Image processing device, image processing method, and program

Abstract

To improve accuracy of region extraction concerning a portion of a subject out of images obtained by imaging the portion.SOLUTION: The image processing device comprises: an image acquisition unit which acquires a plurality of images obtained by imaging a portion of a subject; a primary region acquisition unit which acquires information on a plurality of primary regions corresponding to the portion out of a predetermined number of images among the plurality of images; and a detail region acquisition unit which on the basis of the information on the plurality of primary regions, acquires information on a detail region corresponding to the portion of one of the plurality of images.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an image processing device, an image processing method, and a program. [Background technology]

[0002] In the medical field, there is a technology that assists in understanding the structure and condition of the heart by acquiring a cardiac chamber region from medical images obtained by imaging using various modalities, displaying the cardiac chamber region, and outputting measurement values ​​measured from the cardiac chamber region. Patent Document 1 discloses a technology that efficiently acquires the entire region of an object by acquiring an initial region contained in the object from a low-resolution image and then acquiring a detailed region including the contour of the object from a high-resolution image based on the initial region. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-004441 Summary of the Invention [Problem to be solved by the invention]

[0004] However, with the technology disclosed in Patent Document 1, if there is an error in the initial region to be acquired, there is a possibility that a detailed region of the object cannot be appropriately acquired from the image.

[0005] The technology of the present disclosure has been made in view of the above, and aims to improve the accuracy of extracting a region relating to a part of a subject from an image of the part. [Means for solving the problem]

[0006] An image processing device according to the present disclosure includes an image acquisition unit that acquires a plurality of images of a region of a subject, an initial region acquisition unit that acquires information on a plurality of initial regions corresponding to the region from a predetermined number of images of the plurality of images, and a detailed region acquisition unit that acquires information on a detailed region corresponding to the region in any of the plurality of images based on the information on the plurality of initial regions. The image processing device according to the present disclosure also includes an image acquisition unit that acquires a plurality of images of a region of a subject, a first information acquisition unit that acquires information on a plurality of first regions corresponding to the region from a predetermined number of images of the plurality of images, and a second information acquisition unit that acquires information on a second region that corresponds to the region in any of the plurality of images and that matches the shape of the region more closely than the first region, based on the information on the plurality of first regions.

[0007] An image processing method according to the present disclosure includes the steps of acquiring a plurality of images in which a region of a subject is imaged, acquiring information on a plurality of initial regions corresponding to the region from a predetermined number of images among the plurality of images, and acquiring information on a detailed region corresponding to the region in any of the plurality of images based on the information on the plurality of initial regions.An image processing method according to the present disclosure also includes the steps of acquiring a plurality of images in which a region of a subject is imaged, acquiring information on a plurality of first regions corresponding to the region from a predetermined number of images among the plurality of images, and acquiring information on a second region corresponding to the region in any of the plurality of images that matches the shape of the region more closely than the first regions based on the information on the plurality of first regions. [Effects of the Invention]

[0008] According to the technology of the present disclosure, it is possible to improve the accuracy of extracting a region relating to a part of a subject from an image of the part. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram showing the arrangement of an image processing system according to a first embodiment. [Figure 2] FIG. 2 is a diagram showing the functional configuration of a control unit of the image processing apparatus according to the first embodiment. [Figure 3] FIG. 2 is a flowchart showing an example of processing executed by the image processing apparatus according to the first embodiment. [Figure 4] FIG. 3 is a diagram showing areas acquired from an image in the first embodiment. [Figure 5] FIG. 4 is a diagram showing a display example of a detailed region in the first embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of an image processing device disclosed in this specification will be described with reference to the drawings. The same or equivalent components, members, and processes shown in each drawing will be assigned the same reference numerals, and duplicated descriptions will be omitted as appropriate. In addition, some of the components, members, and processes will be omitted as appropriate in each drawing.

[0011] First Embodiment An image processing system according to a first embodiment of the technique of the present disclosure will be described. Fig. 1 is a diagram showing the configuration of an image processing system 10 according to this embodiment. As shown in Fig. 1, the image processing system 10 includes an image processing device 100 and a data server 130. The image processing device 100 is connected to the data server 130 via a network 120 so as to be able to communicate with the data server 130.

[0012] In the image processing system 10 according to this embodiment, the image processing device 100 acquires multiple initial regions of the cardiac cavity from multiple 3D images of the heart, and acquires detailed regions of the cardiac cavity from a predetermined 3D image based on the multiple initial regions. Here, the cardiac cavity refers to the internal cavity of the heart, and includes the left atrium, left ventricle, right atrium, and right ventricle. The image processing device 100 then displays the acquired detailed regions of the cardiac cavity and provides the user with measurement values ​​measured from the detailed regions of the cardiac cavity. This allows the user to easily understand the structure and condition of the heart using the image processing device 100.

[0013] The configuration of the image processing system 10 will be described below with reference to Fig. 1. The image processing system 10 includes, as its functional components, an image processing device 100, a network 120, and a data server 130. The network 120 includes, for example, a LAN (Local Area Network) or a WAN (Wide Area Network).

[0014] The data server 130 is a picture archiving and communication system (PACS) that holds and manages medical images and information associated with the medical images. The image processing device 100 can acquire medical images stored in the data server 130 via the network 120. The data server 130 stores images captured by medical imaging devices (modalities) and transmits the images to each device in response to a request from the device connected to the network 120. The data server 130 also includes a database that stores received images as well as various data associated with the images.

[0015] In this embodiment, the three-dimensional image is taken by an X-ray CT (Computed Tomography) device. Although an image will be described as an example, other modalities may be used to capture the three-dimensional image. Modalities include, in addition to an X-ray CT device, for example, an MRI (Magnetic Resonance Imaging) device and a SPECT (Single Photon Emission Computed Tomography) device. Modalities also include a PET (Positron Emission Tomography) device, an ultrasound diagnostic device, an OCT (Optical Coherence Tomography) device, etc. In addition, in accordance with this embodiment, The image processing device 100 is applicable to three-dimensional images acquired by various modalities. In addition, in this embodiment, a case where the image processing device 100 processes three-dimensional images will be described, but the image processing device 100 is not limited to three-dimensional images, and may perform the following processing on two-dimensional images such as a long-axis image of a cardiac cavity to be processed.

[0016] The image processing device 100 is a device that performs image processing according to the embodiment. The image processing device 100 is a device that displays the area and measurement values ​​acquired from a three-dimensional image on a display unit 150, and functions as a terminal device for image interpretation operated by a user such as a doctor. The image processing device 100 includes a communication IF (Interface) 111 (communication unit), a ROM (Read Only Memory) 112, and a RA (Radio Control Unit). The device includes a random access memory (M) 113, a storage unit 114, and a control unit 115.

[0017] The communication IF 111 (communication unit) is configured with a LAN card or the like, and realizes communication between an external device (for example, the data server 130) and the image processing device 100. The ROM 112 is configured with a non-volatile memory or the like, and stores various programs. The RAM 113 is configured with a volatile memory or the like, and temporarily stores various information as data. The storage unit 114 is configured with a HDD (Hard Disk Drive) or the like, and temporarily stores various information as data. Remember.

[0018] The instruction unit 140 is configured with a GUI (Graphical User Interface) such as a keyboard, mouse, and touch panel, and inputs instructions from a user (e.g., a doctor) to the image processing device 100. An image to be processed in the image processing device 100 is input to the image processing device 100 in accordance with an instruction from a user who operates the instruction unit 140. Note that the selection of an image does not have to be based on an instruction from the user; for example, the control unit 115 of the image processing device 100 may automatically select an image to be processed in accordance with a predetermined rule.

[0019] 2 is a diagram showing the functional configuration of the control unit 115 of the image processing device 100. The control unit 115 is configured by a CPU (Central Processing Unit) and the like, and controls the image processing device 100. The control unit 115 has, as its functional configuration, an input image acquisition unit 210, an initial region acquisition unit 220, a seed information acquisition unit 230, a detailed region acquisition unit 240, and a display control unit 250.

[0020] The input image acquisition unit 210 acquires multiple 3D images from the data server 130 via the communication IF 111 (communication unit) and the network 120. The initial region acquisition unit 220 acquires an initial region corresponding to the cardiac cavity from each of a predetermined number of the acquired multiple 3D images. As a result, the initial region acquisition unit 220 acquires information about the multiple initial regions corresponding to the cardiac cavity from the multiple 3D images. The seed information acquisition unit 230 acquires seed information based on the acquired multiple initial regions. Here, the seed information is information about the rough location of the cardiac cavity region, and is information used to acquire a detailed region of the cardiac cavity, which will be described later. In this embodiment, the seed information is region information indicating a foreground seed and a background seed, which will be described later. The detailed region acquisition unit 240 acquires information about the detailed region corresponding to the cardiac cavity from one of the acquired multiple 3D images based on the seed information. In this embodiment, the initial region acquisition unit 220 and the seed information acquisition unit 230 are a first information acquisition unit that acquires information about the initial region, which is a first region. Furthermore, the detailed region acquisition unit 240 is a second information acquisition unit that acquires information about the detailed region, which is a second region, based on information about the initial region. The second region, which is a detailed region, has a higher degree of match with the shape of the "heart chamber" corresponding to the anatomical region of the extraction target than the first region, which is an initial region extracted initially. In other words, the second region, which is a detailed region, may have a higher probability of matching the shape of the "heart chamber" corresponding to the anatomical region of the extraction target than the first region, which is an initial region extracted initially.

[0021] The display control unit 250 generates an image to be displayed on the display unit 150 using the acquired detailed region. The display control unit 250 also controls the display of the generated image on the display unit 150. The display unit 150 is configured with any display device such as an LCD (Liquid Crystal Display) or a CRT (Cathode Ray Tube), and displays the image acquired by the control unit 115 and various information to the user. Specifically, the display unit 150 displays the input image acquired by the image processing device 100 and the detailed region of the cardiac cavity.

[0022] Each of the components of the image processing device 100 described above functions according to a computer program. For example, the control unit 115 (CPU) uses the RAM 113 as a work area to read and execute a computer program stored in the ROM 112 or the storage unit 114, thereby realizing the function of each component. Note that some or all of the functions of the components of the image processing device 100 may be realized using dedicated circuits. Also, some of the functions of the components of the control unit 115 may be realized using a cloud computer.

[0023] For example, an arithmetic device external to the image processing device 100 may be communicatively connected to the image processing device 100 via the network 120, and the image processing device 100 may transmit and receive data to and from the arithmetic device, thereby realizing at least some of the functions of the components of the control unit 115.

[0024] Next, an example of processing executed by the image processing device 100 in this embodiment will be described with reference to Fig. 3. Fig. 3 is a flowchart showing processing executed by the control unit 115 of the image processing device 100. In this embodiment, the image processing method of the present disclosure is realized by the control unit 115 executing the processing of each of the following steps. In addition, this embodiment assumes a case where the image processing device 100 processes cardiac CT images of a subject's heart captured by an X-ray CT device.

[0025] (S310: Acquire input image) In step S310, when the user instructs acquisition of multiple 3D images via operation of the instruction unit 140, the input image acquisition unit 210 acquires multiple cardiac CT images of the heart of the subject specified by the user from the data server 130. The input image acquisition unit 210 outputs the acquired multiple cardiac CT images to the initial region acquisition unit 220, the seed information acquisition unit 230, the detailed region acquisition unit 240, and the display control unit 250, respectively.

[0026] In this embodiment, cardiac CT images are images acquired from a video sequence (4D CT images) composed of images acquired (i.e., acquired over time) of different phases of the heart, whose shape changes periodically. The multiple cardiac CT images may be multiple cardiac CT images acquired on different dates and times of the same subject, or may be dynamic CT images acquired on the same subject at different contrast enhancement timings, or a combination of contrast-enhanced and non-contrast-enhanced CT images. The multiple acquired images are not limited to images of the heart of the same subject, but may be images of the hearts of different subjects. The multiple cardiac CT images are not limited to images acquired using the same modality, but may be images acquired using different modalities. The multiple cardiac CT images are not limited to images acquired over time of the same subject, but may be images of the subject acquired at the same time using different means.

[0027] The input image acquisition unit 210 can also function as an image selection unit that selects two or more images from the acquired multiple 3D images as targets for acquiring an initial region. For example, the input image acquisition unit 210 can select multiple images that have a higher degree of similarity in the morphology or phase of the left atrium based on morphological information or phase information of the left atrium in the acquired multiple 3D images. This similarity may also be expressed as compatibility, degree of coincidence in shape, etc. Furthermore, this similarity can be converted into an objective numerical value by using known statistical indices such as the Jaccard coefficient, the Dice coefficient, the Simpson coefficient, etc.

[0028] (S320: Acquire initial area) In step S320, the initial region acquisition unit 220 acquires an initial region corresponding to a cardiac cavity from a predetermined number of three-dimensional images, here at least two or more three-dimensional images, among the multiple cardiac CT images acquired in step S310. Then, the initial region acquisition unit 220 outputs information on the multiple initial regions acquired from the multiple cardiac CT images to the seed information acquisition unit 230. In this embodiment, the processing will be described using the left atrium as an example of a cardiac cavity, but it is also applicable to other cardiac cavities (left ventricle, right atrium, right ventricle).

[0029] In this embodiment, the initial region acquisition unit 220 acquires an initial region of the left atrium from each of the three-dimensional images of multiple cardiac phases included in the 4DCT image. Here, acquisition of the initial region from the three-dimensional image can be achieved using a known image analysis technique. For example, as an example of a technique using a machine learning model, the initial region acquisition unit 220 can estimate the region of the left atrium from the cardiac CT image using UNet, which is a type of deep learning, as a machine learning model, and use the estimated region as the initial region.

[0030] In this embodiment, the image of the initial region acquired by the initial region acquisition unit 220 is a mask image. Here, the mask image is a binary image in the same image coordinate system as the cardiac CT image, and is an image in which a foreground pixel value (e.g., 1) is stored in pixels indicating the initial region of the left atrium, and a background pixel value (e.g., 0) is stored in regions other than the initial region. The initial region acquisition unit 220 may also directly acquire the mask image, which is a binary image, using UNet. Alternatively, the initial region acquisition unit 220 may acquire an image in which a value representing the likelihood of the left atrium (the probability of the left atrium being present) is stored in the pixel value of each pixel using UNet, and then perform threshold processing or the like to acquire the mask image, which is a binary image. Alternatively, the initial region acquisition unit 220 may acquire an image in which a pixel value corresponding to the probability of the left atrium being present is stored in each pixel, rather than a binary mask image.

[0031] Furthermore, the acquisition of the initial region is not limited to the method using UNet, but can also be achieved by using any other machine learning model or known image region extraction technology. For example, the initial region acquisition unit 220 may acquire the initial region of the left atrium by applying a statistical shape model to the subject's cardiac CT image. Based on the results of aligning the subject's cardiac CT image with the statistical shape model representing the statistical trend of the left atrium region calculated from multiple cardiac CT images and the left atrium, the initial region acquisition unit 220 projects the left atrium region indicated by the statistical shape model onto the subject's image coordinate system. The initial region acquisition unit 220 then generates a mask image using the region obtained by the projection as the initial region. Alternatively, information regarding the subject's left atrium region may be stored in a data server in advance, and the initial region acquisition unit 220 may read and acquire the information from the data server and acquire the initial region based on the acquired information.

[0032] In this embodiment, initial regions are acquired from at least two or more of the multiple cardiac CT images acquired by the input image acquisition unit 210. Then, it is preferable that the seed information acquired in subsequent step S330 has similar left atrium morphologies among the multiple cardiac CT images from which initial regions are acquired. Therefore, for example, in the case of 4D CT images obtained by sequentially imaging a beating heart over time, cardiac CT images with similar left atrium morphologies can be obtained by acquiring initial regions from images of a specified frame and the frames before and after it.

[0033] Furthermore, when acquiring 4D CT images of the heart, the initial region can be acquired from each of multiple frame images with similar cardiac phases, even if the images are from different cardiac cycles. Furthermore, the multiple cardiac CT images from which the initial region is acquired do not necessarily have to be images with similar cardiac phases. For example, images from multiple cardiac phases with similar left atrium morphology (shape, volume, etc.), such as a cardiac phase midway between diastole and systole in the period from diastole to systole and a cardiac phase midway between diastole and systole in the period from systole to diastole, may be used.

[0034] The initial region acquisition unit 220 may also perform image analysis on the acquired multiple 3D images to automatically determine the image from which the initial region is to be acquired. For example, the initial region acquisition unit 220 may calculate the image similarity between the acquired image of a predetermined cardiac phase and images of other cardiac phases, and acquire the initial region only from images of cardiac phases with high similarity. This allows the initial region acquisition unit 220 to selectively acquire images with similar left atrium morphologies.

[0035] The initial region acquisition unit 220 may also determine the image from which the initial region is to be acquired depending on the image quality of each image. For example, the initial region acquisition unit 220 may calculate a feature value indicating the image sharpness for each image constituting the multiple cardiac CT images, and acquire the initial region only from images of cardiac phases with high sharpness. This allows the initial region acquisition unit 220 to acquire the initial region from images with high image quality and few artifacts such as motion artifacts. Note that the initial region acquisition unit 220 does not necessarily need to limit the images from which the initial region is to be acquired, and may acquire the initial region from images of all cardiac phases constituting the multiple input cardiac CT images.

[0036] (S330: Acquire seed information) In step S330, the seed information acquisition unit 230 acquires seed information based on the multiple initial regions acquired in step S320. Then, the seed information acquisition unit 230 outputs the acquired seed information to the detailed region acquisition unit 240.

[0037] In this embodiment, the seed information acquisition unit 230 acquires information on foreground seeds and background seeds as seed information. A foreground seed is an area where the multiple initial areas acquired in step S320 overlap. On the other hand, a background seed is an area that is not included in any of the multiple initial areas. First, an example of a process for acquiring foreground seed information will be specifically described with reference to FIG. 4.

[0038] In Figures 4A to 4D, for the sake of simplicity, a predetermined tomographic image (two-dimensional image) included in a three-dimensional image processed by the image processing device 100 is used as an example, but this embodiment is not limited to two-dimensional images and can also be applied to three-dimensional images.

[0039] 4A to 4C are sectional images at the same slice position included in three-dimensional images of consecutive cardiac phases among the 4DCT images acquired in step S310. In step S320, initial regions 401, 411, and 421 of the left atrium (regions in the mask image where foreground pixel values ​​are stored) are acquired from the images 400, 410, and 420, respectively. Then, the seed information acquisition unit 230 acquires regions where the initial regions 401, 411, and 421 overlap in the image space as foreground seeds. FIG. 4D shows a foreground seed 432 acquired from the image 410 as a predetermined image by the seed information acquisition unit 230 using the initial regions 401, 411, and 421.

[0040] Generally, when extracting an object region from an image, portions within the object tend to be extracted with higher accuracy than portions near the object's edges. Therefore, in step S320, the regions acquired as initial regions from images of multiple cardiac phases in which the left atrium has a similar shape have a high probability (likelihood or internal likelihood) of being regions contained within the left atrium. Therefore, the seed information acquisition unit 230 can identify regions with a high likelihood of being contained within the left atrium as foreground seeds. Here, the likelihood representing the probability that the initial region is included in the left atrium in the image from which the initial region is acquired may be said to indicate the degree of probability that the region of interest in the image is included in the left atrium. That is, when the likelihood representing the probability that the initial region is included in the left atrium in the image from which the initial region is acquired is equal to or greater than a predetermined threshold, this may be said to indicate that the region of interest in the image is likely to be included in the left atrium at a predetermined probability or higher.

[0041] Specifically, for example, in an image of a predetermined cardiac phase, a motion artifact is detected near the edge of the left atrium. Assume that the boundary of the left atrium is unclear due to artifacts or CT value variations caused by contrast agents. In this case, there is a possibility that regions other than the actual left atrium are extracted as the initial region due to the unclear boundary, resulting in so-called over-extraction. However, even if the edge of the left atrium is over-extracted in an image of a certain cardiac phase, the initial region may be appropriately extracted in another cardiac phase because there are fewer artifacts or the CT value variations caused by contrast agents are different. According to this embodiment, by acquiring a region where the initial region of the left atrium overlaps between a predetermined phase and another phase, the over-extracted region can be excluded and the region contained within the left atrium in the image of the predetermined cardiac phase can be identified as a foreground seed. That is, the seed information acquisition unit 230 can acquire foreground seed information by using each 3D image to identify regions with a high probability (likelihood) of being inside the left atrium as foreground seeds through the above process.

[0042] As an example, the seed information acquisition unit 230 acquires a first likelihood distribution of the left atrium region for the multiple initial regions acquired in step S320. In the first likelihood distribution, the likelihood increases as the probability that the initial region is located inside the left atrium increases. Since the first likelihood distribution for the initial region can be calculated using well-known techniques, detailed explanation of the calculation of the likelihood distribution will be omitted here. Furthermore, in the first likelihood distribution, the likelihood for a portion where at least two of the multiple initial regions overlap is higher than the likelihood for a portion where the at least two regions do not overlap. Therefore, the first likelihood distribution indicates that the overlapping portion of the multiple initial regions is more likely to be located inside the left atrium. As a result, for example, the seed information acquisition unit 230 can identify a portion in the acquired first likelihood distribution where the likelihood is equal to or greater than a threshold as the initial region and acquire information about the initial region.

[0043] Next, an example of a process for acquiring background seed information in this embodiment will be described with reference to FIGS. 4A to 4D. In this embodiment, the seed information acquisition unit 230 identifies, from an image 410 as a predetermined image, an area in image space that is not included in any of the initial areas 401, 411, and 421, as a background seed 433. An area that is not included in any of the initial areas has a high probability (likelihood or external likelihood) of being an area that is not included in the left atrium. Therefore, the seed information acquisition unit 230 identifies an area with a high likelihood among areas that are not included in the left atrium as a background seed, and acquires information about the background seed.

[0044] As explained in the description of the foreground seed acquisition method above, the boundary of the left atrium region may be unclear in an image of a predetermined cardiac phase for various reasons. In this case, the initial region may be extracted with the actual left atrium region missing due to the unclear boundary edge portion. However, even if the left atrium edge portion is extracted with a missing portion in an image of a certain cardiac phase, an appropriate initial region may be acquired in an image of another phase due to fewer artifacts or different CT value unevenness caused by contrast agents. According to this embodiment, by acquiring a region that is not included in either of the initial regions of the left atrium between the predetermined phase and other phases, the missing region as described above can be excluded, and a region that is not included in the left atrium in the image of the predetermined phase can be identified as a background seed. In other words, by the above processing, a region that has a high probability (likelihood) of being outside the left atrium in each 3D image can be identified as a background seed.

[0045] As an example, the seed information acquisition unit 230 acquires a second likelihood distribution of the region being outside the region for the multiple initial regions acquired in step S320, with the left atrium as the target region. Unlike the first likelihood distribution described above, in the second likelihood distribution, the higher the probability that the initial region is outside the left atrium, the higher the likelihood. The second likelihood distribution for the initial region can be calculated using well-known techniques, and therefore, detailed explanation of the calculation of the likelihood distribution will be omitted here. Furthermore, in the second likelihood distribution, at least two of the multiple initial regions are selected as the region outside the left atrium. The likelihood for a portion not included in any of the initial regions is higher than the likelihood for a portion included in one of the at least two regions. Therefore, this likelihood distribution indicates that a portion not included in the multiple initial regions is more likely to be outside the left atrium. As a result, for example, the seed information acquisition unit 230 can identify a portion in the acquired second likelihood distribution where the likelihood is equal to or greater than a threshold as a portion outside the left atrium, and acquire information about the portion outside the left atrium as information about the initial region.

[0046] In this embodiment, a method for acquiring seed information (information on the foreground seed 432 and background seed 433) for the image 410 has been described, but seed information for images of other cardiac phases can also be acquired by a method similar to that described above. Alternatively, the seed information for the acquired image 410 may be used as seed information for images of other phases. This allows for more efficient processing because seed information only needs to be acquired from one of the multiple three-dimensional images acquired by the input image acquisition unit 210.

[0047] The above-described process of acquiring a region where multiple initial regions overlap in image space as foreground seed information is relatively simple and efficient among various processes for acquiring foreground seed information. However, because the shape of the left atrium differs between multiple cardiac CT images, even if the initial region of the left atrium is properly acquired from an image of a certain cardiac phase, the region at the same image coordinates in another cardiac phase is not necessarily a region contained within the left atrium. For example, a region contained within the left atrium in an image of one cardiac phase may be located outside the left atrium in an image of another cardiac phase. In this case, if this position is over-extracted as the initial region, the above method may not be able to exclude the over-extracted region from the foreground seed. Therefore, a possible method is to use deformation information of the initial region, which can be acquired by, for example, aligning the contour position of the left atrium between multiple images corresponding to different phases of a region whose shape periodically changes in the image (e.g., the left atrium). In this case, the shape of the region corresponding to the left atrium in one cardiac phase and the initial region in another cardiac phase are deformed based on deformation information acquired, and the overlapping region between these regions is acquired as a foreground seed. This allows for more accurate acquisition of foreground seeds contained in the left atrium of an image at a predetermined cardiac phase.

[0048] Similarly to foreground seeds, background seeds may also be obtained using deformation information of an initial region that can be obtained by performing registration so that the contour position of the left atrium and the like coincide. In this case, a region that is not included in either the initial region of a certain cardiac phase or a deformed region obtained by deforming the initial region of another cardiac phase may be obtained as a background seed. This allows for more accurate acquisition of background seeds that are not included in the left atrium of an image of a specific cardiac phase.

[0049] Here, the alignment of an object (such as the left atrium) between multiple 3D images can be realized using a known method. For example, alignment is performed so that the similarity of the object between the images after deformation due to alignment is high. In this case, the image deformation model includes a Thin It is possible to use a deformation model based on a radial basis function such as Plate Spline (TPS) or a known deformation model such as Free Form Deformation (FFD). As an example of a method using a machine learning model, VoxelMorph, which is a type of deep learning, may be used for alignment.

[0050] Alternatively, instead of the above, a region obtained by contracting an overlapping region of multiple initial regions may be used as a foreground seed, or a predetermined pixel value threshold may be used to exclude from the foreground seed a region where multiple initial regions overlap, a region with a pixel value less than the threshold. Here, the threshold may be set based on the CT value of the left atrium acquired in a general contrast-enhanced CT image. The threshold is not necessarily set in advance, and the control unit 115 may adaptively set the threshold based on the distribution of CT values ​​(for example, the average or median) within multiple initial regions. This allows the left atrium to be over-extracted, compared to a method of simply acquiring overlapping regions of initial regions in image space. This is expected to have the effect of more appropriately excluding the areas that have been detected.

[0051] Similarly, a region that is not included in any of the multiple initial regions may be contracted and used as a background seed, or a predetermined pixel value threshold may be used to exclude from the background seed any region that is not included in any of the multiple initial regions and has a pixel value equal to or greater than the threshold. This is expected to have the effect of more appropriately excluding regions contained in the left atrium compared to a method that simply obtains regions that are not included in any of the multiple initial regions in image space.

[0052] In addition, when multiple cardiac CT images of the hearts of different subjects are taken, the position and shape of the heart in image space differ significantly between the images compared to multiple cardiac CT images of the same patient. Therefore, it is preferable to use deformation information obtained by aligning the images as described above, and identify the overlapping area between the initial area of ​​the specified image and the area obtained by deforming the initial area of ​​the other image based on the deformation information as a foreground seed. This allows over-extracted areas in the specified image to be excluded from the foreground seed.

[0053] Furthermore, when initial regions are acquired from multiple cardiac CT images of the hearts of different subjects, the difference in the positions of the cardiac contours deformed based on the deformation information may be greater between the images than when initial regions are acquired from multiple cardiac CT images of the same subject. In this case, even if overlapping areas between multiple initial regions are used as foreground seeds, it may be impossible to exclude over-extracted areas in a given image from the foreground seeds. Therefore, when initial regions are acquired from multiple cardiac CT images of the hearts of different subjects, it is preferable to also perform a process in which a contracted area where multiple initial regions overlap is used as the foreground seed. This allows for more accurate acquisition of foreground seeds contained within the left atrium of a given image, even when multiple cardiac CT images of the hearts of different subjects are used. When acquiring background seed information, initial regions can be acquired more accurately from cardiac CT images of different subjects, as in the case of foreground seeds.

[0054] The initial regions used to acquire foreground seeds and background seeds may be all of the initial regions acquired in step S320, or only some of the initial regions. When acquiring seed information using only some of the initial regions, the seed information acquisition unit 230 may calculate the degree of abnormality for each initial region and exclude initial regions with a high degree of abnormality from the targets for acquiring seed information. Here, the degree of abnormality is an index that indicates, for example, the difference in the volume of an initial region or the position of the initial region in image space (e.g., the center of gravity position) compared to other initial regions. If the degree of abnormality is high, it can be considered that there is a high possibility that acquisition of the initial region has failed.

[0055] Furthermore, the multiple initial regions used to obtain the foreground seed and background seed information do not necessarily have to be initial regions obtained from the same image. That is, the image from which the initial region used to obtain the foreground seed information is obtained and the image from which the initial region used to obtain the background seed information is obtained may be different images.

[0056] In this embodiment, it is assumed that foreground seed information is acquired based on an area where all of the multiple initial areas overlap. However, an area where a predetermined number or a predetermined percentage of the multiple initial areas used to acquire foreground seed information overlap may be used as the foreground seed. For example, an area where two or more initial areas overlap may be used as the foreground seed, or an area where more than half (50%) of the multiple initial areas used to acquire foreground seeds overlap may be used as the foreground seed.

[0057] Similarly, a region that is not included in any of the multiple initial regions used to obtain background seeds may be used as a background seed. Alternatively, a region that is not included in a predetermined number of initial regions or a predetermined percentage of the multiple initial regions used to obtain background seeds may be used as a background seed. For example, a region that is not included in two or more initial regions may be set as a background seed, or a region that is not included in half (50%) or more of the initial regions used to obtain background seeds may be set as a background seed.

[0058] Also, assume that step S320 acquires an image in which a value indicating the existence probability of an object (such as the left atrium) is stored as a pixel value in each pixel of the initial region. In this case, the seed information acquisition unit 230 may calculate a statistical value of the probability based on the pixel value of each pixel and acquire seed information based on the calculated statistical value. For example, the seed information acquisition unit 230 may determine whether the sum or average of the existence probabilities of the object is equal to or greater than a threshold value at each corresponding pixel position in multiple initial regions, and identify pixel positions determined to be equal to or greater than the threshold value as foreground seeds. Alternatively, the seed information acquisition unit 230 may determine whether the sum or average of the existence probabilities of the object is less than a threshold value at each corresponding pixel position in multiple initial regions, and identify pixel positions determined to be less than the threshold value as background seeds.

[0059] Furthermore, it is not necessary to acquire both the foreground seed and the background seed by the above method; information on either the foreground seed or the background seed may be acquired by the above method, and information on the other seed may be acquired by a method other than the above. For example, the seed information acquisition unit 230 may acquire information on the foreground seed by one of the above methods, and identify an area input by the user via the instruction unit 140 as the background seed. Alternatively, information on the background seed may be stored in advance in the data server 130, and the seed information acquisition unit 230 may acquire information on the background seed from the data server 130 and identify the background seed based on the acquired information.

[0060] Note that it is not always necessary to acquire background seed information, and the seed information acquisition unit 230 may omit the process of acquiring the background seed information. In this case, the processing time required to acquire background seed information can be reduced, and the seed information acquisition unit 230 outputs only the acquired foreground seed information to the detailed region acquisition unit 240.

[0061] (S340: Get detailed area) In step S340, the detailed region acquisition unit 240 acquires information about the detailed region corresponding to the cardiac cavity from a predetermined image among the images acquired in step S310, using the seed information acquired based on the information about the initial region in step S330. Then, the detailed region acquisition unit 240 outputs the acquired information about the detailed region to the display control unit 250.

[0062] In this embodiment, the detailed region can be identified from the image using a known segmentation method. For example, the detailed region acquisition unit 240 can identify the detailed region by graph cutting, regarding each pixel of a predetermined image from which the detailed region is acquired as a node and using foreground seeds and background seeds as the initial values ​​of each node. The detailed region acquisition unit 240 can also identify the region obtained by applying a region growing method or an active contour method to the foreground seeds as the detailed region.

[0063] Alternatively, the detailed region acquisition unit 240 may identify the detailed region based on the distribution of pixel values ​​contained in each of the foreground seed and background seed regions in a predetermined image from which the detailed region is acquired. For example, the detailed region acquisition unit 240 calculates a value (e.g., median or average) that represents each seed based on the distribution of pixel values ​​of each of the foreground seed and background seed, and uses the calculated value to acquire a threshold value of pixel values ​​that separates the regions of the two seeds. The detailed region acquisition unit 240 can then identify the detailed region based on the acquired threshold value. More specifically, the detailed region acquisition unit 240 may use the distribution of pixel values ​​of the foreground seed and the distribution of pixel values ​​of the background seed as initial clusters (foreground cluster and background cluster), and identify the detailed region based on these using a known method such as the k-means method. In this case, the detailed region acquisition unit 240 Of the identified detailed regions, regions included in the background seeds acquired in step S330 may be excluded from the detailed regions.

[0064] Therefore, in step S340, the detailed region acquisition unit 240 acquires a detailed region such that the probability of including the left atrium inside the region is higher than in the initial region, and the probability of including the left atrium outside the region is lower than in the initial region.

[0065] The predetermined image from which the detailed region is acquired is at least one image of the input image acquired in step S310. It may be the image of the cardiac phase from which the initial region was acquired in step S320, or an image of any other cardiac phase. The image from which the detailed region is acquired may be an image selected by the user from the multiple input images acquired in step S310 by operating the instruction unit 140. Alternatively, the image from which the detailed region is acquired may be selected based on the header information of the input image acquired in step S310. For example, assume that the header information of the input image includes information indicating the cardiac phase. In this case, the detailed region acquisition unit 240 can select an image with few motion artifacts, such as the end-systole or end-diastole cardiac phase, as the image from which the detailed region is acquired. Alternatively, the detailed region acquisition unit 240 may acquire the detailed region from images of all phases of the input image acquired in step S310.

[0066] (S350:Display) In step S350, the display control unit 250 controls the display of the cardiac CT image acquired in step S310 and the detailed region acquired in step S340 on the display unit 150. Specifically, the display control unit 250 generates an image to be displayed on the display unit 150 based on the cardiac CT image and the detailed region, and controls the display of the generated image on the display unit 150. For example, the display control unit 250 controls the display unit 150 to generate a CG-rendered image of the detailed region of the cardiac cavity. This allows the user to observe the three-dimensional shape of the cardiac cavity of the subject.

[0067] Furthermore, the display control unit 250 can provide functions similar to those of a general medical image viewer. For example, the display control unit 250 selects a 2D slice image from the cardiac CT image displayed on the display unit 150 in response to a user's operation of the instruction unit 140. Then, the display control unit 250 can generate an image in which a detailed region of the cardiac cavity corresponding to the selected slice image is superimposed on the selected slice image, and display the image on the display unit 150. Fig. 5 shows an example of an image in which a detailed region is superimposed on a slice image displayed on the display unit 150.

[0068] In the above description of this embodiment, information about the detailed region is acquired from image 410, which is a slice image. Therefore, the display control unit 250 generates an image in which the detailed region 501 is superimposed on image 410, using the detailed region acquired by the above processing on image 410. Then, the display control unit 250 controls the display unit 150 to display the generated image, thereby displaying the image in which the detailed region 501 is superimposed on image 410, as shown in FIG. 5 . This allows the user to accurately identify the region of the left atrium in image 410 from the detailed region 501 displayed on the display unit 150.

[0069] The control unit 115 may perform control to store the detailed region acquired in step S340 in the storage unit 114 of the image processing device 100, or may perform control to store it in the data server 130 via the network 120. This allows the control unit 115 to display an image of the detailed region on the display unit 150 using any other medical image viewer, or to read and use the detailed region in any other surgery support software, etc. In this case, the control unit 115 may omit the display process of step S350. Furthermore, the process of displaying and saving the detailed region is not essential, and the control unit 115 may perform control to store the detailed region acquired in step S340 in the storage unit 114 of the image processing device 100, or to store it in the data server 130 via the network 120. This allows the control unit 115 to display an image of the detailed region on the display unit 150 using any other medical image viewer, or to read and use the detailed region in any other surgery support software, etc. In this case, the control unit 115 may omit the display process of step S350. Furthermore, the process of displaying and saving the detailed region is not essential, and the control unit 115 may perform control to store the detailed region acquired in step S340 in the storage unit 114 of the image processing device 100. For example, the control unit 115 may calculate or estimate measurement values ​​(volume, diameter, etc.) related to the shape of the left atrium and attribute values ​​(values ​​indicating the state of blood flow, etc.) related to the state of the left atrium based on the detailed region.

[0070] As described above, the image processing device 100 of this embodiment can acquire multiple initial regions of the cardiac cavity from multiple 3D images obtained by imaging the heart of a subject, and can acquire detailed regions of the cardiac cavity from a predetermined 3D image based on the acquired multiple initial regions. A user of the image processing device 100 can then check the acquired detailed regions of the cardiac cavity and measurement values ​​measured from the detailed regions of the cardiac cavity, thereby understanding the structure and condition of the subject's heart more easily and accurately than ever before.

[0071] Next, a modified example of the first embodiment will be described. In the following description, the same configurations and processes as those of the image processing system 10 according to the first embodiment will be denoted by the same reference numerals, and detailed description thereof will be omitted.

[0072] (Variation 1) In the image processing device 100 according to the first embodiment described above, the detailed region acquired in step S340 is output to the display control unit 250. However, in the image processing device 100 according to this modification, for example, the detailed region acquisition unit 240 outputs the detailed region of the left atrium of the predetermined image acquired in step S340 to the initial region acquisition unit 220. Then, the initial region acquisition unit 220 may replace the initial region with the detailed region input from the detailed region acquisition unit 240, and perform the processing of step S320 again.

[0073] As a specific example, assume that the image from which the detailed region is acquired in step S340 is image 410 shown in Fig. 4D. In this case, in step S320, the initial region acquisition unit 220 replaces the initial region 411 acquired from image 410 with the detailed region. This allows the initial region acquisition unit 220 to acquire the initial region as a region in which the target object (such as the left atrium) can be identified with higher accuracy. Therefore, according to this modification, it is expected that in step S340, the detailed region acquisition unit 240 will be able to acquire information about the detailed region with higher accuracy than the detailed region acquired initially.

[0074] Furthermore, in the image processing device 100 of this modification, after replacing the detailed region with the initial region in step S320, when step S340 is executed again to acquire the detailed region, it is not necessary to acquire the detailed region from the same image used in the previous step S340. For example, in the case of 4DCT images obtained by imaging a beating heart over time, the control unit 115 acquires the detailed region from the first frame image in the initial step S340. In the subsequent step S320, the control unit 115 replaces the initial region acquired in the first frame image with the acquired detailed region. Then, in the next step S340, the control unit 115 acquires the detailed region from the image of the frame (second frame) subsequent to the first frame image used in the previous step S340. Furthermore, in the subsequent step S320, the control unit 115 replaces the initial region acquired in the second frame image with the acquired detailed region. Then, the control unit 115 may repeatedly execute the processes from step S320 to step S340 until detailed regions are acquired for all frame images acquired by the input image acquisition unit 210.

[0075] Therefore, according to the image processing device 100 of this modified example, an area that provides higher accuracy in extracting the object from the multiple three-dimensional images acquired by the input image acquisition unit 210 is used as the initial area, thereby making it possible to more accurately acquire information about the detailed area from a specified three-dimensional image.

[0076] (Variation 2) Next, a second modification of the first embodiment will be described. In step 00, a detailed region of the left atrium, which is the cavity of the heart, is acquired. However, the image processing device 100 according to this modification can acquire detailed regions of organs other than the heart.

[0077] As an example of this modification, assume that the image processing device 100 processes a dynamic CT image of a liver. In contrast CT images of the liver, images of a phase (e.g., arterial phase) in which a short time has elapsed since the administration of a contrast agent to the subject have a large difference in CT value between the liver and surrounding blood vessels, so the image processing device 100 tends to extract the initial region near the boundary between the liver and blood vessels with high accuracy. On the other hand, the image processing device 100 tends to extract the initial region near the boundary between the liver and surrounding soft tissue with low accuracy, because the difference in CT value between the liver parenchyma and surrounding soft tissue is small.

[0078] Furthermore, in the image processing device 100, in an image of a phase where a long time has elapsed since the administration of a contrast agent to a subject (for example, the portal vein phase or the equilibrium phase), the difference in CT value between the liver and the surrounding blood vessels is small, so the initial region near the boundary between the liver and the blood vessels tends to be extracted with low accuracy.On the other hand, in the image processing device 100, the difference in CT value between the liver parenchyma and the surrounding soft tissues is large, so the initial region near the boundary between the liver and the surrounding soft tissues tends to be extracted with high accuracy.

[0079] The image processing device 100 acquires an overlapping region of initial regions acquired from images of a short phase and a long phase after administration of a contrast agent to the subject. This allows the image processing device 100 to exclude regions extracted due to over-extraction near the boundary between the liver and the surrounding blood vessels and soft tissues. That is, the image processing device 100 according to this modification is expected to be effective in acquiring regions with a high probability (likelihood) of being inside the liver as foreground seeds. Similarly, the image processing device 100 according to this modification is expected to be effective in acquiring regions with a high probability (likelihood) of being outside the liver as background seeds, excluding missing regions near the boundary between the liver and the surrounding blood vessels and soft tissues.

[0080] In this modified example, we have described the case where detailed regions are acquired by targeting the liver, which is different from the cardiac cavity in the first embodiment, but other parts of the subject (e.g., the lungs) may also be the target, or parts other than the human body (e.g., industrial components) may also be the target.

[0081] Therefore, according to the image processing device 100 of this modification, detailed regions can be acquired from a plurality of three-dimensional images acquired by the input image acquisition section 210 for various parts of a subject or an object other than a human body.

[0082] In the above embodiment and modified examples, the various controls described as being performed by control unit 115 may or may not be performed by a single piece of hardware (e.g., a processor or a circuit). The entire device may be controlled by multiple pieces of hardware (e.g., multiple processors, multiple circuits, or a combination of one or more processors and one or more circuits) sharing the processing.

[0083] The above processor is a processor in a broad sense, and includes general-purpose processors and dedicated processors. General-purpose processors include, for example, CPUs (Central Processing Units), MPs, U (Micro Processing Unit), DSP (Digital Signal Processor), etc. The processor is, for example, a GPU (Graphics Processing Unit). The dedicated processor is, for example, an ASIC (Application Specific Integrated Circuit), LD (Programmable Logic Device) and so on. Programmable logic devices are, for example, Examples include FPGA (Field Programmable Gate Array) and CPLD (Complex Programmable Logic Device).

[0084] Furthermore, although the embodiments and modifications of the present invention have been described in detail, the present invention is not limited to these specific embodiments, and various forms within the scope of the gist of the present invention are also included in the present invention. Furthermore, the above-described embodiments and modifications merely represent one embodiment of the present invention, and the above-described embodiments and modifications can be combined as appropriate.

[0085] <Other embodiments> Furthermore, the disclosed technology can be embodied as, for example, a system, a device, a method, a program, or a recording medium (storage medium), etc. Specifically, it may be applied to a system consisting of multiple devices (for example, a host computer, an interface device, an imaging device, a web application, etc.), or it may be applied to an apparatus consisting of a single device.

[0086] Needless to say, the object of the present invention can be achieved by the following: Namely, a recording medium (or storage medium) on which software program code (computer program) that realizes the functions of the above-described embodiments is recorded is supplied to a system or device. Needless to say, such a recording medium is a computer-readable recording medium. Then, a computer (or CPU or MPU) of the system or device reads and executes the program code stored on the recording medium. In this case, the recording medium on which the program code read from the recording medium is recorded constitutes the present invention.

[0087] The present invention can also be realized by supplying a program that realizes one or more functions of the above-described embodiments to a system or device via a network or a storage medium, and having one or more processors in the computer of the system or device read and execute the program. It can also be realized by a circuit (e.g., ASIC) that realizes one or more functions.

[0088] The disclosure of this embodiment includes the following configuration, method, and program. (Configuration 1) an image acquisition unit that acquires a plurality of images of a region of a subject; an initial region acquisition unit that acquires information about a plurality of initial regions corresponding to the region from a predetermined number of images among the plurality of images; a detailed region acquisition unit that acquires information about a detailed region corresponding to the part in any of the plurality of images based on information about the plurality of initial regions; An image processing device having: (Configuration 2) The image processing device described in configuration 1, wherein the detailed region acquisition unit acquires the detailed region so that the probability of including the part inside the region is higher than in the initial region, and the probability of including the part outside the region is lower than in the initial region. (Configuration 3) an image acquisition unit that acquires a plurality of images of a region of a subject; a first information acquisition unit that acquires information about a plurality of first regions corresponding to the part from a predetermined number of images among the plurality of images; a second information acquisition unit that acquires, based on information about the plurality of first regions, information about a second region that corresponds to the region in any of the plurality of images and has a higher degree of matching with the shape of the region than the first region; An image processing device having: (Configuration 4) The image processing device described in configuration 3, characterized in that the first information acquisition unit acquires a first likelihood distribution that the region is the part from the predetermined number of images, and acquires information about the first region based on the first likelihood distribution. (Configuration 5) The image processing device described in configuration 4, characterized in that in the first likelihood distribution, the likelihood for a portion that overlaps between at least two of the plurality of first regions is higher than the likelihood for a portion that does not overlap between the at least two regions. (Configuration 6) 6. The image processing device according to configuration 4 or 5, wherein the first information acquisition unit acquires information about a portion in the first likelihood distribution where the likelihood is equal to or greater than a threshold as information about the first region. (Configuration 7) The image processing device described in any one of configurations 3 to 6, characterized in that the first information acquisition unit acquires a second likelihood distribution that the area is outside the part from the predetermined number of images, and acquires information about the first area based on the second likelihood distribution. (Configuration 8) The image processing device described in configuration 7, characterized in that in the second likelihood distribution, the likelihood for a portion that is not included in any of at least two of the multiple first regions is higher than the likelihood for a portion that is included in any of the at least two regions. (Configuration 9) The image processing device according to configuration 7 or 8, wherein the first information acquisition unit acquires information about the area outside the region where the likelihood in the second likelihood distribution is equal to or greater than a threshold as information about the first region. (Configuration 10) 10. The image processing device according to any one of configurations 3 to 9, further comprising an image selection unit that selects, from the plurality of images acquired by the image acquisition unit, two or more images from which information about the first region is acquired by the first information acquisition unit. (Configuration 11) 11. The image processing device according to claim 10, wherein the image selection unit selects the two or more images based on morphological information or phase information of the region in the plurality of images acquired by the image acquisition unit. (Configuration 12) 12. The image processing device according to any one of configurations 3 to 11, wherein the plurality of images acquired by the image acquisition unit include a plurality of images of the region of the subject captured over time. (Configuration 13) the plurality of images include images corresponding to different phases of the part whose shape periodically changes, The first information acquisition unit acquires information about the first region using an image of the part corresponding to a predetermined phase among the plurality of images. 13. The image processing device according to any one of configurations 3 to 12. (Configuration 14) The image processing device described in configuration 13, characterized in that the first information acquisition unit acquires information about the first region by changing the shape of the region corresponding to the part in the image corresponding to the specified phase. (Method 1) acquiring a plurality of images of a region of a subject; acquiring information about a plurality of initial regions corresponding to the region from a predetermined number of images among the plurality of images; acquiring information about a detailed region corresponding to the part in any of the plurality of images based on information about the plurality of initial regions; An image processing method comprising: (Method 2) acquiring a plurality of images of a region of a subject; acquiring information about a plurality of first regions corresponding to the part from a predetermined number of images among the plurality of images; acquiring information about a second region that corresponds to the region in any of the plurality of images and has a higher degree of matching with the shape of the region than the first region, based on information about the plurality of first regions; An image processing method comprising: (program) A program for causing a computer to execute each step of the image processing method according to Method 1 or 2. [Explanation of symbols]

[0089] 100 Image processing device, 115 Control unit, 210 Input image acquisition unit, 220 Initial region acquisition unit, 230 Seed information acquisition unit, 240 Detailed region acquisition unit

Claims

1. an image acquisition unit that acquires a plurality of images of a region of a subject; an initial region acquisition unit that acquires information about a plurality of initial regions corresponding to the region from a predetermined number of images among the plurality of images; a detailed region acquisition unit that acquires information about a detailed region corresponding to the part in any of the plurality of images based on information about the plurality of initial regions; An image processing device having:

2. The image processing device according to claim 1 , wherein the detailed region acquisition unit acquires the detailed region so that the probability of including the part inside the region is higher than that of the initial region and the probability of including the part outside the region is lower than that of the initial region.

3. an image acquisition unit that acquires a plurality of images of a region of a subject; a first information acquisition unit that acquires information about a plurality of first regions corresponding to the part from a predetermined number of images among the plurality of images; a second information acquisition unit that acquires, based on information about the plurality of first regions, information about a second region that corresponds to the region in any of the plurality of images and has a higher degree of matching with the shape of the region than the first region; An image processing device having:

4. The image processing device according to claim 3, characterized in that the first information acquisition unit acquires a first likelihood distribution that the region is the part from the predetermined number of images, and acquires information about the first region based on the first likelihood distribution.

5. 5. The image processing device according to claim 4, wherein in the first likelihood distribution, the likelihood for a portion that overlaps between at least two of the plurality of first regions is higher than the likelihood for a portion that does not overlap between the at least two regions.

6. The image processing device according to claim 4 , wherein the first information acquisition unit acquires information about a portion in the first likelihood distribution where the likelihood is equal to or greater than a threshold value as information about the first region.

7. The image processing device according to claim 3, characterized in that the first information acquisition unit acquires a second likelihood distribution that the region is outside the part from the predetermined number of images, and acquires information about the first region based on the second likelihood distribution.

8. 8. The image processing device according to claim 7, wherein in the second likelihood distribution, the likelihood for a portion that is not included in any of at least two of the plurality of first regions is higher than the likelihood for a portion that is included in any of the at least two regions.

9. The image processing device according to claim 7 , wherein the first information acquisition unit acquires, as the information about the first region, information about the outside of the part where the likelihood in the second likelihood distribution is equal to or greater than a threshold value.

10. 5. The image processing device according to claim 3, further comprising an image selection unit that selects, from the plurality of images acquired by the image acquisition unit, two or more images from which information regarding the first region is acquired by the first information acquisition unit.

11. The image selection unit selects the image from the plurality of images acquired by the image acquisition unit. The image processing apparatus according to claim 10 , wherein the two or more images are selected based on morphological information or phase information of the region.

12. 5. The image processing apparatus according to claim 3, wherein the plurality of images acquired by the image acquisition unit include a plurality of images of the region of the subject captured over time.

13. the plurality of images include images corresponding to different phases of the part whose shape periodically changes, The first information acquisition unit acquires information about the first region using an image of the part corresponding to a predetermined phase among the plurality of images.

5. The image processing device according to claim 3, wherein the image processing device is a computer.

14. The image processing device according to claim 13 , wherein the first information acquisition unit acquires information about the first region by changing a shape of a region corresponding to the part in the image corresponding to the predetermined phase.

15. acquiring a plurality of images of a region of a subject; acquiring information about a plurality of initial regions corresponding to the region from a predetermined number of images among the plurality of images; acquiring information about a detailed region corresponding to the part in any of the plurality of images based on information about the plurality of initial regions; An image processing method comprising:

16. acquiring a plurality of images of a region of a subject; acquiring information about a plurality of first regions corresponding to the part from a predetermined number of images among the plurality of images; acquiring information about a second region that corresponds to the region in any of the plurality of images and has a higher degree of matching with the shape of the region than the first region, based on information about the plurality of first regions; An image processing method comprising:

17. A program for causing a computer to execute each step of the image processing method according to claim 15 or 16.

Images