Medical image generating device, treatment plan creating device, treatment system, medical image generating method, treatment plan creating method, medical image generating program, and treatment plan creating program

Abstract

To provide a medical image generation device, a medical image generation method, and a medical image generation program for reducing a burden of measurement on an observation individual (a patient) and simply generating accurate four-dimensional image data concerning the observation individual, and to provide a treatment planning preparation device, a treatment system, a treatment planning preparation method, and a treatment planning preparation program.SOLUTION: The medical image generation device and the method and program related to the device are provided to acquire image data concerning an observation individual, on the other hand, learn data concerning a four-dimensional image concerning a contrast individual other than the observation individual, deform the acquired image data on the basis of the learned data, and estimate and output the deformed image data as four-dimensional image data concerning the observation individual. There are also provided the treatment planning preparation device, the treatment system, the treatment planning preparation method, and the treatment planning preparation program using the medical image generation device.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a medical image generating device, a medical image generating method, and a medical image generating program for generating images used in treatment or diagnosis, and also to a treatment planning device, a treatment system, a treatment planning method, and a treatment planning program that use images generated by the medical image generating device. [Background technology]

[0002] In general, images of the affected area are acquired and utilized in the treatment or diagnosis of patients. In particular, in radiation therapy, in which radiation is irradiated to the affected area, it is necessary to accurately determine the position of the affected area and then irradiate the radiation. For this reason, images of the patient's affected area are acquired in advance as 3D data, and treatment plans and treatment of the affected area are created based on these images.

[0003] On the other hand, when the affected area of ​​a patient is located in a place where organ movement occurs due to breathing or heartbeat (such as the chest or abdomen), it is difficult to grasp the exact location of the affected area using normal 3D images. For this reason, image acquisition is carried out taking into account the movement of the affected area due to the patient's breathing.

[0004] For example, Patent Document 1 describes a treatment planning device and method that uses four-dimensional computed tomography images (hereinafter referred to as "four-dimensional CT images") consisting of a collection of three-dimensional images taken for each patient's respiratory cycle, cardiac cycle, and arterial pulse.

[0005] Furthermore, Patent Document 2 describes a method for estimating a patient's condition in real time during radiation therapy (claim 1, paragraphs 0018, 0023-0026, 0069-0075, 0085, and 0099-0105).

[0006] Furthermore, Non-Patent Document 1 describes the use of a multiscale deformable image registration (DIR) framework (MJ-CNN) that employs unsupervised joint training of convolutional neural networks to achieve accurate and rapid non-rigid image registration (DIR) for lung CT scans. This MJ-CNN uses 4D CT scans of 22 lung cancer patients as training data, and a pair of image data (training data) representing the inspiratory and expiratory phases of the 4D CT scans as input. The network then outputs a predicted DVF (Φ3) from the pair of image data and generates an image (Φ3·S) by deforming the input data (S) using the predicted DVF (Φ3) (see Abstract, page 1; Column 2.2, page 3; Figure 1, page 4; and Columns 2.3.1 and 2.3.2, page 5).

[0007] Non-Patent Document 2 also describes a supervised non-rigid image registration method based on a 3D convolutional neural network, along with a framework for training the network. It also describes that this methodology was verified on seven pairs of inspiration-expiration lung CT images, and that two full images are used as input to the network, with the output being a displacement vector field related to the full image domain. It also describes that this training method generates a training set using artificial transformations applied to a small set of representative images, so that expert-annotated data is not required to train the network (Abstract, page 1; left column, lines 10-30, column II, paragraph 1; column III.B., page 6; column VI, paragraphs 1 and 5). [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Japanese Patent Application Laid-Open No. 2008-80131 [Patent Document 2] International Publication No. 2020 / 086976 [Non-patent literature]

[0009] [Non-Patent Document 1] Zhuoran JIANG et al.,Phys.Med.Biol.,65,015011,(13 January 2020) [Non-patent document 2] EPPENHOF,KAJ,PLUIM,JPW,IEEE Trans.Med.Imaging, 38,1097-1105,Article8510836,(2019) Summary of the Invention [Problem to be solved by the invention]

[0010] As described in Patent Document 1, it is known to acquire a four-dimensional CT image of a patient and use the image to create a treatment plan.

[0011] However, acquiring the 4D CT image described in Patent Document 1 requires a longer imaging time than acquiring a normal 3D CT image, and in particular when imaging using X-rays, the patient is exposed to a large amount of radiation, resulting in a significant burden on the patient. Another technical problem occurs when the patient's breathing becomes irregular, significantly reducing the accuracy of the 4D CT image.

[0012] Furthermore, although measuring devices for acquiring four-dimensional image data of patients are already known, it is difficult to create an environment in which four-dimensional image data of patients can be acquired at all times due to factors such as the cost of the device itself and the large amount of data that must be handled.

[0013] On the other hand, 4D image data is extremely useful in the treatment and diagnosis of patients. Therefore, to solve the above-mentioned problems, there is a need to reduce the burden on patients involved in measurements and easily obtain highly accurate 4D image data.

[0014] Therefore, the object of the present invention is to provide a medical image generation device, a medical image generation method, and a medical image generation program that reduce the burden of measurement on the observed individual (patient) and easily generate highly accurate four-dimensional image data of a specific area of ​​the observed individual.

[0015] Another object of the present invention is to provide a treatment plan creation device, treatment system, treatment plan creation method, treatment method, and treatment plan creation program that use images created by the above-mentioned medical image generation device to reduce the burden on the observed individual and enable the creation of highly accurate treatment plans or treatment of the treatment target area. [Means for solving the problem]

[0016] As a result of intensive research into the above-mentioned problems, the inventors have discovered that by using four-dimensional image data relating to a specific region of a control individual other than the observed individual and transforming two-dimensional or three-dimensional image data relating to a specific region of the observed individual, it is possible to reduce the burden on the observed individual and easily obtain highly accurate four-dimensional image data, and have completed the present invention.

[0017] That is, the present invention provides the following medical image generating apparatus, treatment planning apparatus, treatment system, medical image generating method, treatment planning method, treatment method, medical image generating program, and treatment planning program.

[0018] The medical image generating device of the present invention, which solves the above-mentioned problems, is a medical image generating device that generates four-dimensional image data relating to a specific region of an observed individual, and is characterized by comprising: an image acquisition unit that acquires two-dimensional or three-dimensional image data relating to the specific region of the observed individual; a learning unit that learns multiple data relating to four-dimensional images relating to a specific region of a control individual different from the observed individual; and an image transformation processing unit that transforms the image data acquired by the image acquisition unit based on the data from the learning unit, and estimates and outputs four-dimensional image data relating to the specific region of the observed individual.

[0019] According to this feature, it is possible to estimate 4D image data of a specific region of an observed individual by acquiring 2D or 3D image data of the specific region of the observed individual while performing image deformation using 4D image data of the specific region of a control individual that has already been measured. This reduces the burden of measurement on the observed individual itself. Furthermore, by performing image deformation using the measurement results of the control individual, it is possible to easily generate highly accurate 4D image data of the specific region of the observed individual without being affected by factors such as respiratory disturbances of the observed individual.

[0020] In addition, one embodiment of the medical image generating apparatus of the present invention is characterized in that the image acquiring section acquires three-dimensional image data.

[0021] This feature reduces the computational load during image deformation compared to when using two-dimensional image data, making it easier to generate four-dimensional image data.

[0022] Furthermore, one embodiment of the medical image generating device of the present invention is characterized in that it includes a deformation amount calculation unit that sets one two-dimensional or three-dimensional image data as a reference phase for a four-dimensional image relating to a specific region of a control individual, and calculates the deformation amount of the two-dimensional or three-dimensional image data in other phases relative to the reference phase, and the learning unit learns the deformation amount calculated by the deformation amount calculation unit and the image data of the reference phase.

[0023] This feature makes it possible to reduce the amount of data required compared to learning from images themselves, thereby reducing the computational load.

[0024] In addition, one embodiment of the medical image generating device of the present invention is characterized in that the learning unit creates a model that derives a group of deformation amounts for each phase from the deformation amount calculated by the deformation amount calculation unit and image data of the reference phase, and the image deformation processing unit uses the model to calculate image data for each phase in the image data acquired by the image acquisition unit, and outputs it as four-dimensional image data relating to a specific region of the observed individual.

[0025] According to this feature, by calculating the amount of deformation in reference phase image data and other respiratory phases for a specific region of a control individual and creating a model based on these, it is possible to reduce the amount of data required for calculation and the computational load. Furthermore, since it is possible to create a model that takes data continuity into consideration, when deforming 3D image data for a specific region of an observed individual, it is possible to output highly accurate 4D image data that ensures the structural continuity of the observed individual.

[0026] Furthermore, one embodiment of the medical image generating device of the present invention is characterized in that it sets one two-dimensional or three-dimensional image data as a reference phase for a four-dimensional image relating to a specific region of a control individual, and is equipped with a comparison calculation unit that compares the image data acquired by the image acquisition unit with image data of the reference phase, and the image transformation processing unit outputs data relating to the four-dimensional image relating to the specific region of the control individual, including image data of the reference phase that is highly similar to the image data acquired by the image acquisition unit, as four-dimensional image data relating to the specific region of the observed individual.

[0027] According to this feature, the image data of the specific region of the control individual can be directly compared with the image data of the specific region of the observed individual, and based on the comparison result, the 4D image data of the specific region of the control individual can be output as the 4D image data of the specific region of the observed individual, which makes it easy to construct an arithmetic formula (program) for image transformation processing.

[0028] The treatment planning device of the present invention, which solves the above problem, is characterized by comprising the above medical image generation device and a treatment planning unit that uses four-dimensional image data relating to a specific region of the observed individual generated by the medical image generation device to create a treatment plan for the observed individual.

[0029] According to this feature, by using the images generated by the medical image generating device, it is possible to reduce the burden of measuring the observed individual and to create a treatment plan based on highly accurate 4D image data, thereby improving the accuracy of the treatment plan for the observed individual.

[0030] The treatment system of the present invention, which solves the above-mentioned problems, is characterized by comprising the above-mentioned medical image generating device and a treatment unit that performs treatment on the treatment target area of ​​the observed individual using four-dimensional image data related to a specific area of ​​the observed individual generated by the medical image generating device.

[0031] According to this feature, by using the images generated by the medical image generating device, it is possible to reduce the burden of measuring the observed individual and to perform treatment on the treatment target area of ​​the observed individual based on highly accurate 4D image data, thereby reducing the overall treatment burden on the observed individual and increasing the accuracy of treatment on the treatment target area.

[0032] The medical image generation method of the present invention for solving the above problem is a medical image generation method for generating four-dimensional image data relating to a specific region of an observed individual, and is characterized by comprising an image acquisition step for acquiring two-dimensional or three-dimensional image data relating to the specific region of the observed individual, a learning step for learning multiple pieces of data relating to four-dimensional images relating to a specific region of a control individual different from the observed individual, and an image transformation processing step for transforming the image data acquired by the image acquisition means based on the data learned in the learning step, and estimating and outputting four-dimensional image data relating to the specific region of the observed individual.

[0033] According to this feature, while acquiring two-dimensional or three-dimensional image data relating to a specific region of the observed individual, it is possible to estimate four-dimensional image data relating to the specific region of the observed individual by performing image deformation based on data relating to a four-dimensional image relating to the specific region of a control individual that has already been measured. This reduces the burden of measurement on the observed individual itself. Furthermore, by performing image deformation using the results of measurements on the control individual, it is possible to easily generate highly accurate four-dimensional image data relating to the specific region of the observed individual without being affected by factors such as respiratory disturbances of the observed individual.

[0034] In one embodiment of the medical image generating method of the present invention, the image acquiring step is characterized in that three-dimensional image data is acquired.

[0035] This feature reduces the computational load during image deformation compared to when using two-dimensional image data, making it easier to generate four-dimensional image data.

[0036] Furthermore, one embodiment of the medical image generation method of the present invention includes a deformation amount calculation step in which one two-dimensional or three-dimensional image data is set as a reference phase for a four-dimensional image relating to a specific region of a control individual, and the deformation amount of the two-dimensional or three-dimensional image data in other phases relative to the reference phase is calculated, and the learning step is characterized by learning the deformation amount calculated in the deformation amount calculation step and the image data of the reference phase.

[0037] This feature makes it possible to reduce the amount of data required compared to learning from images themselves, thereby reducing the computational load.

[0038] Furthermore, in one embodiment of the medical image generation method of the present invention, the learning step creates a model that derives a group of deformation amounts for each phase from the deformation amounts calculated in the deformation amount calculation step and image data of the reference phase, and the image deformation processing step uses the model to calculate image data for each phase in the image data acquired in the image acquisition step, and outputs it as four-dimensional image data relating to a specific region of the observed individual.

[0039] According to this feature, by calculating the amount of deformation in reference phase image data and other respiratory phases for a specific region of a control individual and creating a model based on these, it is possible to reduce the amount of data required for calculation and the computational load. Furthermore, since it is possible to create a model that takes data continuity into consideration, when deforming 3D image data for a specific region of an observed individual, it is possible to output highly accurate 4D image data that ensures the structural continuity of the observed individual.

[0040] Furthermore, one embodiment of the medical image generation method of the present invention includes a comparison calculation step in which one two-dimensional or three-dimensional image data serving as a reference phase is set for a four-dimensional image relating to a specific region of a control individual, and the image data acquired in the image acquisition step is compared with image data of the reference phase, and the image transformation processing step is characterized in that data relating to the four-dimensional image relating to the specific region of the control individual, including image data of the reference phase that is highly similar to the image data acquired in the image acquisition step, is output as four-dimensional image data relating to the specific region of the observed individual.

[0041] According to this feature, the image data of the specific region of the control individual can be directly compared with the image data of the specific region of the observed individual, and based on the comparison result, the 4D image data of the specific region of the control individual can be output as the 4D image data of the specific region of the observed individual, which makes it easy to construct an arithmetic formula (program) for image transformation processing.

[0042] The treatment planning method of the present invention for solving the above problem is characterized by comprising a medical image acquisition step of acquiring four-dimensional image data relating to a specific region of the observed individual using the above medical image generation method, and a treatment planning step of using the four-dimensional image data acquired in the medical image acquisition step to create a treatment plan for the observed individual.

[0043] According to this feature, by using the images generated by the above medical image generation method, it is possible to reduce the burden of measuring the observed individual and to create a treatment plan based on highly accurate four-dimensional image data, thereby improving the accuracy of the treatment plan for the observed individual.

[0044] The treatment method of the present invention for solving the above problem is characterized by comprising a medical image acquisition step of acquiring four-dimensional image data relating to a specific region of an observed individual using the above medical image generation method, and a treatment step of performing treatment on the treatment target area of ​​the observed individual using the four-dimensional image data acquired in the medical image acquisition step.

[0045] According to this feature, by using the images generated by the above medical image generation method, it is possible to reduce the burden of measuring the observed individual and to perform treatment on the treatment target area of ​​the observed individual based on highly accurate 4D image data, thereby reducing the overall treatment burden on the observed individual and increasing the accuracy of treatment on the treatment target area.

[0046] A medical image generating program according to the present invention for solving the above problem is characterized by executing the above medical image generating method.

[0047] This feature allows for rapid acquisition and calculation of large amounts of data, making it possible to smoothly carry out a series of processes related to medical image generation.

[0048] A treatment planning program according to the present invention for solving the above problems is characterized by executing the above treatment planning method.

[0049] This feature allows for rapid acquisition and calculation of large amounts of data, making it possible to smoothly carry out a series of processes related to the creation of a treatment plan. [Effects of the Invention]

[0050] According to the present invention, it is possible to provide a medical image generation device, a medical image generation method, and a medical image generation program that reduce the burden of measurement on the observed individual (patient) and easily generate highly accurate four-dimensional image data of a specific area of ​​the observed individual.

[0051] Furthermore, according to the present invention, it is possible to provide a treatment plan creation device, a treatment system, a treatment plan creation method, a treatment method, and a treatment plan creation program that use images created by the above-mentioned medical image generation device to reduce the burden on the observed individual and enable the creation of a highly accurate treatment plan or treatment of the treatment target area. [Brief explanation of the drawings]

[0052] [Figure 1] 1 is a schematic explanatory diagram showing the structure of a medical image generating apparatus according to a first embodiment of the present invention. [Figure 2] 4 is a flowchart of deformation amount calculation in the deformation amount calculation unit of the medical image generating apparatus according to the first embodiment of the present invention. FIG. [Figure 3] FIG. 2 is a diagram showing an example of a neural network constructed in a learning unit of the medical image generating apparatus according to the first embodiment of the present invention. [Figure 4] FIG. 2 is a diagram showing an example of calculation in an image transformation processing unit of the medical image generating apparatus according to the first embodiment of the present invention. [Figure 5] FIG. 2 is a diagram showing an example of image data output from an image transformation processing unit of the medical image generating apparatus according to the first embodiment of the present invention. [Figure 6] FIG. 1 is a flow diagram of a medical image generation process using a medical image generation apparatus according to a first embodiment of the present invention. [Figure 7]1 is a schematic explanatory diagram showing the structure of a treatment planning device according to a first embodiment of the present invention. [Figure 8] FIG. 1 is a schematic explanatory diagram showing the structure of a treatment system according to a first embodiment of the present invention. [Figure 9] FIG. 10 is a schematic explanatory diagram showing the structure of a medical image generating apparatus according to a second embodiment of the present invention. [Figure 10] FIG. 10 is a flowchart of calculations in a comparison calculation unit of a medical image generating apparatus according to a second embodiment of the present invention. [Figure 11] FIG. 10 is a flow diagram of a medical image generation process using a medical image generation apparatus according to a second embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0053] Hereinafter, embodiments of the medical image generating device, treatment planning device, treatment system, medical image generating program, and treatment planning program of the present invention will be described in detail. Furthermore, the medical image generating method, treatment planning method, and treatment method of the present invention will be substituted for the description of the configurations and operations of the medical image generating device, treatment planning device, and treatment system.

[0054] It should be noted that the medical image generating device, treatment plan creating device, treatment system, medical image generating method, treatment plan creating method, treatment method, medical image generating program, and treatment plan creating program described in the embodiments are merely examples used to explain the present invention, and are not limited to these.

[0055] In this specification, the term "observed individual" refers to an object to which the present invention is applied and which requires 4D image data for treatment or diagnosis, particularly a human or non-human animal. Specific examples of non-human animals include non-human primates, cows, pigs, sheep, goats, rabbits, dogs, cats, guinea pigs, hamsters, mice, rats, and chickens. That is, the device, system, program, and method of the present invention can be applied to humans. They can also be applied to livestock and pets, excluding humans.

[0056] Furthermore, in this specification, a "control individual" refers to an entity that is separate from the observed individual but of the same species as the observed individual. For example, if the observed individual is human, the control individual is also human. Furthermore, there is no particular relationship or association between the control individual and the observed individual. For example, there may be many common or similar elements (age, physique, case, etc.) between the control individual and the observed individual, or there may be almost no common or similar elements.

[0057] In this specification, the term "specific region" refers to a location where four-dimensional image data needs to be acquired for treatment or diagnosis, and particularly refers to a location where organ movement occurs due to breathing or heartbeat. Specific examples include the chest and abdomen. It is preferable to set the "specific region" as a location that includes an affected area (such as a tumor) that will be the treatment target, but this is not limitative. It is also possible to set a location where it is determined that image data acquisition is necessary for diagnosis, regardless of whether or not there is an affected area.

[0058] [First embodiment] (Medical image generation device) FIG. 1 is a schematic explanatory diagram showing the structure of a medical image generating apparatus according to a first embodiment of the present invention.

[0059] A medical image generating apparatus 1A according to this embodiment generates four-dimensional image data relating to a specific region of an observed individual. As shown in FIG. 1, the medical image generating apparatus 1A includes an image acquisition unit 2, a learning unit 3, and an image transformation processing unit 4. The medical image generating apparatus 1A also includes a transformation amount calculation unit 5 as an example of the calculation means in the image transformation processing unit 4. In FIG. 1, the arrows indicated by dashed dotted lines indicate connections that allow control or input.

[0060] The image acquisition unit 2 acquires two-dimensional or three-dimensional image data relating to a specific region of the observed individual. The image data acquired by the image acquisition unit 2 is input to the image transformation processing unit 4 as basic data relating to the specific region of the observed individual.

[0061] The image acquisition unit 2 is not particularly limited as long as it can acquire two-dimensional or three-dimensional image data for a specific region of the observed individual. Here, examples of image data acquired by the image acquisition unit 2 include X-ray fluoroscopic images (X-ray CT images), MRI images, and ultrasound images. Therefore, examples of the image acquisition unit 2 include X-ray fluoroscopic devices, X-ray CT devices, MRI devices, and ultrasound diagnostic imaging devices as long as it can acquire the above image data.

[0062] It is more preferable that the image acquisition unit 2 be capable of acquiring three-dimensional image data relating to a specific region of the observed individual, thereby reducing the computational load on the image transformation processing unit 4, which will be described later.

[0063] Here, it is preferable that the image data acquired by the image acquisition unit 2 be acquired in a state where it is not affected by organ movement due to breathing of the observed individual. More specifically, the image data is acquired by the image acquisition unit 2 in a state where the observed individual holds its breath. This makes it possible to improve the image quality of the image data acquired by the image acquisition unit 2.

[0064] The image acquisition unit 2 only acquires two-dimensional or three-dimensional image data relating to a specific region of the observed individual. Therefore, it is possible to complete measurements of the observed individual in a shorter time than with conventional four-dimensional image data acquisition. This reduces the burden on the observed individual.

[0065] The learning unit 3 learns a plurality of data relating to a four-dimensional image of a specific region of a control individual different from the observed individual. Furthermore, based on the data learned by the learning unit 3, calculations are performed in the image transformation processing unit 4, which will be described later.

[0066] The data learned by the learning unit 3 relates to the measurement results for a specific region of the control individual, and is data acquired prior to the image data of the observed individual acquired by the image acquisition unit 2.

[0067] The data learned by the learning unit 3 may be obtained by measuring multiple control individuals at a facility equipped with an environment for acquiring 4D image data. In other words, the learning unit 3 aggregates and compiles 4D image data that has been measured on various control individuals at various facilities. Therefore, the learning unit 3 may be capable of collecting 4D image data, and the means for acquiring the data learned by the learning unit 3 may be installed in a remote location separate from the medical image generation device 1A. This eliminates the need for the medical image generation device 1A itself to have a function for acquiring data related to 4D images, thereby reducing the cost of the device.

[0068] Furthermore, the data learned by the learning unit 3 may be data relating to a four-dimensional image of a specific region of the control individual, and the type of data is not particularly limited, but it is preferable that the data be a combination of the type of data acquired by the image acquisition unit 2 and other aspects other than the dimensions. For example, when the image acquisition unit 2 acquires three-dimensional X-ray fluoroscopic images (X-ray CT images), the data learned by the learning unit 3 may be four-dimensional X-ray fluoroscopic images (X-ray CT images). When the image acquisition unit 2 acquires three-dimensional MRI images, the data learned by the learning unit 3 may be four-dimensional MRI images.

[0069] The image transformation processing unit 4 transforms the image data acquired by the image acquisition unit 2 based on the data from the learning unit 3, and estimates and outputs four-dimensional image data relating to a specific region of the observed individual. In other words, the image transformation processing unit 4 receives the two-dimensional or three-dimensional image data acquired by the image acquisition unit 2 as input and performs a calculation to output it as four-dimensional image data.

[0070] The calculation means performed by the image transformation processing unit 4 is not particularly limited as long as it performs calculations based on the data from the learning unit 3. For example, it may be possible to grasp a trend in image changes over time from image data relating to a specific region of a control individual and transform the image data acquired by the image acquisition unit 2 based on this trend, or to directly compare the image data acquired by the image acquisition unit 2 with the image data relating to a specific region of a control individual and transform the image data acquired by the image acquisition unit 2 based on the comparison result.

[0071] In the following, in this embodiment, as an example of a calculation means performed by the image transformation processing unit 4, we will explain one that is equipped with a deformation amount calculation unit 5 and grasps trends in image changes over time from image data related to a specific region of a control individual.

[0072] The deformation amount calculation unit 5 sets one 2D or 3D image data as a reference phase for a 4D image relating to a specific region of a control individual, and calculates the deformation amount of the 2D or 3D image data in other phases relative to the reference phase. Note that, although the following description will be given using X-ray CT images as the type of image data, the invention is not limited to this.

[0073] FIG. 2 shows a flow of calculation of the amount of deformation in the amount of deformation calculation unit 5.

[0074] First, image data (3D image data in Figure 2) that will serve as the reference phase is set from the 4D CT images of the specific region of the control individual learned by the learning unit 3. The 4D CT image is a collection of 3D image data for one respiratory cycle, and the reference phase is set by dividing one respiratory cycle into equal parts at any interval. For example, one respiratory cycle is divided into 10 equal parts, with the beginning and end of the cycle (inhalation phase) set as T00 and T90, and the midpoint of the cycle (exhalation phase) set as T50, and the 4D CT image is divided by respiratory phase (@Tn). One of these parts is then set as the reference phase (@Tref).

[0075] The reference phase (@Tref) set at this time is preferably matched with the respiratory phase of image data relating to a specific region of the observed individual acquired by the image acquisition unit 2. For example, the reference phase (@Tref) may be selected to be a point (@T50) corresponding to the expiratory phase in the respiratory cycle, and the image acquisition unit 2 may acquire an image of the observed individual in a state where it has completely exhaled (expiratory phase).

[0076] Next, the deformation amount in the image data of other respiratory phases (4DCT@T00 to 4DCT@T90 (excluding 4DCT@T50)) is calculated from the set reference phase image data (4DCT@Tref (=4DCT@T50)) among the 4D image data of a specific region of the control individual.

[0077] The amount of deformation can be calculated using the deformable image registration (DIR) method. DIR is a type of image registration that aligns two images, and is a processing method that nonlinearly deforms one image to match the other.

[0078] An example of processing related to calculation of the amount of deformation using DIR will be described.

[0079] In two images (image A and image B), the movement vectors of the feature points on image B that correspond to the feature points on image A are calculated so that similar parts (feature points) between image A and image B best match.

[0080] This movement vector is defined as the deformation amount (hereinafter also referred to as "DVF (deformable vector file)") in this embodiment. For example, the deformation amount of 4DCT@T00 with respect to 4DCT@Tref is defined as DVF@T00, and similarly the deformation amounts of 4DCT@T10 to 4DCT@T90 are defined as DVF@T10 to DVF@T90.

[0081] The algorithm for calculating the amount of deformation by DIR is not particularly limited. For example, the algorithm shown in the "2018 Guidelines for the Use of Non-rigid Image Registration in Radiation Therapy (abbreviated as DIR Guidelines 2018)," which is an official guideline of the Japanese Society for Radiation Oncology (JASTRO), can be used.

[0082] Then, the deformation amount (DVF@T00 to T90) calculated by the deformation amount calculation unit 5 is combined with the reference phase image data (4DCT@Tref) and trained by the training unit 3. This allows the data trained by the training unit 3 in this embodiment to be smaller in size than when the image data itself is input. This reduces the load on data storage and calculation.

[0083] Here, it is preferable that the learning unit 3 uses a combination of reference phase image data (4DCT@Tref) relating to a specific region of a control individual and deformation amounts at other phases (DVF@T00 to T90) as training data to create a model that derives a group of deformation amounts for each respiratory phase.

[0084] The calculation means for creating the model is not particularly limited. -La One example is to construct a neural network and optimize the network parameters so that the relationship between the input data and the output data is closest to the relationship between the reference phase image data (4DCT@Tref) and the deformation amount (DVF@T00 to T90) that is the training data.

[0085] Figure 3 shows the network structure constructed in the learning unit 3. -La An example of a network is shown.

[0086] As shown in Figure 3, an autoencoder uses one encoder and multiple decoders to optimize the network parameters that connect the relationship between input data and output data. Methods such as stochastic gradient descent can be used for this optimization.

[0087] Based on the calculation results, it is possible to create a model that can derive a group of deformation amounts for each respiratory phase. In addition, calculations can be performed by associating all deformation amounts (DVF@T00-T90) with one reference phase image data (4DCT@Tref). This makes it possible to create a model that takes into account the relationship (continuity) between respiratory phases.

[0088] FIG. 4 shows an example of the calculation performed by the image transformation processing unit 4.

[0089] As shown in Fig. 4A, the image transformation processing unit 4 inputs image data (3DCT) relating to a specific region of the observed individual acquired by the image acquisition unit 2 into the model obtained by the learning unit 3, and calculates a transformation amount (DVF@T00-T90) corresponding to this image data (3DCT). Then, as shown in Fig. 4B, this transformation amount is used to transform the image data (3DCT) relating to the specific region of the observed individual. As a result, image data (4DCT@T00-T90) for each respiratory phase relating to the specific region of the observed individual is obtained.

[0090] FIG. 5 shows an example of image data output from the image transformation processing unit 4. As shown in FIG.

[0091] 5 shows image data corresponding to 4DCT@T00, T20, T50, T70, and T90 obtained by the image transformation processing unit 4. In this case, 4DCT@T50 in FIG. 5 corresponds to the image data (3DCT) input to the image transformation processing unit 4.

[0092] Then, a collection of image data such as that shown in FIG. 5 is output as four-dimensional image data relating to a specific region of the observed individual.

[0093] The relationship between respiratory phases is taken into consideration in the model obtained by the learning unit 3. Therefore, as shown in Fig. 5, the 4D image data output from the image transformation processing unit 4 is output as highly accurate 4D image data in which the image data for each respiratory phase (4DCT@T00 to T90) ensures continuity in the structure of the observed individual (such as the human skeleton and organ shapes).

[0094] The four-dimensional image data relating to the specific region of the observed individual output from the image transformation processing unit 4 is stored in a storage unit (not shown) and output to the outside as required.

[0095] The learning unit 3, image transformation processing unit 4, and transformation amount calculation unit 5 in this embodiment are realized by, for example, a hardware processor such as a CPU executing a program (software). Furthermore, some or all of these may be realized by hardware (including circuitry) such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a GPU (Graphics Processing Unit), or may be realized by a combination of software and hardware.

[0096] (Medical image generation process) 6 is a flow diagram of the medical image generation process in this embodiment. Note that the steps of this medical image generation process correspond to the medical image generation method in this embodiment.

[0097] As shown in Fig. 6, first, four-dimensional image data (4DCT) relating to a specific region of a control individual is collected and input to the deformation amount calculation unit 5. The deformation amount calculation unit 5 sets reference phase image data (4DCT@Tref) of the four-dimensional image data and calculates deformation amounts (DVF@T00 to T90) in other respiratory phases. This corresponds to the deformation amount calculation step.

[0098] Next, data relating to a combination of the reference phase image data (4DCT@Tref) and the deformation amounts (DVF@T00 to T90) is input from the deformation amount calculation unit 5 to the learning unit 3. This corresponds to the learning step. At this time, the number of data to be input to the learning unit 3 is not particularly limited, but the greater the number, the greater the generalizability and versatility of the model created by the learning unit 3.

[0099] The learning unit 3 uses the input data as training data to create a model for deriving a group of deformation amounts for each respiratory phase. -La Using machine learning techniques such as neural networks, a model that takes into account the continuity of the data is created by learning and calculating while maintaining the relationship between one input data (reference phase image data) and multiple output data (deformation amounts).

[0100] The model created by the learning unit 3 is then input to the image transformation processing unit 4.

[0101] Meanwhile, image data (3DCT) relating to a specific region of the observed individual is acquired by the image acquisition unit 2, and this image data (3DCT) is input to the image transformation processing unit 4. This corresponds to the image acquisition step.

[0102] The image transformation processing unit 4 calculates a transformation amount (DVF@T00-90) corresponding to the image data (3DCT) using the model input from the learning unit 3. Using this transformation amount, the image data (3DCT) relating to a specific region of the observed individual is transformed to obtain image data (4DCT@T00-T90) for each respiratory phase relating to the specific region of the observed individual.

[0103] The resulting set of image data is then output as four-dimensional image data relating to a specific region of the observed individual, which corresponds to the image transformation processing step.

[0104] Furthermore, the four-dimensional image data relating to the specific region of the observed individual output from the image transformation processing unit 4 is saved, and the medical image generation process ends.

[0105] It is preferable to create a necessary program for the series of operations related to the medical image generation process based on the flow shown in Figure 6 and execute it using a hardware processor such as a CPU. This makes it possible to quickly acquire and calculate large amounts of data, and to smoothly execute the series of processes related to medical image generation. In addition, this program should be stored in advance on a hard disk (HDD). The software may be stored in a storage device such as a hard disk drive (HDD) or flash memory, or may be stored on a removable storage medium such as a DVD or CD-ROM, and installed in the storage device when the storage medium is inserted into a drive device.

[0106] As described above, the medical image generating device 1A in this embodiment acquires two-dimensional or three-dimensional image data relating to a specific region of the observed individual, while performing image deformation using four-dimensional image data relating to a specific region of a control individual that has already been measured, thereby making it possible to estimate four-dimensional image data relating to the specific region of the observed individual. This reduces the burden of measurement on the observed individual itself. Furthermore, performing image deformation using the results of measurements on the control individual makes it possible to easily generate highly accurate four-dimensional image data relating to the specific region of the observed individual without being affected by factors such as respiratory disturbances of the observed individual.

[0107] In particular, in the medical image generating device 1A of this embodiment, by calculating the amount of deformation in reference phase image data and other respiratory phases for a specific region of a control individual and creating a model based on these, it becomes possible to reduce the amount of data required for calculation and reduce the calculation load. Furthermore, since it becomes possible to create a model that takes data continuity into consideration, when deforming 3D image data for a specific region of an observed individual, it becomes possible to output highly accurate 4D image data that ensures the structural continuity of the observed individual.

[0108] (treatment planning device) FIG. 7 is a schematic explanatory diagram showing the structure of a treatment planning device in the first embodiment of the present invention.

[0109] As shown in FIG. 7, a treatment planning device 10 in this embodiment includes the above-mentioned medical image generating device 1A and a treatment planning unit 11.

[0110] The treatment planning unit 11 uses four-dimensional image data relating to a specific region of the observed individual, generated by the medical image generating apparatus 1A, to generate a treatment plan for the observed individual.

[0111] The treatment planning unit 11 is not particularly limited as long as it can acquire 4D image data relating to a specific region of the observed individual from the medical image generation device 1A and create a treatment plan for the observed individual based on this 4D image data. For example, if the treatment for the observed individual is radiation therapy, the treatment planning unit 11 may include a target identification unit that identifies a treatment target (target) from the 4D image data generated by the medical image generation device 1A, a radiation irradiation simulation unit that simulates the energy, irradiation direction, and shape of the irradiation area of ​​the radiation to be irradiated based on the position and shape of the target, and a treatment plan evaluation unit based on the results of the radiation irradiation simulation unit, and may create a treatment plan using these units. The treatment plan created by the treatment planning unit 11 is stored in a storage unit (not shown) and output to an external device as needed.

[0112] The treatment planning unit 11 is realized by, for example, a hardware processor such as a CPU executing a program (software).

[0113] In the treatment planning device 10, it is preferable to create a necessary program and execute it on a hardware processor such as a CPU for a series of operations related to the acquisition of 4D image data (corresponding to an image acquisition step) of a specific region of the observed individual generated by the medical image generation device 1A and the creation of a treatment plan (corresponding to a treatment planning step) in the treatment planning unit 11. This makes it possible to quickly acquire and calculate large amounts of data, and to smoothly execute a series of processes related to the creation of a treatment plan. Furthermore, this program may be stored in advance in a storage device such as an HDD (Hard Disk Drive) or flash memory, or may be stored on a removable storage medium such as a DVD or CD-ROM and installed in the storage device when the storage medium is inserted into a drive device.

[0114] Furthermore, a series of operations in the treatment planning device 10 corresponds to the treatment planning method in this embodiment. For example, acquisition of four-dimensional image data relating to a specific region of the observed individual generated by the medical image generation device 1A corresponds to an image acquisition step, and creation of a treatment plan in the treatment planning unit 11 corresponds to a treatment planning step.

[0115] The treatment planning device 10 in this embodiment reduces the burden of measuring the observed individual by using images generated by the medical image generation device 1A, and can create a treatment plan based on highly accurate four-dimensional image data, thereby improving the accuracy of the treatment plan for the observed individual.

[0116] (Treatment System) FIG. 8 is a schematic explanatory diagram showing the structure of a treatment system according to the first embodiment of the present invention.

[0117] As shown in FIG. 8, a treatment system 20 in this embodiment includes the above-mentioned medical image generating apparatus 1A and a treatment unit .

[0118] The treatment unit 21 uses four-dimensional image data relating to a specific region of the observed individual, generated by the medical image generating apparatus 1A, to perform treatment on a treatment target site of the observed individual.

[0119] The treatment unit 21 is not particularly limited as long as it can acquire four-dimensional image data relating to a specific region of the observed individual from the medical image generating device 1A and perform treatment on the treatment target portion of the observed individual based on this four-dimensional image data. For example, if the treatment on the observed individual is radiation therapy, the treatment unit 21 may include a bed for fixing the observed individual, a radiation irradiation device for irradiating radiation, and an output control means for controlling the output of the radiation irradiation device, and perform treatment on the treatment target portion of the observed individual based on the four-dimensional image data generated by the medical image generating device 1A. Alternatively, the treatment unit 21 may be controlled based on the treatment plan by the treatment planning unit 11 described above, and the treatment may be performed on the treatment target portion of the observed individual.

[0120] Furthermore, a series of operations in the treatment system 20 corresponds to the treatment method in this embodiment. For example, acquisition of four-dimensional image data relating to a specific region of the observed individual generated by the medical image generating device 1A corresponds to an image acquisition step, and treatment of the treatment target site of the observed individual by the treatment unit 21 corresponds to a treatment step.

[0121] The treatment system 20 in this embodiment reduces the burden of measurements on the observed individual by using images generated by the medical image generating device 1A, and enables treatment of the treatment target area of ​​the observed individual based on highly accurate 4D image data. This reduces the overall treatment burden on the observed individual and also makes it possible to increase the accuracy of treatment of the treatment target area.

[0122] [Second embodiment] FIG. 9 is a schematic explanatory diagram showing the structure of a medical image generating apparatus according to the second embodiment of the present invention.

[0123] 9, the medical image generating apparatus 1B according to the second embodiment is provided with a comparison calculation unit 6 instead of the deformation amount calculation unit 5 in the medical image generating apparatus 1A according to the first embodiment. Note that a description of the same structures as those in the first embodiment will be omitted.

[0124] The comparison calculation unit 6 is an example of a calculation means in the image transformation processing unit 4, and sets one two-dimensional or three-dimensional image data that serves as a reference phase for a four-dimensional image relating to a specific region of the control individual, and compares the image data acquired by the image acquisition unit 2 with the image data of the reference phase.

[0125] FIG. 10 shows the calculation flow in the comparison calculation unit 6.

[0126] First, image data (3D image data in FIG. 10) that will serve as the reference phase is set from the 4D CT images of the specific region of the control individual learned by the learning unit 3. The 4D CT image is a collection of 3D image data for one respiratory cycle, and the reference phase is set by dividing one respiratory cycle into equal parts at any interval. For example, one respiratory cycle is divided into 10 equal parts, with the beginning and end of the cycle (inhalation phase) set as T00 and T90, and the midpoint of the cycle (exhalation phase) set as T50, and the 4D CT image is divided by respiratory phase (@Tn). One of these parts is then set as the reference phase (@Tref).

[0127] The reference phase (@Tref) set at this time is preferably matched with the respiratory phase of image data relating to a specific region of the observed individual acquired by the image acquisition unit 2. For example, the reference phase (@Tref) may be selected to be a point (@T50) corresponding to the expiratory phase in the respiratory cycle, and the image acquisition unit 2 may acquire an image of the observed individual in a state where it has completely exhaled (expiratory phase).

[0128] Next, the set reference phase image data (4DCT@Tref (=4DCT@T50)) is compared with three-dimensional image data (3DCT) of a specific region of the observed individual acquired by the image acquisition unit 2. As shown in FIG. 10, the three-dimensional image data (3DCT) of the specific region of the observed individual is compared in order with the reference phase image data (4DCT@Tref) of the 1st to nth control individuals, and the respective differences Δd are calculated.

[0129] The method for calculating the difference Δd is not particularly limited, and examples thereof include a method for quantifying the similarity between two images. More specifically, examples include a method for extracting representative points in two images and calculating the difference in their positions, or a method for calculating the difference in signal intensity of each pixel or each voxel between two images.

[0130] Next, the comparison result from the comparison operation unit 6 is input to the image transformation processing unit 4. Then, the image transformation processing unit 4 extracts the reference phase image data (4DCT@Tref) of the 1st to nth control individuals that has the smallest value of difference Δd. In this case, the number of reference phase image data (4DCT@Tref) of the control individuals to be extracted may be one, or the one with the mth smallest value of difference Δd may be extracted.

[0131] Then, four-dimensional image data (4DCT) relating to a specific region of the control individual, which includes the extracted reference phase image data (4DCT@Tref) of the control individual, is obtained from the data of the learning unit 3, and this is output as four-dimensional image data relating to a specific region of the observed individual.

[0132] In addition, when reference phase image data (4DCT@Tref) of multiple control individuals are extracted, the four-dimensional image data (4DCT) relating to specific regions of the corresponding control individuals may be averaged and output as the four-dimensional image data relating to the specific region of the observed individual.

[0133] The four-dimensional image data relating to the specific region of the observed individual output from the image transformation processing unit 4 is stored in a storage unit (not shown) and output to the outside as required.

[0134] The learning unit 3, image transformation processing unit 4, and comparison operation unit 6 in this embodiment are realized by, for example, a hardware processor such as a CPU executing a program (software). Furthermore, some or all of these may be realized by hardware (including circuitry) such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a GPU (Graphics Processing Unit), or may be realized by a combination of software and hardware.

[0135] (Medical image generation process) 11 is a flow diagram of the medical image generation process in this embodiment. Note that the steps of this medical image generation process correspond to the medical image generation method in this embodiment.

[0136] As shown in FIG. 11, first, four-dimensional image data (4DCT) relating to a specific region of a control individual is collected and input to the learning unit 3. The learning unit 3 sets reference phase image data (4DCT@Tref) of the four-dimensional image data. This corresponds to the learning step. Then, the set reference phase image data (4DCT@Tref) of the four-dimensional image data is input to the comparison calculation unit 6.

[0137] Meanwhile, image data (3DCT) relating to a specific region of the observed individual is acquired by the image acquisition unit 2, and this image data (3DCT) is input to the comparison calculation unit 6. This corresponds to the image acquisition step.

[0138] The comparison calculation unit 6 compares the image data (3DCT) acquired by the image acquisition unit 2 with the reference phase image data (4DCT@Tref) of the four-dimensional image data set by the learning unit 3 to find the difference Δd. This is performed for the data of the 1st to nth control individuals. This corresponds to the comparison calculation step.

[0139] In this case, the number of data points of the control individual is not particularly limited, but in order to reduce the calculation load, a predetermined number of control individuals having many common or similar elements to the observed individual may be selected in advance, and reference phase image data (4DCT@Tref) of the 4D image data relating to a specific region of the selected control individuals may be input to the comparison calculation unit 6.

[0140] Next, the reference phase image data (4DCT@Tref) of the 1st to nth control individuals is extracted with the smallest difference Δd.

[0141] Then, 4D image data (4DCT) of the specific region of the control individual, which includes the extracted reference phase image data (4DCT@Tref) of the control individual, is extracted from the data of the learning unit 3, and this is output as 4D image data of the specific region of the observed individual. This corresponds to the image deformation processing step.

[0142] Furthermore, the four-dimensional image data relating to the specific region of the observed individual output from the image transformation processing unit 4 is saved, and the medical image generation process ends.

[0143] It is preferable to create a necessary program for the series of operations related to the medical image generation process based on the flow shown in Figure 11 and execute it using a hardware processor such as a CPU. This makes it possible to quickly acquire and calculate large amounts of data, and to smoothly execute the series of processes related to medical image generation. In addition, this program may be stored in advance in a storage device such as an HDD (Hard Disk Drive) or flash memory, or may be stored in a removable storage medium such as a DVD or CD-ROM and installed in the storage device when the storage medium is inserted into a drive device.

[0144] As described above, the medical image generating device 1B of this embodiment, like the medical image generating device 1A shown in the first embodiment, can estimate 4D image data relating to a specific region of an observed individual by using 4D image data relating to a specific region of an already measured control individual and performing image deformation of the image data relating to a specific region of an observed individual. This reduces the measurement burden on the observed individual itself. Furthermore, by performing image deformation using the measurement results of the control individual, it becomes possible to easily generate highly accurate 4D image data relating to a specific region of an observed individual without being affected by breathing disturbances of the observed individual, etc.

[0145] In particular, in the medical image generating device 1B, in the image deformation of the image data relating to the specific region of the observed individual, the image data relating to the specific region of the control individual is directly compared with the image data relating to the specific region of the observed individual, and based on the comparison result, the 4D image data relating to the specific region of the control individual can be output as the 4D image data relating to the specific region of the observed individual as is, which makes it easy to construct an arithmetic formula (program) for the image deformation process.

[0146] Furthermore, the medical image generating device 1B in this embodiment may be used as a medical image generating device in the treatment planning device or treatment system shown in the first embodiment. This reduces the burden associated with measuring the observed individual, and makes it possible to create a treatment plan and treat the observed individual based on highly accurate four-dimensional image data. This also makes it possible to improve the accuracy of the treatment plan and treatment for the observed individual.

[0147] The above-described embodiments show examples of the medical image generating device, treatment planning device, treatment system, medical image generating method, treatment planning method, treatment method, medical image generating program, and treatment planning program. The medical image generating device, treatment planning device, treatment system, medical image generating method, treatment planning method, treatment method, medical image generating program, and treatment planning program according to the present invention are not limited to the above-described embodiments, and the medical image generating device, treatment planning device, treatment system, medical image generating method, treatment planning method, treatment method, medical image generating program, and treatment planning program according to the above-described embodiments may be modified within the scope of the gist of the claims. [Industrial Applicability]

[0148] The medical image generating device, treatment planning device, treatment system, medical image generating method, treatment planning method, treatment method, medical image generating program, and treatment planning program of the present invention are used for the treatment and diagnosis of an observed individual (patient). In particular, they are used to acquire four-dimensional image data relating to a specific region of an observed individual that is useful for treatment and diagnosis, and are suitably used for treatment and diagnosis utilizing this four-dimensional image data. [Explanation of symbols]

[0149] 1A, 1B Medical image generation device, 2 Image acquisition unit, 3 Learning unit, 4 Image transformation processing unit, 5 Deformation amount calculation unit, 6 comparison calculation unit, 10 treatment plan creation device, 11 treatment plan creation unit, 20 treatment system, 21 treatment unit

Claims

1. 1. A medical image generating apparatus for generating four-dimensional image data relating to a specific region of an observed individual, comprising: an image acquisition unit that acquires two-dimensional or three-dimensional image data relating to a specific region of the observed individual; a deformation amount calculation unit that sets one two-dimensional or three-dimensional image data as a reference phase for a four-dimensional image relating to a specific region of a control individual different from the observed individual, and calculates a deformation amount of the two-dimensional or three-dimensional image data in another phase relative to the reference phase; a learning unit that learns the deformation amounts in other phases relative to the one reference phase calculated by the deformation amount calculation unit as teacher data and creates a model that derives a group of deformation amounts of the other phases in consideration of a relationship (continuity) between the phases; an image transformation processing unit that uses the model created by the learning unit to derive a group of deformation amounts for the other phases corresponding to image data of the observed individual acquired by the image acquisition unit, and that uses the derived group of deformation amounts to transform the image data of the observed individual for each of the other phases, and outputs the result as four-dimensional image data relating to a specific region of the observed individual.

2. 2. The medical image generating device according to claim 1, wherein the learning unit learns the deformation amounts at other phases relative to the one reference phase calculated by the deformation amount calculation unit and the image data at the reference phase as training data.

3. 3. The medical image generating apparatus according to claim 1, wherein the image acquisition unit acquires three-dimensional image data.

4. A medical image generating device as described in any one of claims 1 to 3, wherein the reference phase and the other phase are a reference phase of a respiratory phase and another phase of a respiratory phase, respectively, and nine other phases of the respiratory phase are set.

5. A medical image generating device according to any one of claims 1 to 4; a treatment planning unit that uses four-dimensional image data relating to a specific region of the observed individual generated by the medical image generation device to generate a treatment plan for the observed individual.

6. A medical image generating device according to any one of claims 1 to 4; a treatment unit that performs treatment on a treatment target site of the observed individual using four-dimensional image data relating to a specific region of the observed individual generated by the medical image generation device.

7. 1. A medical image generation method for generating four-dimensional image data relating to a specific region of an observed individual, comprising: an image acquisition step of acquiring two-dimensional or three-dimensional image data relating to a specific region of the observed individual; a deformation amount calculation step of setting one two-dimensional or three-dimensional image data as a reference phase for a four-dimensional image relating to a specific region of a control individual other than the observed individual, and calculating a deformation amount of the two-dimensional or three-dimensional image data in another phase relative to the reference phase; a learning step of learning the deformation amounts in other phases relative to the one reference phase calculated in the deformation amount calculation step as teacher data and creating a model for deriving a group of deformation amounts of the other phases in consideration of the relationship (continuity) between the phases; an image deformation processing step of deriving a group of deformation amounts for the other phases corresponding to the image data of the observed individual acquired in the image acquisition step using the model created in the learning step, and deforming the image data of the observed individual for each of the other phases using the derived group of deformation amounts, and outputting the image data as four-dimensional image data relating to a specific region of the observed individual.

8. 8. The medical image generating method according to claim 7, wherein in the learning step, the deformation amount at another phase relative to the one reference phase calculated in the deformation amount calculation step and the image data at the reference phase are learned as teacher data.

9. 9. The medical image generating method according to claim 7, wherein the image acquiring step acquires three-dimensional image data.

10. A medical image generation method described in any one of claims 7 to 9, wherein the reference phase and the other phase are a reference phase of a respiratory phase and another phase of a respiratory phase, respectively, and nine other phases of the respiratory phase are set.

11. a medical image acquisition step of acquiring four-dimensional image data relating to a specific region of an observed individual by the medical image generating method according to any one of claims 7 to 10; a treatment planning step of using the four-dimensional image data acquired in the medical image acquisition step to create a treatment plan for the observed individual, said treatment planning method being executed by a computer.

12. A medical image generating program that causes a computer to execute the medical image generating method according to any one of claims 7 to 10.

13. A treatment planning program for causing a computer to execute the treatment planning method according to claim 11.

Images