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

Abstract

A medical image processing apparatus according to an embodiment includes a first image acquisition unit, a second image acquisition unit, a region acquisition unit, an image similarity calculation unit, a cost calculation unit, and a registration unit. The first image acquisition unit acquires a first image obtained by capturing an image of the body of a patient. The second image acquisition unit acquires a second image of the body of the patient captured at a time different from that of the first image. The region acquisition unit acquires two or more regions corresponding to the first image or the second image. The image similarity calculation unit calculates the similarity between the first image and the second image. The cost calculation unit calculates a cost based on the positional relationship of the regions. The registration unit finds the relative position of the first image with respect to the second image so that the image similarity is high and the cost is low.

Description

Technical Field

[0001] Embodiments of the present invention relate to a medical image processing device, a treatment system, a medical image processing method, and a storage medium. Background Art

[0002] Radiation therapy is a treatment method that destroys a tumor (lesion) in a patient's body by irradiating it with radiation. Since radiation may affect normal tissue in a patient's body if it is irradiated, it is necessary to accurately irradiate the tumor with radiation in radiation therapy. Therefore, when performing radiation therapy, first, at the stage of treatment planning, for example, computed tomography (CT) is performed in advance to grasp the position of the tumor in the patient's body in three dimensions. Then, based on the grasped position of the tumor, the direction of the irradiation of the radiation and the intensity of the irradiated radiation are planned. After that, at the stage of treatment, the patient's position is aligned with the patient's position at the stage of treatment planning, and the tumor is irradiated with radiation according to the irradiation direction and irradiation intensity planned at the stage of treatment planning.

[0003] In the patient alignment during the treatment phase, the patient's in-vivo fluoroscopic image taken with the patient lying on the bed just before the start of treatment is matched with a digitally reconstructed radiograph (DRR) image obtained by virtually reconstructing the fluoroscopic image based on the three-dimensional CT image taken during the treatment plan, and the patient's position offset between the images is calculated. Then, the bed is moved based on the calculated offset to align the position of the tumor or bone in the patient's body with the position during the treatment plan.

[0004] The patient's position offset is obtained by searching the position in the CT image by reconstructing a DRR image that is most similar to the fluoroscopic image. Currently, a variety of methods for automating the search for the patient's position by computer have been proposed. However, the results of the automatic search have always been confirmed by the user (doctor, etc.) by comparing the fluoroscopic image with the DRR image.

[0005] At this time, it is sometimes difficult to visually confirm the position of the tumor reflected in the fluoroscopic image. This is because the X-ray transmittance of the tumor is higher than that of bones, etc., so the tumor will not be clearly reflected in the fluoroscopic image. Therefore, during treatment, CT images are also taken instead of fluoroscopic images to confirm the position of the tumor. In this case, the displacement of the patient's position is calculated by image matching the CT images taken during the treatment planning and the CT images taken during the treatment stage, that is, image matching of the CT images with each other.

[0006] In the image matching between CT images, the position of one CT image is offset while the position that is most similar to the other CT image is found. As an example of a method for image matching between CT images, there is a method disclosed in Patent Document 1. In the method disclosed in Patent Document 1, an image of the periphery of the tumor contained in the CT image taken during the treatment plan is prepared as a template, and template matching is performed with respect to the CT image taken during the treatment phase, thereby searching for the position of the most similar image as the position of the tumor. Then, based on the searched position, the offset of the patient's position is calculated, and the bed is moved according to the offset in the same manner as described above to align the patient's position with the same posture as during the treatment plan. In the method disclosed in Patent Document 1, not only is the prepared template scanned three-dimensionally, but a search method is also mentioned in which the posture is changed by tilting the template, etc. for scanning.

[0007] However, in the method disclosed in Patent Document 1, it is important to align the position of the tumor periphery of interest with the CT image of the tumor periphery prepared as a template. Therefore, in the method disclosed in Patent Document 1, the position of the patient's internal tissue is not necessarily accurately aligned at a position other than the tumor periphery. That is, even if the position of the patient is aligned by the method disclosed in Patent Document 1, the energy of the planned radiation may not be imparted to the tumor due to the patient's internal tissue located on the path through which the radiation passes.

[0008] In fact, radiation used in radiotherapy loses energy when passing through matter. Therefore, in conventional treatment planning, the radiation irradiation method is determined by calculating the energy loss of the radiation that is hypothetically irradiated based on the captured CT images. If this is taken into account, when aligning the patient's position during the treatment phase, it is important to align the tissues in the patient's body that are on the path through which the irradiated radiation passes.

[0009] As an example of a method for performing image matching between CT images focusing on this point, there is a method disclosed in Patent Document 2. In the method disclosed in Patent Document 2, image matching of CT images is performed using a CT image obtained by calculating the arrival energy of radiation for each pixel and performing conversion. However, in the method disclosed in Patent Document 2, when performing image matching, a DRR image reconstructed from the converted CT image is also used for image matching. That is, in the method disclosed in Patent Document 2, the image used for image matching is also in a state where the stereoscopic image information possessed by the CT image is lost.

[0010] Furthermore, a method of combining the method disclosed in Patent Document 1 with the method disclosed in Patent Document 2, and using the converted CT image to perform patient alignment by template matching can be considered. However, since the calculation method of the arrival energy changes depending on the direction of the irradiated radiation, if the posture of the template used in template matching is changed, the arrival energy needs to be recalculated each time. Therefore, even in the case of combining the method disclosed in Patent Document 1 with the method disclosed in Patent Document 2, if it is considered that a plurality of templates need to be prepared in advance according to the posture or that the position needs to be aligned by focusing on the periphery of the tumor, it is not easy to perform alignment including the patient's internal tissue on the path through which the radiation passes.

[0011] Patent document 3 discloses a method of calculating the water equivalent thickness associated with the energy attenuation on the path through which radiation passes from a CT image, and correcting the deviation of the patient's position so that the energy imparted to the tumor by the irradiated radiation is close to the energy amount in the treatment plan.

[0012] Thus, the CT image alignment methods of the above-mentioned Patent Documents 1 to 3 only consider the degree of consistency between the images, and do not use the positions of the tumor and the irradiation field during treatment, or the so-called organs at risk, to align the CT images. As a result, the accuracy of the CT image alignment is sometimes low.

[0013] Prior art literature

[0014] Patent Literature

[0015] Patent Document 1: Japanese Patent No. 5693388

[0016] Patent Document 2: U.S. Patent Application Publication No. 2011 / 0058750

[0017] Patent Document 3: Japanese Patent Application Publication No. 2022-029277 Summary of the invention

[0018] Problems to be solved by the invention

[0019] The problem to be solved by the present invention is to provide a medical image processing device, a treatment system, a medical image processing method, a program and a storage medium. When CT images of a patient taken during treatment planning and treatment are aligned with each other, they can perform high-speed and high-precision image matching by also utilizing the positional relationship of the tumor, irradiation field, organs at risk, etc. during treatment.

[0020] Means for solving problems

[0021] The medical image processing device of the embodiment comprises a first image acquisition unit, a second image acquisition unit, a region acquisition unit, an image similarity calculation unit, a cost calculation unit and a registration unit. The first image acquisition unit acquires a first image obtained by photographing the inside of a patient's body. The second image acquisition unit acquires a second image of the inside of a patient's body photographed at a different time from the first image. The region acquisition unit acquires two or more regions corresponding to either or both of the first image or the second image. The image similarity calculation unit calculates the similarity between the first image and the second image. The cost calculation unit calculates the cost based on the positional relationship of the regions. The registration unit calculates the relative position of the first image with respect to the second image in such a way that the image similarity is high and the cost is low.

[0022] Effects of the Invention

[0023] According to the present invention, there can be provided a medical image processing device, a treatment system, a medical image processing method, a program and a storage medium that can perform high-speed and high-precision image matching by also utilizing the positional relationship of the tumor, irradiation field, organs at risk, etc. during treatment when aligning CT images of a patient taken during treatment planning and during treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a block diagram showing a schematic configuration of a treatment system including the medical image processing apparatus according to the first embodiment.

[0025] Figure 2 These are diagrams for explaining CTV and PTV in the treatment planning stage and treatment stage.

[0026] Figure 3 This is a block diagram showing a schematic configuration of the medical image processing apparatus 100 according to the first embodiment.

[0027] Figure 4 This is a diagram for explaining an example of the relationship between the emission of radiation and the irradiation target of the radiation in the treatment system.

[0028] Figure 5 This is a diagram for explaining another example of the relationship between the emission of radiation and the irradiation target of the radiation in the treatment system.

[0029] Figure 6 This is a flowchart showing an example of the flow of processing executed by the medical image processing apparatus according to the first embodiment.

[0030] Figure 7 This is a flowchart showing another example of the flow of processing executed by the medical image processing apparatus 100 according to the first embodiment.

[0031] Figure 8 This is a block diagram showing a schematic configuration of a medical image processing apparatus 100B according to the second embodiment.

[0032] Fig. 9 This is a flowchart showing the flow of processing executed by the medical image processing apparatus 100B according to the second embodiment. DETAILED DESCRIPTION

[0033] Hereinafter, a medical image processing apparatus, a treatment system, a medical image processing method, a program, and a storage medium according to embodiments will be described with reference to the accompanying drawings.

[0034] (First Embodiment)

[0035] (Overall structure)

[0036] Figure 1 This is a block diagram showing a schematic structure of a treatment system including a medical image processing apparatus according to a first embodiment. The treatment system 1 includes, for example, a treatment apparatus 10 and a medical image processing apparatus 100. The treatment apparatus 10 includes, for example, a bed 12, a bed control unit 14, a computed tomography (CT) apparatus 16 (hereinafter referred to as "CT imaging apparatus 16"), and a treatment beam irradiation gate 18.

[0037] The bed 12 is a movable treatment table that fixes the subject (patient) P who receives radiotherapy in a lying position by, for example, fixing devices. The bed 12 moves the patient P in a fixed state to a ring-shaped CT imaging device 16 having an opening according to the control from the bed control unit 14. The bed control unit 14 controls the translation mechanism and the rotation mechanism provided on the bed 12 according to the movement amount signal output by the medical image processing device 100 to change the direction of irradiating the treatment beam B to the patient P fixed on the bed 12. The translation mechanism can drive the bed 12 in three-axis directions, and the rotation mechanism can drive the bed 12 around three axes. Therefore, the bed control unit 14 controls the translation mechanism and the rotation mechanism of the bed 12, for example, to move the bed 12 with six degrees of freedom. The degrees of freedom with which the bed control unit 14 controls the bed 12 may not be six degrees of freedom, but may be 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).

[0038] The CT imaging device 16 is an imaging device for performing three-dimensional computed tomography. The CT imaging device 16 is provided with a plurality of radiation sources inside the annular opening, and radiation for performing perspective on the body of the patient P is irradiated from each radiation source. That is, the CT imaging device 16 irradiates radiation from a plurality of positions around the patient P. The radiation irradiated from each radiation source in the CT imaging device 16 is, for example, X-rays. The CT imaging device 16 detects the radiation irradiated from the corresponding radiation source and reaching the body of the patient P through the plurality of radiation detectors arranged inside the annular opening. The CT imaging device 16 generates a CT image obtained by imaging the body of the patient P based on the energy of the radiation detected by each radiation detector. The CT image of the patient P generated by the CT imaging device 16 is a three-dimensional digital image in which the energy of the radiation is represented by a digital value. The CT imaging device 16 outputs the generated CT image to the medical image processing device 100. The CT imaging device 16 performs three-dimensional imaging of the inside of the patient P, that is, irradiation of radiation from each radiation source and generation of CT images based on radiation detected by each radiation detector, which is controlled by, for example, an imaging control unit (not shown).

[0039] The treatment beam irradiation gate 18 irradiates radiation for destroying the part of the treatment object existing in the body of the patient P, that is, the tumor (lesion), as a treatment beam B. The treatment beam B is, for example, X-rays, gamma rays, electron beams, proton beams, neutron beams, heavy particle beams, etc. The treatment beam B is irradiated linearly from the treatment beam irradiation gate 18 to the patient P (more specifically, the tumor in the body of the patient P). The irradiation of the treatment beam B in the treatment beam irradiation gate 18 is controlled, for example, by a treatment beam irradiation control unit (not shown). In the treatment system 1, the treatment beam irradiation gate 18 is an example of an "irradiation unit" in the claims.

[0040] In the treatment room where the treatment system 1 is installed, the Figure 1Then, in the treatment room where the treatment beam B is irradiated to the patient P, the setting position of the treatment beam irradiation gate 18, the direction (irradiation direction) of irradiating the treatment beam B, the setting position of the bed 12, the setting position of the CT imaging device 16, the imaging position of the CT image obtained by imaging the body of the patient P, etc. are grasped based on the three-dimensional coordinates of the preset reference position. In the following description, the three-dimensional coordinate system of the reference position preset in the treatment room is defined as the "indoor coordinate system". Moreover, in the following description, "position" is the coordinates of the three-axis directions (three dimensions) realized by the translation mechanism of the bed 12 according to the indoor coordinate system, and "posture" is the rotation angle around the three axes realized by the rotation mechanism of the bed 12 according to the indoor coordinate system. For example, the position of the bed 12 represents the position of a predetermined point included in the bed 12 with three-dimensional coordinates, and the posture of the bed 12 represents the rotation angle of the bed 12 with yaw, roll, and pitch.

[0041] In radiotherapy, a treatment plan is formulated under the conditions of a simulated treatment room. That is, in radiotherapy, the state of the patient P on the bed 12 in the treatment room is simulated, and the irradiation direction, intensity, etc. when irradiating the treatment beam B to the patient P are planned. Therefore, the CT image of the treatment plan stage (treatment plan stage) is given information such as parameters indicating the position and posture of the bed 12 in the treatment room. This is also the case with the CT image taken just before the radiotherapy or the CT image taken during the previous radiotherapy. That is, the CT image obtained by taking the patient P's body by the CT imaging device 16 is given parameters indicating the position and posture of the bed 12 at the time of the shooting.

[0042] exist Figure 1 The structure of the treatment device 10 including a CT imaging device 16 and a fixed treatment beam irradiation gate 18 is shown in the figure, but the structure of the treatment device 10 is not limited to the above structure. For example, the treatment device 10 may also be a CT imaging device that replaces the CT imaging device 16 and has a structure that allows a group of radiation sources and radiation detectors to rotate inside a circular opening, and a structure of an imaging device that generates an image obtained by three-dimensionally imaging the body of the patient P, such as a cone beam (Cone-Beam: CB) CT device, a magnetic resonance imaging (Magnetic Resonance Imaging: MRI) device, an ultrasonic diagnostic device, etc. For example, the treatment device 10 may also be a structure that includes multiple treatment beam irradiation gates, such as a treatment beam irradiation gate that irradiates the patient P from a horizontal direction. For example, the treatment device 10 may also be a structure that includes multiple treatment beam irradiation gates. Figure 1 A treatment beam irradiation gate 18 is shown relative to Figure 1The treatment device 10 can be configured to irradiate the patient P with a treatment beam from various directions by rotating around the patient P, such as rotating the rotation axis of the horizontal direction X by 360 degrees. For example, the treatment device 10 includes one or more imaging devices consisting of a set of a radiation source and a radiation detector instead of the CT imaging device 16. The imaging device can also be a device that is relatively stable. Figure 1 The rotation axis in the horizontal direction X shown rotates 360 degrees to capture the internal structure of the patient P from all directions. This structure is called a rotating gantry type treatment device. In this case, for example, Figure 1 The treatment beam irradiation gate 18 shown may also be configured to rotate simultaneously with the imaging device on the same rotation axis.

[0043] The medical image processing apparatus 100 performs processing for aligning the position of the patient P when performing radiotherapy based on the CT image output by the CT imaging device 16. More specifically, the medical image processing apparatus 100 performs processing for aligning the position of a tumor or tissue existing in the body of the patient P based on the CT image of the patient P captured before performing radiotherapy, for example, at the treatment planning stage, and the current CT image of the patient P captured by the CT imaging device 16 at the stage of treatment (treatment stage) of performing radiotherapy. In addition, the medical image processing apparatus 100 outputs to the bed control unit 14 a movement amount signal for moving the bed 12 in order to align the irradiation direction of the treatment beam B irradiated from the treatment beam irradiation gate 18 with the direction set at the treatment planning stage. That is, the medical image processing apparatus 100 outputs to the bed control unit 14 a movement amount signal for moving the patient P in a direction in which the treatment beam B is appropriately irradiated to the tumor or tissue to be treated in the radiotherapy.

[0044] The medical image processing device 100 and the bed control unit 14 and the CT imaging device 16 of the treatment device 10 may be connected via wired means or wirelessly via a LAN (Local Area Network) or a WAN (Wide Area Network).

[0045] (Treatment Plan)

[0046] Next, the treatment plan performed before the movement amount calculation process is performed in the medical image processing device 100 is described. In the treatment plan, the energy of the treatment beam B (radiation) irradiated to the patient P, the irradiation direction, the shape of the irradiation range, the distribution of the dose when the treatment beam B is irradiated multiple times, etc. are determined. More specifically, first, the planner of the treatment plan (doctor, etc.) specifies the boundary between the area of ​​the tumor (lesion) and the area of ​​the normal tissue, the boundary between the tumor and the important organs around it, etc. for the first image captured in the treatment planning stage (for example, a CT image captured by the CT imaging device 16). Then, in the treatment plan, the direction (the path through which the treatment beam B passes), the intensity, etc. of the irradiation treatment beam B are determined based on the depth from the body surface of the patient P to the position of the tumor and the size of the tumor calculated based on the information related to the tumor specified by the planner of the treatment plan (doctor, etc.).

[0047] Designating the boundary between the tumor area and the normal tissue area is equivalent to designating the position and volume of the tumor. The volume of the tumor is called the gross tumor volume (Gross Tumor Volume: GTV), clinical target volume (Clinical Target Volume: CTV), internal target volume (Internal Target Volume: ITV), planning target volume (PTV), etc. GTV is the volume of the tumor that can be confirmed by the naked eye from the image. In radiotherapy, it is the volume of the treatment beam B that needs to irradiate a sufficient dose. CTV is the volume that includes GTV and the potential tumor to be treated. ITV is a volume obtained by adding a predetermined margin (margin) to CTV, taking into account that CTV will move due to predictable physiological activities of patient P, etc. PTV (an example of "irradiation field") is a volume obtained by adding a margin to ITV, taking into account the error in the positioning of patient P during treatment. In these volumes, the relationship of the following formula (1) holds.

[0048] [Formula 1]

[0049] GTV∈CTV∈ITV∈PTV …(1)

[0050] On the other hand, the volume of an important organ located around a tumor that is highly sensitive to radiation and thus strongly shows the influence of the dose of radiation irradiated is called an organ at risk (OAR). As a volume with a predetermined margin added to the OAR, a planning organ at risk volume (PRV) is specified. The PRV is specified by adding a volume (area) that is irradiated to avoid the OAR that is not desired to be damaged by radiation as a margin. These volumes are related by the following formula (2).

[0051] [Formula 2]

[0052] OAR∈PRV …(2)

[0053] In the treatment planning stage, the direction (path) and intensity of the treatment beam B (radiation rays) to be irradiated to the patient P are determined based on a margin that takes into account errors that may occur in actual treatment.

[0054] Figure 2 These are diagrams for explaining CTV and PTV in the treatment planning stage and treatment stage. Figure 2 (a) represents the CTV and PTV in the treatment planning stage. Figure 2 (b) represents the CTV and PTV in the treatment phase. In the treatment planning phase, the planner of the treatment plan specifies the boundaries of the CTV and PTV on the CT image (more specifically, specifies the boundaries of the CTV and PTV on multiple two-dimensional tomographic images obtained by cutting the CT image from multiple directions, and converts them into three-dimensional CTV and PTV). The CTV and PTV specified in the treatment planning phase are copied to the CT image taken in the treatment phase using methods such as DIR (Deformable Image Registration) or optical flow described later. Figure 2 The CTV and PTV of the treatment phase of (b) are replicated in Figure 2 The CTV and PTV specified in the treatment planning stage of (a) are formed. During treatment, the treatment beam B is irradiated to the copied PTV.

[0055] However, if Figure 2 As shown in the figure mark SV of (b), depending on the type of organ, the error of CTV and PTV associated with replication is sometimes large, and CTV may deviate from PTV (between fractions). In this case, the area SV may deviate from the irradiation range of the treatment beam B, causing inconvenience to the treatment of the tumor. The medical image processing device 100 of the present invention addresses this problem.

[0056] (Structure of Medical Image Processing Apparatus)

[0057] Next, a medical image processing apparatus 100 according to the first embodiment will be described. Figure 3 1 is a block diagram showing a schematic configuration of a medical image processing apparatus 100 according to the first embodiment. The medical image processing apparatus 100 includes, for example, a first image acquisition unit 102, a second image acquisition unit 104, and a registration unit 110. The registration unit 110 includes, for example, a region acquisition unit 112, an image similarity calculation unit 114, and a cost calculation unit 116.

[0058] Some or all of the components of the medical image processing device 100 are implemented, for example, by executing a program (software) by a hardware processor such as a CPU (Central Processing Unit). Some or all of these components can be implemented by hardware (including circuitry) such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), GPU (Graphics Processing Unit), etc., or can be implemented by the collaboration of software and hardware. Some or all of the functions of these components can also be implemented by a dedicated LSI. The program may be pre-stored in a storage device (storage device with non-transitory storage medium) such as ROM (Read Only Memory), RAM (Random Access Memory), HDD (Hard Disk Drive), flash memory, etc., which is provided in the medical image processing apparatus 100, or may be stored in a removable storage medium (non-transitory storage medium) such as DVD or CD-ROM and installed in the HDD or flash memory provided in the medical image processing apparatus 100 by installing the storage medium in a drive device provided in the medical image processing apparatus 100. The program may also be downloaded from another computer device via a network and installed in the HDD or flash memory provided in the medical image processing apparatus 100.

[0059] The first image acquisition unit 102 acquires a first image related to the patient P before treatment and parameters indicating the position and posture when the first image was taken. The first image is a three-dimensional CT image that is taken by, for example, the CT imaging device 16 at the treatment planning stage when performing radiotherapy and indicates the three-dimensional shape of the body of the patient P. The first image is used to determine the direction (including the path such as the inclination and the distance) and the intensity of the treatment beam B irradiated to the patient P in radiotherapy. In the first image, the determined direction (irradiation direction) and the intensity of the treatment beam B are set. The first image is taken in a state where the position and posture (hereinafter referred to as "body position") of the patient P are maintained constant by being fixed on the bed 12. The parameters indicating the body position of the patient P when the first image was taken may be the position and posture (shooting direction, shooting magnification) of the CT imaging device 16 when the first image was taken, or may be, for example, setting values ​​set for the translation mechanism and the rotation mechanism provided on the bed 12 in order to maintain the position and posture of the bed 12, i.e., the body position of the patient P, constant when the first image was taken. The first image acquisition unit 102 outputs the acquired first image and parameters to the registration unit 110 .

[0060] The second image acquisition unit 104 acquires a second image related to the patient P taken immediately before the start of radiotherapy and parameters indicating the position and posture when the second image was taken. The second image is a three-dimensional CT image that is taken by, for example, a CT imaging device 16 in order to align the body position of the patient P when irradiating the treatment beam B during radiotherapy, and represents the three-dimensional shape of the body of the patient P. That is, the second image is an image taken by the CT imaging device 16 in a state where the treatment beam B is not irradiated from the treatment beam irradiation gate 18. In other words, the second image is a CT image taken at a time different from the time when the first image was taken. In this case, the first image and the second image are taken at different times, but the shooting method of each image is the same. Therefore, the second image is taken in a body position close to the same body position as when the first image was taken. The parameter indicating the body position of the patient P when the second image is taken may be the position and posture (shooting direction, shooting magnification) of the CT imaging device 16 when the second image is taken, or may be, for example, the position and posture of the diagnostic bed 12 when the second image is taken, that is, the setting value set for the translation mechanism and rotation mechanism provided on the diagnostic bed 12 in order to make the body position of the patient P close to the body position when the first image is taken. The second image acquisition unit 104 outputs the acquired second image and parameters to the registration unit 110.

[0061] In addition, the first image and the second image are not limited to CT images captured by the CT imaging device 16, and may be three-dimensional images captured by an imaging device different from the CT imaging device 16, such as a CBCT device, an MRI device, or an ultrasonic diagnostic device. For example, the first image may be a CT image, and the second image may be a three-dimensional image captured by an MRI device.

[0062] Furthermore, the first image and the second image may also be two-dimensional images such as X-ray fluoroscopic images. In this case, the first image acquisition unit 102 and the second image acquisition unit 104 may also acquire DRR images obtained by virtually reconstructing fluoroscopic images based on three-dimensional CT images, and use them as the first image and the second image, respectively. When the first image and the second image are two-dimensional images, the parameters representing the position and posture are the position of the image in the treatment room and the rotation angle within the plane.

[0063] (Generation of integral image)

[0064] When aligning the first image and the second image, for example, the position and posture of the second image arranged in the indoor coordinate system are fixed, and the first image is moved while finding the position in space where the image similarity (for example, the difference in pixel values ​​between the first image and the second image) is high. However, in this method, although the difference in pixel values ​​between the first image and the second image becomes smaller, it is not necessarily calculated to the extent that the dose distribution of the treatment beam B relative to the tumor specified by the planner (doctor, etc.) in the important treatment plan is consistent in radiotherapy. Since radiation (here, the treatment beam B) loses energy when passing through matter, the following operation is performed in the treatment plan: the energy loss amount of the radiation that is virtually irradiated is calculated using the CT image to determine the radiation irradiation method. If this is taken into account, when aligning the position of the patient P in the treatment stage, it is important to also align the tissues in the body of the patient P that exist on the path through which the irradiated treatment beam B passes.

[0065] With this situation as the background, in order to make the energy given to the tumor in the body of the patient P by the irradiated treatment beam B closer to the energy planned in the treatment planning stage, the first image acquisition unit 102 and the second image acquisition unit 104 generate an integral image (water equivalent thickness image) obtained by integrating the pixel values ​​(CT values) of the pixels (voxels) existing on the path through which the treatment beam B passes in the CT image, and acquire the generated integral images as the first image and the second image respectively. That is, the first image acquisition unit 102 and the second image acquisition unit 104 also function as the "image conversion unit" in the claims. The first image acquisition unit 102 and the second image acquisition unit 104 output the generated integral images, namely the first image and the second image, to the registration unit 110. Below, taking the first image acquisition unit 102 that calculates the first integral image corresponding to the first image as a CT image as an example, the outline of the calculation method of the integral image is described.

[0066] In the calculation of the first integral image performed by the first image acquisition unit 102, first, pixels located on the path through which the treatment beam B passes are extracted from the pixels contained in the first image. With regard to the path through which the treatment beam B passes, the path through which the treatment beam B irradiated from the treatment beam irradiation gate 18 passes through the patient P can be obtained as three-dimensional coordinates of an indoor coordinate system based on the irradiation direction of the treatment beam B contained in information related to its direction in the treatment room (hereinafter referred to as "direction information"). The direction information includes, for example, information indicating the irradiation direction of the treatment beam B and information indicating the moving direction of the bed 12. The direction information is information represented by a pre-set indoor coordinate system. The path through which the treatment beam B passes can also be obtained as a three-dimensional vector represented by the three-dimensional coordinates of the indoor coordinate system with the position of the treatment beam irradiation gate 18 as the starting point.

[0067] The first image acquisition unit 102 calculates a first integral image by integrating the pixel values ​​(CT values) of pixels (voxels) existing on the path through which the treatment beam B passes within the first image, based on the first image output by the first image acquisition unit 102, parameters representing the position and posture of the first image, and direction information.

[0068] Next, the irradiation direction of the treatment beam B irradiated from the treatment beam irradiation gate 18 will be described. In the following description, the path of the treatment beam B is assumed to be a three-dimensional vector. Figure 4 1 is a diagram illustrating an example of the relationship between the emission of radiation (treatment beam B) in the treatment system 1 and the irradiation target of the radiation (treatment beam B) (a tumor existing in the body of the patient P). Figure 4 , an example of a path of the treatment beam B irradiated from the treatment beam irradiation gate 18 to reach a region (range) of a tumor existing in the body of the patient P which is an irradiation target is shown. Figure 4 This is an example of a configuration in which the treatment beam B is emitted from the treatment beam irradiation gate 18 .

[0069] When the treatment beam irradiation gate 18 is configured to emit the treatment beam B, as shown in FIG. Figure 4 As shown, the treatment beam irradiation gate 18 has a planar emission port. The treatment beam B emitted from the treatment beam irradiation gate 18 reaches the tumor of the irradiation target via the collimator 18-1. That is, among the treatment beam B emitted from the treatment beam irradiation gate 18, only the treatment beam B' after passing through the collimator 18-1 reaches the tumor of the irradiation target. The collimator 18-1 is a metal device for cutting off the useless treatment beam B". The collimator 18-1 adjusts the area through which the treatment beam B passes, for example, according to the shape of the tumor of the irradiation target, so as not to irradiate the treatment beam B to areas other than the tumor present in the body of the patient P. The collimator 18-1 may be, for example, a multi-leaf collimator that can mechanically change the area where the useless treatment beam B" is cut off. Figure 4 An example of a case where a treatment beam B' after passing through the collimator 18-1 is irradiated to a tumor of an irradiation target in a first image is schematically shown. In this case, the starting point in the path of the treatment beam B' is the position of the exit point of the treatment beam B' within the range of the planar exit port of the treatment beam irradiation gate 18. The three-dimensional position of the treatment beam irradiation gate 18 is, for example, the position (coordinates) of the center of the plane of the exit port.

[0070] The first image acquisition unit 102 acquires direction information, in which the irradiation direction of the treatment beam B' is included as information indicating the irradiation direction of the treatment beam B. The first image acquisition unit 102 sets the path of the treatment beam B' to the tumor of the irradiation object in the first image as the path of the treatment beam B' irradiated in a predetermined three-dimensional space. Here, the position of the tumor of the irradiation object is represented by the position i in the indoor coordinate system, and the path b(i) of the treatment beam B' reaching the position can be discretely represented by a set of three-dimensional vectors as shown in the following formula (1).

[0071] [Formula 3]

[0072] B = {b i / i∈Ω} …(3)

[0073] The starting point of each path, i.e., the starting point of the three-dimensional vector b(i), is the position of the exit point of the treatment beam B' that reaches the tumor of the irradiation object in each path b(i). The three-dimensional position of the starting point is represented by S. In addition, Ω is the set of positions of the tumor position of the irradiation object, i.e., the PTV and GTV in the indoor coordinate system.

[0074] Figure 5This is a diagram for explaining another example of the relationship between the emission of radiation and the irradiation target of the radiation in the treatment system. Figure 5 An example of a path of the treatment beam B irradiated from the treatment beam irradiation gate 18 to reach a region (range) of a tumor existing in the body of the patient P which is an irradiation target is also shown. Figure 5 This is an example of a case where the treatment beam irradiation gate 18 is configured to scan the emitted treatment beam B. In the case of this configuration, Figure 5 As shown, the treatment beam irradiation gate 18 has one exit port and does not include a collimator 18-1. The treatment beam B emitted from one exit port of the treatment beam irradiation gate 18 is bent in direction by, for example, a magnet, so as to be scanned in a manner that covers (scans) the entire area of ​​the tumor to be irradiated, thereby irradiating the tumor to be irradiated. Figure 5 An example is schematically shown in which the irradiation direction of the treatment beam B is scanned to irradiate the tumor of the irradiation target in the first image. In this case, the starting point of each path of the scanned treatment beam B is the position of the exit port of the treatment beam irradiation gate 18. The three-dimensional position of the treatment beam irradiation gate 18 is the position (coordinate) of one exit port. In this case, the path b(i) of the treatment beam B arriving at a certain position i in the indoor coordinate system can be discretely expressed in the same way as the above formula (3).

[0075] The first image acquisition unit 102 acquires direction information, in which the irradiation direction of the treatment beam B is included as information indicating the irradiation direction of the treatment beam B. The first image acquisition unit 102 sets the path b(i) of the scanned treatment beam B to the coordinate i of the indoor coordinate system indicating the position of the tumor of the irradiation target in the first image as the path of the treatment beam B irradiated in the prescribed three-dimensional space. The path of the treatment beam B in this case can also be discretely represented by a set of three-dimensional vectors as shown in the above formula (3). The starting point of each path, that is, the starting point of the three-dimensional vector b(i), is the position of the exit port of the treatment beam irradiation gate 18.

[0076] Next, a method for calculating an integral image based on the set path of the treatment beam B is described. In the following description, the position i of a certain point in a predetermined three-dimensional space (indoor coordinate system) is represented as point i. Furthermore, the pixel value of a three-dimensional pixel corresponding to point i included in the first image that is virtually arranged in the predetermined three-dimensional space is represented as I. i Similarly, the pixel value of a three-dimensional pixel corresponding to a point i included in the second image which is virtually arranged in a predetermined three-dimensional space is represented as T i (x). When there is no pixel corresponding to the point i in the first image or the second image, the pixel value is set to "0". x is a parameter of a vector x representing the position and posture of the first image or the second image in a predetermined three-dimensional space.

[0077] The position of the exit port of the treatment beam irradiation gate 18 in the treatment beam B, that is, the vector from the three-dimensional vector 0 of the starting point S to the point i can be expressed by the following equation (4).

[0078] [Formula 4]

[0079] b(i)=i-0 …(4)

[0080] In this case, the first image acquisition unit 102 accumulates the pixel values ​​of each pixel on the path of the treatment beam B in the first image to point i, and the pixel values ​​of the pixels contained in the first integral image (hereinafter referred to as "integral pixel values") obtained by equation (5) can be calculated by the following equation (6).

[0081] [Formula 5]

[0082]

[0083] [Formula 6]

[0084]

[0085] Similarly, the second image acquisition unit 104 integrates the pixel values ​​of each pixel on the path of the treatment beam B to point i in the second image, and the integrated pixel value equation (7) of the pixels included in the second integral image can be calculated by the following equation (8).

[0086] [Formula 7]

[0087]

[0088] [Formula 8]

[0089]

[0090] In the above formula (6) and the above formula (8), t is an intermediate variable, and f(x) is a function for converting the pixel value (CT value) of the CT image. Function f(x) is, for example, a function that complies with a conversion table for converting the energy loss of radiation into water equivalent thickness. As described above, radiation loses energy when passing through a substance. At this time, the energy lost by the radiation is the energy corresponding to the CT value of the CT image. That is, the energy loss of radiation is not uniform, but varies according to the tissue in the body of the patient P, such as bone or fat. The water equivalent thickness is a value that expresses the energy loss of radiation that varies for each tissue (substance) as the thickness of the same substance, i.e., water, and can be converted based on the CT value. For example, when the CT value is a value representing bone, the water equivalent thickness is a larger value because the energy loss of radiation when passing through the bone is large. For example, when the CT value is a value representing fat, the water equivalent thickness is a smaller value because the energy loss of radiation when passing through the fat is small. For example, when the CT value is a value representing air, the water equivalent thickness is "0" because there is no energy loss when radiation passes through the air. By converting each CT value contained in the CT image into a water equivalent thickness, the energy loss calculated for each pixel on the path of the treatment beam B can be expressed in the same standard. As a conversion formula for converting a CT value into a water equivalent thickness, for example, a regression formula based on nonlinear conversion data obtained through experiments is used. Various documents have been published on nonlinear conversion data obtained through experiments. The function f(x) may, for example, be a function for performing identity mapping. Alternatively, the function f(x) may be defined in a switchable manner depending on the treatment site. As described above, the first image acquisition unit 102 and the second image acquisition unit 104 respectively acquire a first image and a second image as integral images.

[0091] (Estimation and acquisition of regions)

[0092] The region acquisition unit 112 acquires two or more regions corresponding to one or both of the first image and the second image from the first image and the second image, and outputs them to the cost calculation unit 116. More specifically, the region acquisition unit 112 acquires a region (PTV, CTV, etc.) including the position and volume of the tumor specified in the treatment plan from the first image, and acquires a region obtained by estimating the movement of the region specified in the first image from the second image.

[0093] To estimate the movement of the region, the region acquisition unit 112 obtains the movement of the region in the second image that is similar to the image in the tumor region specified in the first image. As this method, the region acquisition unit 112 uses, for example, DIR or optical flow technology.

[0094] As an example of a method for obtaining optical flow, the region acquisition unit 112 uses an image representing the region of the tumor specified in the first image as a template, performs template matching on the second image, and searches for the position of the most similar image as the position of the tumor in the second image. Then, the region acquisition unit 112 obtains the motion vector of the position of the tumor in the searched second image, and uses all the obtained motion vectors as a motion model. The region acquisition unit 112 may also divide the tumor region as a template into a plurality of small regions (hereinafter referred to as "small regions"), and use the images representing each of the divided small regions as each template. In this case, the region acquisition unit 112 performs template matching according to the template of each small region, and searches for the position of the tumor in the most similar second image according to each small region. Then, the region acquisition unit 112 obtains the motion vector of the position of the tumor in the corresponding second image according to each searched small region, and uses all the obtained motion vectors as motion models. The region acquisition unit 112 may also use the average vector or median vector of the obtained motion vector as a motion model.

[0095] As another example of a method for obtaining optical flow, the region acquisition unit 112 may also obtain the movement of a region in the second image that is similar to the distribution of pixel values ​​in the region of the tumor specified in the first image. As this method, the region acquisition unit 112 may, for example, use a technique for tracking an object by searching for a position similar to a histogram of pixel values ​​by means of mean shift or medoid shift. At this time, the region acquisition unit 112 generates a movement model using the distribution of a histogram of pixel values ​​obtained using all pixel values ​​in the region of the tumor specified in the first image. The region acquisition unit 112 may also divide the region of the tumor specified in the first image into a plurality of small regions, and generate a movement model corresponding to each small region using the distribution of a histogram of pixel values ​​obtained using pixel values ​​in the region according to each of the divided small regions. In this case, the region acquisition unit 112 may use a plurality of movement models corresponding to each small region together as a movement model group, or may use an average vector or a median vector of the movement model group as a movement model.

[0096] In addition, the region acquisition unit 112 may also set the region acquired from the first image to be the same as or smaller than the PTV set in the second image. Thus, the region acquired from the first image can be reliably included in the PTV of the second image. Furthermore, the region acquired by the region acquisition unit 112 may not be the entire region determined in the treatment plan, but may be a partial region of the PTV, OAR, etc.

[0097] (Calculation of image similarity)

[0098] The image similarity calculation unit 114 acquires the first image and the parameters indicating the position and posture thereof from the first image acquisition unit 102, acquires the second image and the parameters indicating the position and posture thereof from the second image acquisition unit 104, calculates the image similarity between the first image and the second image, and outputs the calculated image similarity to the registration unit 110. More specifically, for example, the image similarity calculation unit 114 obtains the absolute value of the difference between the pixel values ​​at the same spatial position of the first image and the second image according to the following equation (9), and calculates the sum of the differences in the entire image as the image similarity.

[0099] [Formula 9]

[0100] |I(x+Δx)-R(x plan )| 2 …(9)

[0101] In equation (9), Δx represents the offset between the position and posture of the first image and the second image, and x plan represents the coordinates of the PTV in the treatment plan, R(x plan ) represents the coordinate x plan The pixel value below.

[0102] In addition, the image similarity calculation unit 114 may use the normalized mutual correlation at the same spatial position of the first image and the second image as the similarity. In this case, the range of correlation is centered on the pixel of the calculation object, and a small area such as 3×3×3 is taken as the object. In addition, the image similarity calculation unit 114 may use the mutual information amount at the same spatial position of the first image and the second image as the similarity. In addition, when calculating the similarity, the image similarity calculation unit 114 may also focus on the pixels in the area corresponding to each image.

[0103] (Calculation of cost)

[0104] In radiotherapy, if the tumor deviates from the PTV, the dose cannot be given as planned, and there is a risk that the treatment effect cannot be fully achieved. In addition, if the dose given to the OAR exceeds the planned dose, there is a risk that the side effects will increase. These risks cannot be measured by the similarity of the images. Therefore, the indicator "cost" is introduced to realize such risks.

[0105] The cost calculation unit 116 calculates the cost based on the positional relationship between the regions using the two or more regions acquired from the region acquisition unit 112, and outputs the cost to the registration unit 110. More specifically, the cost calculation unit 116 calculates the cost based on the positional relationship between the regions according to the following equation (10).

[0106] [Formula 10]

[0107] λf o(x+Δx) …(10)

[0108] In formula (10), f0 represents the cost function, and λ represents the weight given to the cost. The cost function f0 is defined as: Figure 2 As shown in (b), the more the CTV in the treatment phase deviates from the range of the PTV, the larger the value of the cost function f0 is reflected as the cost. In this case, since the CTV in the treatment phase extends outside the PTV, there is a possibility that the dose to the tumor is lower than planned, and the treatment effect is reduced. Therefore, the value corresponding to the volume SV (or area) of the deviated part is defined as the cost function f0.

[0109] Alternatively, the cost function f0 may also be designed such that the closer the positions of the PTV and the OAR in the first image are, the higher the cost function f0 is. In this way, the cost function can be designed based on the positional relationship between two or more regions determined in the treatment plan. In addition, the cost function f0 may also be defined as the linear sum of multiple cost functions. In addition, for example, the narrower the PTV is, the larger the weight λ is, or the closer the positions of the PTV and the OAR are, the larger the weight λ is.

[0110] (Execution of Registration)

[0111] The registration unit 110 obtains the position of the first image in such a manner that the calculated image similarity is high and the cost is low based on the calculation results of the image similarity calculation unit 114 and the calculation results of the cost calculation unit 116. More specifically, the registration unit 110 first defines the cost function E(Δx) as the sum of the calculation results of the image similarity calculation unit 114 and the calculation results of the cost calculation unit 116 as shown in the following formula (11).

[0112] [Formula 11]

[0113] E(Δx)=|I(x+Δx)-R(x plan )| 2 +λf o (x+Δx) …(11)

[0114] If the Taylor expansion of equation (11) is performed around x, the following equation (12) is obtained.

[0115] [Formula 12]

[0116]

[0117] If the right side is differentiated by the shift amount Δx and set to 0 in order to find the minimum value of the offset amount Δx, the following equation (13) is obtained.

[0118] [Formula 13]

[0119]

[0120] If equation (13) is solved with respect to Δx, the following equation (14) is obtained.

[0121] [Formula 14]

[0122]

[0123] In formula (14), H is a Hessian matrix defined by the following formula (15). In formula (15), V represents a vector of the position and posture when the first image is configured in a predetermined three-dimensional space. The vector V has the same dimension as the number of axes represented by the above-mentioned direction information, for example, a six-dimensional vector in the case of the above-mentioned six degrees of freedom.

[0124] [Formula 15]

[0125]

[0126] The registration unit 110 substitutes the coordinate xplan representing the pre-prepared candidate position as the initial value of x in formula (11) to calculate the cost function E(Δx), and calculates Δx by formula (14). Next, the registration unit 110 uses the calculated Δx to update to x=x+Δx, and recalculates the cost function E(Δx). The registration unit 110 repeatedly calculates the cost function E(Δx) a predetermined number of times, or terminates the processing when the difference between the previous result and the current result of the cost function E(Δx) is lower than a threshold value. The registration unit 110 calculates the movement amount of the diagnostic bed 12 based on the offset Δx at the time when the processing ends, and outputs a movement amount signal.

[0127] In the above description, the registration unit 110 defines the cost function E(Δx) by equation (11). However, the present invention is not limited to this configuration, and the registration unit 110 may define the cost function E(Δx) by, for example, the following equation (16).

[0128] [Formula 16]

[0129] E(Δx)=|I(x+Δx)-R(x plan )| 2 +λ1f o (x+Δx)+λ2|Δx| 2

[0130] …(16)

[0131] Formula (16) is the formula (14) in which λ2|Δx| 2The formula added to the cost function E(Δx). In formula (16), λ1 corresponds to λ in formula (12), and λ2 represents the weight given to the offset |Δx|2. That is, by defining the cost function E(Δx) according to formula (16), the cost can be calculated by taking into account the offset of the CT image.

[0132] If the Taylor expansion of equation (16) is performed around x, the following equation (17) is obtained.

[0133] [Formula 17]

[0134]

[0135] If the right side is differentiated by the shift amount Δx and set to 0 in order to find the minimum value of the offset amount Δx, the following equation (18) is obtained.

[0136] [Formula 18]

[0137]

[0138] If equation (18) is solved with respect to Δx, the following equation (19) is obtained.

[0139] [Formula 19]

[0140]

[0141] In equation (19), H2 is the Hessian matrix defined by the following equation (20). In equation (16), by replacing λ2|Δx| 2 Appended to the cost function E(Δx), unlike the Hessian matrix H in equation (15), λ2E is added to H2 as a remainder.

[0142] [Formula 20]

[0143]

[0144] The registration unit 110 generates a coordinate x representing a candidate position prepared in advance. plan The cost function E(Δx) is calculated by substituting the initial value of x in equation (16), and Δx is calculated by equation (19). Next, the registration unit 110 uses the calculated Δx to update to x=x+Δx, and recalculates the cost function E(Δx). The registration unit 110 repeatedly calculates the cost function E(Δx) a predetermined number of times, or terminates the processing when the difference between the previous result and the current result of the cost function E(Δx) is lower than a threshold value. The registration unit 110 calculates the movement amount of the bed 12 based on the offset Δx at the time when the processing ends, and outputs a movement amount signal.

[0145] Next, refer to Figure 6The flow of processing executed by the medical image processing apparatus 100 according to the first embodiment will be described. Figure 6 This is a flowchart showing the flow of processing executed by the medical image processing apparatus 100 according to the first embodiment.

[0146] First, the medical image processing apparatus 100 acquires a first image and parameters indicating its position and posture, and a second image and parameters indicating its position and posture using the first image acquisition unit 102 and the second image acquisition unit 104 (step S100). Next, the medical image processing apparatus 100 acquires two or more regions corresponding to either or both of the first image and the second image using the region acquisition unit 112 (step S102).

[0147] Next, the medical image processing apparatus 100 uses the image similarity calculator 114 to calculate the similarity between the first image and the second image (step S104 ). Next, the medical image processing apparatus 100 uses the cost calculator 116 to calculate the cost based on the acquired positional relationship between the two or more regions (step S106 ).

[0148] Next, the medical image processing apparatus 100 calculates the sum of the similarity and the cost using the registration unit 110, and determines whether the number of calculations is greater than or equal to a predetermined number of times or whether the difference between the current value and the previous value of the calculated sum is within a threshold value (step S108). When it is determined that the number of calculations is within the predetermined number of times or the difference between the current value and the previous value of the calculated sum is within the threshold value, the medical image processing apparatus 100 calculates and outputs a movement amount signal corresponding to the offset Δx using the registration unit 110 (step S110). On the other hand, when it is determined that the number of calculations is not within the predetermined number of times and the difference between the current value and the previous value of the calculated sum is greater than the threshold value, the medical image processing apparatus 100 updates the parameter to x=x+Δx using the registration unit 110, and returns the process to step S104 again. Thus, the process of this flowchart ends.

[0149] In the above flowchart, the registration unit 110 determines whether the number of calculations is within a predetermined number of times or whether the difference between the current value and the previous value of the calculated sum is within a threshold value. However, the present invention is not limited to such a configuration. Figure 7 is a flowchart showing another example of the flow of processing performed by the medical image processing apparatus 100 according to the first embodiment. Figure 6 The description will focus on the differences in the processing of the flowchart.

[0150] In step S100, if the first image acquisition unit 102 acquires the first image, the first image acquisition unit 102 prepares a plurality of candidates for the position and posture of the first image (step S101). In step S104, the image similarity calculation unit 114 calculates the similarity for each of the prepared plurality of candidates, and in step S106, the cost calculation unit 116 calculates the cost for each of the prepared plurality of candidates. In step S107, the registration unit 110 selects the offset Δx with the smallest sum of the similarity and the cost among the plurality of candidates, and outputs a movement amount signal corresponding to the selected offset Δx (step S110). Thus, the processing of this flowchart ends.

[0151] According to the first embodiment described above, when CT images of a patient taken during treatment planning and treatment are aligned with each other, in addition to the similarity of the CT images, the positional relationship of the tumor, irradiation field, organs at risk, etc. during treatment is used to calculate the cost, and a movement amount signal with high similarity and low cost is output. This enables high-speed and high-precision image matching.

[0152] (Second Embodiment)

[0153] Next, the second embodiment is described. In the first embodiment, a method for obtaining the best position and posture among the candidate positions and postures prepared in advance is described. However, with respect to this method, the higher the degree of freedom of the position and posture, the greater their combination, and the more time-consuming the calculation is. For example, as parameters representing the position in three-dimensional space, there are three: the X-axis, the Y-axis, and the Z-axis, and as parameters representing the posture, there are three: the rotation angle centered on the XYZ axes, which is a combination of a total of six parameters. Against this background, the medical image processing device 100B of the second embodiment obtains the position and posture more efficiently than the first embodiment.

[0154] Figure 8 1 is a block diagram showing a schematic structure of a medical image processing apparatus 100B according to the second embodiment. The medical image processing apparatus 100B includes a first image acquisition unit 102, a second image acquisition unit 104, and a registration unit 110. The registration unit 110 includes a region acquisition unit 112, an approximate image calculation unit 115, and a movement cost calculation unit 117. The following description will focus on the structure that is different from the first embodiment.

[0155] The approximate image calculation unit 115 acquires the first image and the parameters indicating the position and posture thereof, acquires the second image and the parameters indicating the position and posture thereof from the second image acquisition unit 104, and calculates the approximate image at the position and posture of the first image. The approximate image calculation unit 115 outputs the calculated approximate image to the registration unit 110. Next, a specific method for calculating the approximate image will be described.

[0156] (Calculation of approximate images)

[0157] In the following description, pixels (voxels) included in the first image which are virtually arranged in a three-dimensional space defined by an indoor coordinate system are referred to as I. i (V). At pixel I i In (V), mathematical formula (19) represents the three-dimensional position in the indoor coordinate system.

[0158] [Formula 21]

[0159]

[0160] The vector V may also be a vector with fewer dimensions corresponding to the directions of the degrees of freedom when controlling the movement of the bed 12. For example, when the directions of the degrees of freedom for controlling the movement of the bed 12 are directions of four degrees of freedom, the vector V may also be a four-dimensional vector. On the other hand, the vector V may also be increased in dimension by adding the irradiation direction of the treatment beam B to the movement direction of the bed 12 based on the direction information output by the direction acquisition unit 106. For example, when the irradiation direction of the treatment beam B included in the direction information is two directions of the vertical direction and the horizontal direction, and the movement direction of the bed 12 is a direction of six degrees of freedom, the vector V may also be a vector with a total of eight dimensions.

[0161] The approximate image calculation unit 115 calculates an approximate image obtained by moving (translating and rotating) the first image by a small amount ΔV. Here, the amount ΔV is a small amount of movement that is preset as a parameter. The approximate image calculation unit 115 calculates (approximates) the pixel values ​​of each pixel I included in the first image by using the following equation (22) and Taylor expansion. i (V) The corresponding approximate image contains each pixel I i (V+ΔV).

[0162] [Formula 22]

[0163]

[0164] In the above equation (22), the third term ε on the right side is the pixel I i The terms in (V+ΔV) that are concentrated after the second order. It is a value of the first derivative representing the amount of change of the vector that changes for each degree of freedom in the three-dimensional space expanded by the vector V. The vector V is represented by the same dimension as the following vector V: the vector V represents the change in the pixel value (for example, CT value) of the corresponding pixel at the same position i in the indoor coordinate system in the first image before the movement (before the approximation) and the approximation image after the slight movement. For example, when the movement direction of the bed 12 is a direction with six degrees of freedom, the pixel I at the center of the indoor coordinate system in the first image is i (V) corresponds to the six-dimensional vector It is expressed by the following formula (23).

[0165] [Formula 23]

[0166]

[0167] In the above formula (23), Δθx, Δθy, and Δθz represent the rotation angles around the three axes in the indoor coordinate system when the three axes are set as the x-axis, the y-axis, and the z-axis, and Δtx, Δty, and Δtz represent the translational movement along the respective axes. Each element on the right side of the above formula (23) represents the pixel value at position i in the indoor coordinate system of the first image. For example, the first element on the right side of the above formula (23), "Formula (24)", is to make the pixel I at position i in the indoor coordinate system of the first image i (V) The pixel value when rotated by the rotation angle Δθx around the x-axis. In this case, the mathematical formula (25) is represented by the following formula (26). The other elements on the right side of the above formula (23) can also be represented in the same way, but detailed descriptions of each element are omitted.

[0168] [Formula 24]

[0169]

[0170] [Formula 25]

[0171]

[0172] [Equation 26]

[0173]

[0174] The approximate image calculation unit 115 outputs the approximate image obtained by calculating the movement (translation and rotation) of the first image by a small movement amount ΔV as described above to the registration unit 110. When the registration unit 110 outputs information indicating that there is a misalignment between the first image and the second image, the approximate image calculation unit 115 similarly calculates a new approximate image obtained by further moving (translation and rotation) the first image by a small movement amount ΔV, and outputs the calculated new approximate image to the registration unit 110.

[0175] The travel cost calculation unit 117 acquires two or more regions from the region acquisition unit 112, calculates travel costs based on the positional relationship between the regions, and outputs the calculated travel costs to the registration unit 110. Here, the travel costs refer to costs based on the positional relationship between the regions, as in the first embodiment.

[0176] In the second embodiment, the cost is defined as a function f that depends on V representing the position and posture of the first image: i (V). Here, i represents the position in the indoor coordinate system represented by equation (21), and V represents the vector of the position and posture when the first image is arranged in the indoor coordinate system. In addition, the position and posture of the second image are omitted because they are fixed.

[0177] The movement cost calculation unit 117 converts the first image into a region image consisting of pixel values ​​in which flag information indicating whether the image is included in the region is embedded in each pixel, and calculates an approximate image after making it correspond to the positions and postures of the first image and the second image. More specifically, the movement cost calculation unit 117 calculates (approximately) the image f using the following equation (27) and Taylor expansion. i (V) The first region image f when it moves by ΔV i (V+ΔV).

[0178] [Formula 27]

[0179]

[0180] In the above formula (27), the third term ε on the right side is f i The terms in (V+ΔV) that are concentrated after the second order. z i(V) Similarly, f i (V) is a vector of the same dimension as the following V: V represents the change in the pixel value at the same position i in the indoor coordinate system in the first area image after a small shift along each axis of the space expanded by the vector V and the first area image before the shift. Similarly, in f i When (V) is a six-dimensional vector, f i (V) is represented by the following formula (28).

[0181] [Equation 28]

[0182]

[0183] In the above formula (28), Δθx, Δθy, and Δθz represent the rotation angles around the three axes in the indoor coordinate system when the three axes are set as the x-axis, the y-axis, and the z-axis, and Δtx, Δty, and Δtz represent the translational movement along the respective axes. Each element on the right side of the above formula (28) represents the pixel value at position i in the indoor coordinate system of the first image. For example, the first element on the right side of the above formula (28), "Formula (29)", is to make the pixel f at position i in the indoor coordinate system of the first image i (V) The pixel value when rotated by the rotation angle Δθx around the x-axis. In this case, the mathematical formula (30) is represented by the following formula (31). The other elements on the right side of the above formula (28) can also be represented in the same way, but detailed descriptions of each element are omitted.

[0184] [Equation 29]

[0185]

[0186] [Formula 30]

[0187]

[0188] [Equation 31]

[0189]

[0190] (Execution of Registration)

[0191] The registration unit 110 acquires the first image and the parameters indicating the position and posture thereof from the first image acquisition unit 102, acquires the second image and the parameters indicating the position and posture thereof from the second image acquisition unit 104, acquires two or more regions from the region acquisition unit 112, and calculates the offset ΔV between the first image and the second image based on the calculation result of the approximate image calculation unit 115 and the calculation result of the movement cost calculation unit 117. The registration unit 110 outputs a movement amount signal corresponding to the calculated offset ΔV. More specifically, the registration unit 110 calculates the offset ΔV according to the following formula (32).

[0192] [Formula 32]

[0193]

[0194] In the above formula (32), Ω is the total number of pixels I contained in the area where the first image and the second image overlap in the indoor coordinate system. i(V) is a set of positions i. The set Ω may also be a set of positions representing clinically significant spatial regions such as PTV, GTV, OAR, etc. specified by the planner (doctor, etc.) in the treatment plan when irradiating the tumor region with the treatment beam B. In addition, the set Ω may also be a set of positions representing a spatial (sphere, cube, cuboid) region of a specified size centered on the beam irradiation position of the indoor coordinate system. The specified size is set based on the size of the patient P or the average size of the human body. Alternatively, the set Ω may also be set to a range in which the PTV or GTV is enlarged at a specified scale. In formula (32), the cost function is defined by the following formula (33).

[0195] [Formula 33]

[0196] E(ΔV,Ω)-∑ i∈Ω [I i (V+ΔV)-T i (V plan )] 2 +λ∑ i∈Ω f i (V+ΔV)

[0197] …(33)

[0198] In formula (33), T i (V plan ) represents the configuration V at position i in the indoor coordinate system plan The pixel value of the second image. In addition, λ is an adjustment parameter. For example, in the case of attaching importance to the risk of the aforementioned treatment, λ is set to a large value in advance. For example, the larger the number of pixels included in Ω, the larger the value of λ is set. This is because in the region, the pixel value of the part with uniform pixel value is is zero, The non-zero part of is only around the boundary. Therefore, the calculated cost is relatively low.

[0199] The cost function E(ΔV, Ω) used by the registration unit 110 to compare the first image and the second image may be a cost function set in two unconnected spaces as represented by the following equation (34).

[0200] [Equation 34]

[0201] E(ΔV,Ω)=λE(ΔV,Ω1)+(1-λ)E(ΔV,Ω2)+λΣ i∈Ω f i (V+ΔV)

[0202] …(34)

[0203] The cost function E(ΔV, Ω) used by the registration unit 110 to compare the first image and the second image may be a cost function represented by the following equation (36) using the function equation (35) that specifies the weight corresponding to the position i in the indoor coordinate system.

[0204] [Formula 35]

[0205]

[0206] [Equation 36]

[0207] E(ΔV,Ω)=∑ i∈Ω [w(i)(I i (V+ΔV)-T i (V))] 2 +λ∑ i∈Ω f i (V+ΔV)

[0208] …(36)

[0209] In the function mathematical formula (36), w(i) is a function that returns, as a return value, a value corresponding to the position i and the path of the irradiated treatment beam B. The function w(i) is, for example, a binary function that returns "1" when the position i is a position on the path through which the treatment beam B passes, and returns "0" when the position i is not a position on the path through which the treatment beam B passes. The function w(i) may also be, for example, a function that returns a higher value as the distance between the position i and the path through which the treatment beam B passes is closer.

[0210] Function w(i), for example, may also be a function that returns a value corresponding to the position i and a set of positions that represent a clinically significant spatial region when irradiating the tumor region with the treatment beam B, such as PTV, GTV, OAR, etc., specified by the planner (doctor, etc.) in the treatment plan. Function w(i), for example, may also be a function that returns a binary value, and returns "1" when position i is a set of positions representing a spatial region, and returns "0" when position i is not a set of positions representing a spatial region. Function w(i), for example, may also be a function that returns a higher value the closer the distance between position i and the spatial region.

[0211] When equation (32) is rewritten using the approximate image acquired by the approximate image calculation unit 115, the following equation (37) is obtained.

[0212] [Formula 37]

[0213]

[0214] In the above equation (37), the pixel I of the approximate image output by the approximate image calculation unit 115 is ignored. i The third term ε on the right side of the above equation (22) represented by (V+ΔV) is because ε, which is concentrated after the second order in the above equation (22), is a very small value, so even if it is ignored, it will not have a significant impact on the subsequent processing.

[0215] Regarding the right side of the above formula (37), if the right side is differentiated by the movement amount ΔV and set to 0 in order to obtain the minimum value of the movement amount ΔV, the movement amount ΔV is expressed by the following formula (38).

[0216] [Equation 38]

[0217]

[0218] Here, H on the right side of the above equation (38) is the Hessian matrix defined by equation (15).

[0219] The registration unit 110 uses the movement amount ΔV obtained by the above equation (38) to update the vector V of the position and posture of the first image as shown in the following equation (39).

[0220] [Equation 39]

[0221] V←V+ΔV …(39)

[0222] In the above formula (39), the vector V of the position and posture of the updated first image is set to vector V1.

[0223] The registration unit 110 repeatedly calculates the movement amount ΔV using the above formula (38) until the change in the vector V1 of the updated first image decreases. The so-called until the change in the vector V1 decreases means that the norm of the movement amount ΔV, that is, the offset between the position and posture of the first image and the second image is below a predetermined threshold. In other words, it is determined that the body position of the patient P shown in the second image is consistent with the body position of the patient P shown in the treatment planning stage of the first image. The norm of the movement amount ΔV can be any norm of a vector, for example, any one of the l0 norm, l1 norm or l2 norm can be used.

[0224] In the case where the set Ω is the area of ​​PTV and GTV as described above, if the position and posture of the first image are updated, the elements of the set Ω also need to be updated. That is, this is because the set Ω is a set of coordinate positions in the indoor coordinate system, and its position will change as the first image moves in the indoor coordinate system. In order to eliminate the need for such an update, it is ideal that the first image whose position and posture are updated does not contain the area of ​​the specified set Ω. For example, it is also possible to set the CT image taken immediately before treatment (the second image so far) as the first image, and replace the CT image containing the treatment plan information (the first image so far) with the second image.

[0225] The calculation of the movement amount ΔV in the registration unit 110 may also be repeated until the number of repeated calculations that is set in advance exceeds. In this case, the time required for the registration unit 110 to calculate the movement amount ΔV can be shortened. However, in this case, although the registration unit 110 will end the calculation of the movement amount ΔV at the moment when the number of repeated calculations that is set in advance exceeds, the norm of the movement amount ΔV may not necessarily become below the prescribed threshold. In other words, it can also be considered that the possibility of failure in the calculation of the alignment of the patient P is high. In this case, the registration unit 110 may also output a warning signal indicating that the calculation of the movement amount ΔV has been ended due to exceeding the number of repeated calculations that is set in advance to a warning unit (not shown) of the medical image processing device 100B or the treatment system 1. Thus, the warning unit (not shown) can notify the implementer of the radiotherapy, such as a doctor, that is, the user of the treatment system 1, that the calculation of the alignment of the patient P may have failed.

[0226] The registration unit 110 calculates the movement amount ΔV calculated as described above, that is, the offset of the position and posture between the first image and the second image, according to each degree of freedom in the above formula (5). Then, the registration unit 110 determines the movement amount (translation amount and rotation amount) of the bed 12 based on the calculated offset amount in each degree of freedom. At this time, the registration unit 110, for example, sums up the movement amount ΔV moved when calculating the approximate image based on the first image according to each degree of freedom. Then, the registration unit 110 determines the movement amount of the bed 12 by the summed movement amount that moves the position of the current patient P according to each degree of freedom. Then, the registration unit 110 outputs a movement amount signal representing the determined movement amount of the bed 12 to the bed control unit 14.

[0227] Next, refer to Fig. 9 The flow of processing executed by the medical image processing apparatus 100B will be described. Fig. 91 is a flowchart showing the flow of processing performed by the medical image processing apparatus 100B. First, the medical image processing apparatus 100 uses the first image acquisition unit 102 and the second image acquisition unit 104 to acquire a first image and parameters indicating its position and posture, and a second image and parameters indicating its position and posture (step S200). Next, the medical image processing apparatus 100B uses the approximate image calculation unit 115 to calculate an approximate image of the first image using the above-mentioned method (step S202).

[0228] Next, the medical image processing apparatus 100 calculates the difference between the first image and the second image using the image similarity calculation unit 114 (step S204). Next, the medical image processing apparatus 100 uses the movement cost calculation unit 117 to convert two or more regions corresponding to one or both of the first image or the second image acquired from the region acquisition unit 112 into a region image consisting of pixel values ​​in which flag information indicating whether each pixel is included in the region is embedded, and calculates an approximate image thereof after making it correspond to the positions and postures of the first image and the second image (step S206).

[0229] Next, the medical image processing apparatus 100B uses the registration unit 110 and the above-mentioned method to calculate the movement amount ΔV based on the approximate image of the first image, the difference between the first image and the second image, and the approximate image of the regional image (step S208). Next, the medical image processing apparatus 100B uses the registration unit 110 to determine whether the calculated movement amount ΔV satisfies the termination condition (for example, Figure 6 Similarly to the flowchart of , it is determined whether the number of calculations is greater than the specified number of times, or whether the difference with the previous value is within the threshold value) (step S210). When it is determined that the calculated movement amount ΔV satisfies the end condition, the medical image processing device 100B outputs the determined movement amount ΔV as a movement amount signal. On the other hand, when it is determined that the end condition is not met, the medical image processing device 100B uses the registration unit 110 to update the parameter to x=x+Δx (step S212), and returns the processing to step S202 again. Thus, the processing of this flowchart ends.

[0230] According to the second embodiment described above, when aligning CT images of a patient taken during treatment planning and treatment, the approximate image is used to calculate the cost and output a high-similarity and low-cost movement amount signal, thereby enabling high-speed and high-precision image matching.

[0231] Some embodiments of the present invention have been described, but these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the scope of the main purpose of the invention. These embodiments and their variations are included in the scope and main purpose of the invention, and are also included in the invention described in the claims and their equivalents.

Claims

1. A medical image processing device, characterized in that: have: A first image acquisition unit that acquires a first image obtained by photographing the inside of a patient's body; a second image acquisition unit for acquiring a second image captured inside the patient's body at a different time than the first image; A region acquisition unit, which acquires two or more regions corresponding to either or both of the first image or the second image; an image similarity calculation unit, which calculates the similarity between the first image and the second image; a cost calculation unit that calculates a cost based on the positional relationship of the areas; as well as A registration unit is configured to obtain a relative position of the first image with respect to the second image so that the similarity is high and the cost is low.

2. The medical image processing device according to claim 1, characterized in that: The registration unit comprises: an approximate image calculation unit that calculates an approximate image generated by shifting the first image by a predetermined width for each degree of freedom of the patient's body position change; and a movement cost calculation unit that moves the region by a predetermined width for each degree of freedom of the patient's body position change and calculates a cost change amount relative to before the movement as the movement cost, The registration unit determines a movement amount of the first image based on a shift amount between the first image and the second image and a shift amount calculated from the movement cost using the approximate image, and outputs a movement amount signal corresponding to the determined movement amount.

3. The medical image processing device according to claim 1, characterized in that: The medical image processing apparatus further includes an image converter that converts an image acquired by imaging the inside of the patient's body based on an irradiation path of a treatment beam irradiated to the patient, thereby acquiring the first image and the second image as integral images.

4. The medical image processing apparatus according to claim 1, characterized in that: The cost is a value corresponding to the area or volume of the non-overlapping portion of the two or more regions.

5. The medical image processing apparatus according to claim 4, characterized in that: The larger the area or volume of the non-overlapping portion is, the higher the cost calculated by the cost calculation unit is.

6. The medical image processing apparatus according to claim 1, characterized in that: The cost is the shortest distance between the two or more areas.

7. The medical image processing apparatus according to claim 1, characterized in that: The region includes a first region corresponding to the first image and a second region corresponding to the second image, The region acquisition section acquires the second region by deforming the first region.

8. The medical image processing apparatus according to claim 1, characterized in that: The region includes an irradiation field of a treatment beam irradiating the patient.

9. The medical image processing apparatus according to claim 1, characterized in that: The region includes a tumor within the patient's body.

10. A treatment system, characterized in that: have: The medical image processing device according to any one of claims 1 to 9; and A treatment device includes an irradiation unit for irradiating the patient with radiation, an imaging device for imaging the first image and the second image, a bed for mounting and fixing the patient, and a bed control unit for controlling movement of the bed based on a movement amount signal.

11. A medical image processing method, characterized in that: The following processing is performed by the computer: acquiring a first image obtained by photographing the inside of a patient's body, acquiring a second image of the patient's body captured at a different time than the first image, acquiring two or more regions corresponding to either or both of the first image or the second image, calculating the similarity between the first image and the second image, Calculate the cost based on the location relationship of the areas, The relative position of the first image with respect to the second image is obtained so that the similarity is high and the cost is low.

12. A program, characterized in that Causes the computer to perform the following processing: acquiring a first image obtained by photographing the inside of a patient's body, acquiring a second image of the patient's body captured at a different time than the first image, acquiring two or more regions corresponding to either or both of the first image or the second image, calculating the similarity between the first image and the second image, Calculate the cost based on the location relationship of the areas, The relative position of the first image with respect to the second image is obtained so that the similarity is high and the cost is low.

13. A storage medium, characterized in that: A program is stored which causes the computer to execute the following processing: acquiring a first image obtained by photographing the inside of a patient's body, acquiring a second image of the patient's body captured at a different time than the first image, acquiring two or more regions corresponding to either or both of the first image or the second image, calculating the similarity between the first image and the second image, Calculate the cost based on the location relationship of the areas, The relative position of the first image with respect to the second image is obtained so that the similarity is high and the cost is low.

Images