracking model is a machine learning model using the image pattern of the tumor in the first image.

6. The medical image processing apparatus according to claim 1, wherein the tracking model is a machine learning model that determines the relationship between the position of the tumor and the position of the diaphragm in the first image.

7. The medical image processing apparatus according to claim 1, wherein the error is the three-dimensional Euclidean distance between the centroid position of the object in the second image and the tracking position by the tracking unit.

8. The medical image processing apparatus according to claim 1, wherein the error is the distance between the trajectory obtained from the tumor position of the first image, which is two or more still images or a video, and the tracking position by the tracking unit.

9. The medical image processing apparatus according to claim 2, wherein the model correction unit corrects the tracking model based on the difference between the image pattern of the tumor in the first image and the image pattern of the tumor in the second image.

10. The medical image processing apparatus according to claim 2, wherein the model correction unit corrects the tracking model based on the relationship between the position of the tumor and the boundary position of the diaphragm in the second image.

11. A treatment system comprising: a medical image processing device according to any one of claims 1 to 10; an irradiation unit for irradiating the patient with radiation; a first imaging device for capturing the second image; a bed for placing and fixing the patient; and a bed control unit for controlling the movement of the bed.

12. A medical image processing method comprising: a computer acquiring a first image, which is a three-dimensional fluoroscopic image of a patient taken in a first stage; acquiring a second image, which is a three-dimensional fluoroscopic image of the patient taken in a second stage after the first stage; performing 3D-3D positioning to align the position of the second image with the position of the first image based on the pixel values ​​of the first and second images; acquiring region information corresponding to the region of the first image; estimating the region in the second image corresponding to the region information; acquiring a tracking model created based on the first image; tracking an object in the second image using the tracking model; and calculating the error between the tracking position, which is the position of the object in the second image tracked by the tracking model, and the position of the object in the second image.

13. A program that causes a computer to acquire a first image, which is a three-dimensional fluoroscopic image of the patient taken in the first stage; acquire a second image, which is a three-dimensional fluoroscopic image of the patient taken in the second stage, which is after the first stage; perform 3D-3D positioning to align the position of the second image with the position of the first image based on the pixel values ​​of the first and second images; acquire region information corresponding to the region of the first image; estimate the region in the second image corresponding to the region information; acquire a tracking model created based on the first image; track an object in the second image using the tracking model; and calculate the error between the tracking position, which is the position of the object in the second image tracked by the tracking model, and the position of the object in the second image.

14. A recording medium on which a program is recorded, wherein the program causes a computer to: acquire a first image, which is a three-dimensional fluoroscopic image of a patient taken in a first stage; acquire a second image, which is a three-dimensional fluoroscopic image of the patient taken in a second stage after the first stage; perform 3D-3D positioning to align the position of the second image with the position of the first image based on the pixel values ​​of the first and second images; acquire region information corresponding to the region of the first image; estimate the region in the second image corresponding to the region information; acquire a tracking model created based on the first image; track an object in the second image using the tracking model; and calculate the error between the tracking position, which is the position of the object in the second image tracked by the tracking model, and the position of the object in the second image.

Images