Contrast enhancement state determination device, contrast enhancement state determination method, and program

Abstract

Provided are a contrast state determination device, a contrast state determination method, and a program that determine a contrast state at high speed, with high accuracy, and with robustness even if there is an organ that is not included in an image. The present invention acquires, from a first series of images which have been captured before injection of a contrast agent into a subject or after the injection, a plurality of two-dimensional images that include respective pieces of information pertaining to slice images of mutually different locations in the subject, estimates, for the plurality of two-dimensional images, index values pertaining to a contrast state from each of the two-dimensional images, and determines the contrast state of the first series of images on the basis of a plurality of the index values.

Description

[Technical Field]

[0001] The present invention relates to an imaging condition determination device, an imaging condition determination method, and a program for determining an imaging condition from an image obtained by imaging, and more particularly to a technique for determining an imaging condition from a two-dimensional image. [Background technology]

[0002] Dynamic contrast-enhanced CT images taken using CT (Computed Tomography) examinations using contrast agents appear significantly different depending on the contrast phase, and correctly understanding the contrast phase is necessary for processes that depend on contrast, such as blood vessel extraction. Although contrast information is sometimes included in DICOM (Digital Imaging and Communications in Medicine) tags, it is not always included, and it may even be incorrect. Therefore, it is necessary to understand the contrast phase from the image.

[0003] Patent Document 1 discloses a technique for determining whether or not a contrast image of the liver is present. Furthermore, contrast images have multiple phases (for example, pre-contrast, arterial phase, portal venous phase, and equilibrium phase), and Non-Patent Document 1 also discloses a technique for automatically determining these phases from an image. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 5357818 [Non-patent literature]

[0005] [Non-Patent Document 1] Automatic Contrast Phase Estimation in CT Volumes, Michal Sofka, Dijia Wu, Michael S¨uhling, David Liu, Christian Tietjen, Grzegorz Soza, and S. Kevin Zhou<URL:https: / / link.springer.com / content / pdf / 10.1007%2F978-3-642-23626-6_21.pdf> Summary of the Invention [Problem to be solved by the invention]

[0006] The information on the contrast-enhanced phase is required for subsequent processing that depends on the contrast-enhanced phase, and high-speed operation is required. However, the technology disclosed in Non-Patent Document 1 has the problem that it takes time because it processes the contrast-enhanced image as it is in 3D (dimensional) form. In addition, the technology disclosed in Patent Document 1 has the problem that it cannot determine the contrast-enhanced phase when a specific region is not shown in the contrast-enhanced image.

[0007] The present invention has been made in view of the above circumstances, and aims to provide an enhancement state determination device, an enhancement state determination method, and a program that can determine the enhancement state quickly, accurately, and robustly even when there are organs not included in the image. [Means for solving the problem]

[0008] One aspect of a contrast state discrimination device for achieving the above-mentioned object is a contrast state discrimination device that includes at least one processor and at least one memory that stores instructions to be executed by the at least one processor, wherein the at least one processor acquires a plurality of two-dimensional images, each containing information on slice images at different positions of the subject, from a first image series taken before or after injection of a contrast agent into the subject, estimates an index value related to the contrast state from each of the plurality of two-dimensional images, and discriminates the contrast state of the first image series based on the estimated index values.

[0009] That is, the contrast state determination device acquires, from a first image series captured before or after injection of a contrast agent into the subject, a first two-dimensional image including information on at least a first slice image at a first position on the subject and a second two-dimensional image including information on a second slice image at a second position different from the first position, estimates a first index value related to the contrast state from the first two-dimensional image, estimates a second index value related to the contrast state from the second two-dimensional image, and determines the contrast state of the first image series based on at least the first index value and the second index value. The contrast state determination device can further acquire, from the first image series, a third two-dimensional image including information on a third slice image at a third position different from the first and second positions on the subject, estimate a third index value related to the contrast state from the third two-dimensional image, and determine the contrast state of the first image series based on the first to third index values.

[0010] According to this aspect, the enhancement state of an image series is determined based on index values ​​estimated from two-dimensional images of multiple different positions on the subject, so that the enhancement state can be determined quickly, accurately, and robustly even if there are organs not included in the images.

[0011] The index values ​​are likelihoods that each of the multiple contrast states belongs to a particular contrast state, and it is preferable that at least one processor derives an index value that integrates the multiple index values ​​and determines the contrast state of the first image series based on the integrated index value.

[0012] It is preferable that the system includes a first learning model that outputs the likelihood that a two-dimensional image based on an image series taken before or after injection of a contrast agent into a subject is input, and that at least one processor inputs the multiple two-dimensional images into the first learning model and estimates multiple index values.

[0013] The index value is the elapsed time since the contrast agent was injected into the subject, and it is preferable that the at least one processor derives an elapsed time by integrating the multiple elapsed times that are the multiple index values, and determines the contrast state of the first image series based on the integrated elapsed time.

[0014] The index value is the elapsed time since the injection of the contrast agent into the subject and the likelihood of the elapsed time, and it is preferable that at least one processor derives an elapsed time by integrating the multiple elapsed times, which are the multiple index values, based on the multiple likelihoods, which are the multiple index values, and determines the contrast state of the first image series based on the integrated elapsed time.

[0015] Preferably, the at least one processor derives the integrated elapsed time based on the product of a plurality of probability distribution models, each of which has elapsed time and likelihood as parameters.

[0016] It is preferable that the system is provided with a second learning model that outputs the elapsed time since the injection of the contrast agent into the subject and the likelihood of the elapsed time when a two-dimensional image based on an image series taken before or after the injection of the contrast agent into the subject is input, and that at least one processor inputs multiple two-dimensional images into the second learning model to estimate multiple elapsed times and multiple likelihoods.

[0017] It is preferable that a conversion table be provided in which the elapsed time since the injection of contrast agent into the subject and the contrast state are associated, and that at least one processor determine the contrast state of the first image series based on the integrated elapsed time and the conversion table.

[0018] It is preferable that a conversion table be provided in which the elapsed time since the injection of the contrast agent into the subject and the contrast state are associated, and that at least one processor estimates multiple elapsed times since the injection of the contrast agent into the subject as multiple index values, determines multiple contrast states of the first image series based on the multiple elapsed times and the conversion table, and determines the contrast state of the first image series based on the multiple contrast states.

[0019] The two-dimensional image is preferably at least one of a slice image, a maximum intensity projection (MIP) image of a plurality of slice images, and an average image of a plurality of slice images.

[0020] Preferably, the first image series is a three-dimensional image including the liver of the subject, and the contrast-enhanced states include at least one of a non-contrast, an arterial phase, a portal venous phase, and an equilibrium phase.

[0021] The first image series is preferably a three-dimensional image including the kidneys of the subject, and the contrast-enhanced state preferably includes at least one of a non-contrast, corticomedullary phase, parenchymal phase, and excretory phase.

[0022] One aspect of a contrast state discrimination method for achieving the above-mentioned object is a contrast state discrimination method comprising the steps of: acquiring a plurality of two-dimensional images, each containing information on slice images at different positions of the subject, from a first image series taken before or after injection of a contrast agent into the subject; estimating an index value relating to the contrast state from each of the plurality of two-dimensional images; and discriminating the contrast state of the first image series based on the estimated index values.

[0023] One aspect of a program for achieving the above object is a program for causing a computer to execute the above-mentioned contrast state determination method. This aspect may also include a computer-readable non-transitory storage medium on which this program is recorded. [Effects of the Invention]

[0024] According to the present invention, it is possible to quickly, highly accurately, and robustly determine the enhancement state even when there are organs not included in the image. [Brief explanation of the drawings]

[0025] [Figure 1] FIG. 1 is a conceptual diagram showing an outline of the processing performed by the contrast state determination device according to the first embodiment. [Figure 2] FIG. 2 is a flowchart showing a contrast state determination method by the contrast state determination device. [Figure 3] FIG. 3 is a block diagram schematically illustrating an example of the hardware configuration of the contrast condition determination apparatus according to the first embodiment. [Figure 4] FIG. 4 is a functional block diagram showing an outline of the processing functions of the contrast condition determination apparatus according to the first embodiment. [Figure 5] FIG. 5 is a conceptual diagram showing an outline of the processing performed by the contrast state determination device according to the second embodiment. [Figure 6] FIG. 6 is a functional block diagram showing an outline of the processing functions of the contrast condition determination apparatus according to the second embodiment. [Figure 7] FIG. 7 is a conceptual diagram showing an outline of the processing performed by the contrast state determination device according to the third embodiment. [Figure 8] FIG. 8 is an explanatory diagram showing an example of processing in the number-of-seconds distribution estimation unit. [Figure 9] FIG. 9 is a graph of the function used in the variable transformation of equation (2). [Figure 10] FIG. 10 is a graph of the distribution of seconds estimated using the parameters estimated by the distribution of seconds estimation unit. [Figure 11]FIG. 11 is an explanatory diagram showing an example of processing in the integration unit and maximum point identification unit. [Figure 12] FIG. 12 is a diagram illustrating an example of a machine learning method for generating a regression model to be applied to the number-of-seconds distribution estimation unit. [Figure 13] FIG. 13 is an explanatory diagram of the loss function used during training. [Figure 14] FIG. 14 is a functional block diagram showing an outline of the processing functions of the contrast condition determination apparatus according to the third embodiment. [Figure 15] FIG. 15 is an explanatory diagram showing an example of processing in the number-of-seconds distribution estimation unit of the contrast enhancement state determination apparatus according to the fourth embodiment. [Figure 16] FIG. 16 is a graph of the distribution of seconds estimated using the parameters estimated by the distribution of seconds estimation unit. [Figure 17] FIG. 17 is an explanatory diagram showing an example of processing in the integrating unit and maximum point identifying unit of the contrast condition determining apparatus according to the fourth embodiment. [Figure 18] FIG. 18 is an explanatory diagram that schematically illustrates an example of a machine learning method for generating a regression model that is applied to the number-of-seconds distribution estimation unit in the fourth embodiment. [Figure 19] FIG. 19 is an explanatory diagram showing a first modification of data used for input to the contrast state determination device. [Figure 20] FIG. 20 is an explanatory diagram showing a second modification of the data used for inputting to the contrast state determination device. [Figure 21] FIG. 21 is a block diagram showing an example of the configuration of a medical information system to which the contrast state determination device is applied. DETAILED DESCRIPTION OF THE INVENTION

[0026] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] Overview of the contrast state determination device 10 according to the first embodiment FIG. 1 is a conceptual diagram showing an overview of processing by a contrast state determination device 10 according to a first embodiment. Here, an example of a contrast state determination device 10 that estimates the contrast state (contrast enhancement phase) of an image series based on multiple slice images of the same image series obtained by contrast imaging using a CT (Computed Tomography) device is described. In the first embodiment, multiple slice images in the same series are used as input, and the contrast state is estimated by image analysis. "By image analysis" means by processing based on pixel values ​​constituting image data.

[0028] The contrast state of liver contrast imaging includes at least one of non-contrast, arterial phase (early arterial phase / late arterial phase), portal venous phase, and equilibrium phase. The contrast state of kidney contrast imaging includes at least one of non-contrast, corticomedullary phase, parenchymal phase, and excretory phase. Here, an example of determining the contrast state of an image series obtained by liver contrast imaging will be described.

[0029] The contrast condition determination apparatus 10 can be realized using computer hardware and software. As shown in FIG.

[0030] A plurality of slice images are input to the contrast state determination device 10. The plurality of slice images are images sampled at equal intervals from 3D CT data of the same image series captured before or after injection of a contrast agent into a patient SB (an example of a "subject"). In the example shown in FIG. 1, three images IM1, IM2, and IM3 (examples of "multiple slice images" and examples of "multiple two-dimensional images") of the patient SB at different positions SBP1, SBP2, and SBP3, respectively, are input to the contrast state determination device 10. The slice images may also be referred to as tomographic images. The slice images may be understood to be substantially two-dimensional images (cross-sectional images).

[0031] The phase estimation unit 30 estimates an index value (an example of an "index value related to the contrast state") that is the likelihood that an image series of input images belongs to each of a plurality of contrast phases. The numerical range of the likelihood that each image belongs to each contrast phase output from the phase estimation unit 30 may be "0% to 100%." ​​In FIG. 1, three phase estimation units 30 are shown to show the processing flow when three different images IM1, IM2, and IM3 are input, but the phase estimation units 30 to which the images IM1 to IM3 are input are the same (single) processing unit.

[0032] The phase estimation unit 30 includes a trained model 30A (an example of a "first learning model") trained by machine learning. The trained model 30A is a multi-class classification model that, when two-dimensional images based on three-dimensional images obtained by contrast imaging are input, outputs the likelihood that the input image series of two-dimensional images belongs to each of a plurality of contrast imaging phases. The trained model 30A is configured using, for example, a convolutional neural network (CNN). The trained model 30A is trained by a well-known method using pairs of two-dimensional images based on an image series acquired before or after the injection of a contrast agent into a subject and the contrast imaging phases of the image series as training data. The phase estimation unit 30 inputs the input two-dimensional images to the trained model 30A and estimates the index value of each image.

[0033] 1, the phase estimation unit 30 outputs an estimation result PS1 as an index value for image IM1, in which the likelihood that the image belongs to the arterial phase among multiple enhancement states is 10%, the likelihood that the image belongs to the portal vein phase is 50%, and the likelihood that the image belongs to the balance phase is 40%. The phase estimation unit 30 also outputs an estimation result PS2 as an index value for image IM2, in which the likelihood that the image belongs to the arterial phase is 5%, the likelihood that the image belongs to the portal vein phase is 80%, and the likelihood that the image belongs to the balance phase is 15%. The phase estimation unit 30 also outputs an estimation result PS3 as an index value for image IM3, in which the likelihood that the image belongs to the arterial phase is 30%, the likelihood that the image belongs to the portal vein phase is 35%, and the likelihood that the image belongs to the balance phase is 35%.

[0034] The integrating unit 32 derives an index value by integrating the input index values ​​and determines the enhancement state of the image series. The integrating unit 32 may integrate the index values ​​by the maximum value, by majority vote, or by probability sum. The index value integrated by the integrating unit 32 is output as a final estimation result. In the example shown in FIG. 1, the integrating unit 32 integrates the estimation results PS1, PS2, and PS3, and outputs the portal vein phase as the final estimation result.

[0035] <<Explanation of medical images used for input>> The DICOM (Digital Imaging and Communications in Medicine) standard, which defines the format and communication protocol for medical images, defines a series ID within a unit called a study ID, which is an identification code (ID) used to specify the type of examination.

[0036] For example, in liver angiography for a certain patient, CT scans of the area including the liver are performed multiple times (four times in this example) with different scan timings as shown below. [First scan] Before contrast injection [Second scan] 35 seconds after contrast injection [Third shot] 70 seconds after contrast injection [4th scan] 180 seconds after contrast injection

[0037] These four imaging sessions yield four types of CT data. The term "CT data" as used herein refers to three-dimensional data composed of multiple consecutive slice images (tomographic images), and a collection of multiple slice images (a group of consecutive slice images) that make up the three-dimensional data is called an "image series." CT data is an example of a "three-dimensional image" in this disclosure.

[0038] The four types of CT data obtained by a series of imaging including the above four imaging sessions are each assigned the same study ID and a separate series ID.

[0039] For example, a study ID for a liver contrast imaging test for a specific patient is assigned "Study 1," and the series ID for CT data obtained by imaging before contrast injection is "Series 1," the CT data obtained by imaging 35 seconds after contrast injection is "Series 2," the CT data obtained by imaging 70 seconds after contrast injection is "Series 3," and the CT data obtained by imaging 180 seconds after contrast injection is "Series 4." A unique ID is assigned to each series. Therefore, CT data can be identified by the combination of the study ID and series ID. However, in actual CT data, the correspondence between the series ID and the imaging timing (the time elapsed after contrast injection) may not be clearly understood.

[0040] Furthermore, because three-dimensional CT data is large in size, it may be difficult to use the CT data as input data as is to perform processing to determine the contrast enhancement state.

[0041] <<Method for determining contrast state>> 2 is a flowchart showing the steps of the enhancement state determination method performed by the enhancement state determination device 10. Here, an example will be described in which the enhancement state of an image series obtained by liver angiography is determined.

[0042] In step S1, the contrast state determination device 10 acquires an image IM1 (an example of a "first slice image", an example of a "first two-dimensional image") of a position SBP1 (an example of a "first position") of the patient SB from an image series (an example of a "first image series") taken before or after the injection of a contrast agent into the patient SB.

[0043] In step S2, the phase estimation unit 30 inputs the image IM1 acquired in step S1 into the trained model 30A to estimate a first index value related to the enhancement state. Here, the phase estimation unit 30 calculates the probability of non-enhancement (non-enhancement likelihood), the probability of the arterial phase (arterial phase likelihood), the probability of the portal vein phase (portal vein phase likelihood), and the probability of the equilibrium phase (equilibrium phase likelihood) as the first index values, and outputs the estimation results PS1.

[0044] In step S3, the contrast condition determination device 10 acquires an image IM2 (an example of a "second slice image", an example of a "second two-dimensional image") of position SBP2 (an example of a "second position") of patient SB from the same image series as image IM1.

[0045] In step S4, the phase estimation unit 30 inputs the image IM2 acquired in step S3 into the trained model 30A to estimate a second index value related to the enhancement state. Here, the phase estimation unit 30 calculates the probability of non-enhanced, the probability of the arterial phase, the probability of the portal venous phase, and the probability of the equilibrium phase as the second index values, and outputs the estimation results PS2.

[0046] Similarly, the enhancement state determination device 10 acquires an image IM3, and the time phase estimation unit 30 outputs an estimation result PS3 as an index value related to the enhancement state from the image IM3.

[0047] In step S5, the integrating unit 32 determines the enhancement state of the image series based on the estimation results PS1 to PS3, and outputs the determined enhancement state as the final estimation result.

[0048] As described above, the contrast enhancement state determination method by the contrast enhancement state determination device 10 determines the contrast enhancement phase based on multiple two-dimensional images, which is more accurate and robust than determining the contrast enhancement phase based on a single two-dimensional image. Furthermore, the inference is faster than determining the contrast enhancement phase based on a three-dimensional image. Furthermore, since the estimation is not based on a specific region, estimation is possible even if there are organs not visible in the image.

[0049] Here, the contrast state determination device 10 determines the contrast phase based on three images IM1, IM2, and IM3, but may determine the contrast phase based on two images, or may determine the contrast phase based on four or more images. A relatively small number of images results in relatively fast processing, while a relatively large number of images results in relatively high accuracy and robustness of the determination result.

[0050] <<Example of hardware configuration>> FIG. 3 is a block diagram schematically illustrating an example of the hardware configuration of the contrast condition determination apparatus 10 according to the first embodiment. The contrast condition determination apparatus 10 can be realized by a computer system configured using one or more computers. Here, an example will be described in which one computer executes a program to realize various functions of the contrast condition determination apparatus 10. Note that the form of the computer that functions as the contrast condition determination apparatus 10 is not particularly limited, and may be a server computer, a workstation, a personal computer, a tablet terminal, or the like.

[0051] The imaging condition determination device 10 includes a processor 102 , a non-transitory tangible computer-readable medium 104 , a communication interface 106 , an input / output interface 108 , and a bus 110 .

[0052] The processor 102 includes a CPU (Central Processing Unit). The processor 102 may also include a GPU (Graphics Processing Unit). The processor 102 is connected to a computer-readable medium 104, a communication interface 106, and an input / output interface 108 via a bus 110. The processor 102 reads various programs, data, etc. stored in the computer-readable medium 104 and executes various processes.

[0053] The computer-readable medium 104 includes, for example, a memory 104A that is a primary storage device and a storage 104B that is an auxiliary storage device. The storage 104B is configured using, for example, a hard disk drive (HDD) device, a solid state drive (SSD) device, an optical disk, a magneto-optical disk, or a semiconductor memory, or an appropriate combination of these. Various programs, data, etc. are stored in the storage 104B. The computer-readable medium 104 is an example of a "storage device" in this disclosure.

[0054] The memory 104A is used as a working area for the processor 102 and as a storage unit that temporarily stores programs read from the storage 104B and various data. A program stored in the storage 104B is loaded into the memory 104A, and the processor 102 executes instructions of the program, causing the processor 102 to function as a means for performing various processes defined by the program. The memory 104A stores an enhancement state discrimination program 130 executed by the processor 102, various data, and the like. The enhancement state discrimination program 130 includes a learned model 30A (see FIG. 1) trained by machine learning, and causes the processor 102 to execute the processes described in FIGS. 1 and 2.

[0055] The communication interface 106 performs communication processing with an external device via a wired or wireless connection, and exchanges information with the external device. The contrast condition determination device 10 is connected to a communication line (not shown) via the communication interface 106. The communication line may be a local area network or a wide area network. The communication interface 106 can serve as a data acquisition unit that accepts input of data such as images.

[0056] The contrast condition determination apparatus 10 may further include an input device 114 and a display device 116. The input device 114 and the display device 116 are connected to the bus 110 via the input / output interface 108. The input device 114 may be, for example, a keyboard, a mouse, a multi-touch panel, or other pointing device, or a voice input device, or an appropriate combination of these.

[0057] The display device 116 is an output interface that displays various types of information. The display device 116 may be, for example, a liquid crystal display, an organic electro-luminescence (OEL) display, a projector, or an appropriate combination of these.

[0058] <Functional Configuration of the Contrast Condition Determination Device 10> 4 is a functional block diagram showing an outline of the processing functions of the contrast condition determination apparatus 10 according to the first embodiment. The processor 102 of the contrast condition determination apparatus 10 executes a contrast condition determination program 130 stored in a memory 104A, thereby functioning as a data acquisition unit 12, a time phase estimation unit 30, an integration unit 32, and an output unit 19.

[0059] The data acquisition unit 12 receives input of data to be processed. In the example of FIG. 4, the data acquisition unit 12 acquires an image IMi, which is a slice image sampled from CT data. The subscript i represents an index number that identifies multiple images, and in FIG. 4, it indicates that n different images, i=1 to n, can be input. n may be an integer equal to or greater than 2. The data acquisition unit 12 may execute a process of extracting slice images at equal intervals from the CT data, or may acquire slice images that have been sampled in advance by a processing unit (not shown) or the like.

[0060] The image IMi captured via the data acquisition unit 12 is input to the time phase estimation unit 30.

[0061] The output unit 19 is an output interface for displaying the enhancement state estimated by the integrating unit 32 and providing it to other processing units. The output unit 19 may include processing units for generating data for display and / or converting data for transmitting data to the outside. The enhancement phase estimated by the enhancement state determination device 10 may be displayed on a display device (not shown).

[0062] The contrast enhancement condition determination device 10 may be incorporated into a medical image processing device for processing medical images acquired in a medical institution such as a hospital. The processing function of the contrast enhancement condition determination device 10 may be provided as a cloud service. The contrast enhancement condition determination processing method executed by the processor 102 is an example of a "contrast enhancement condition determination method" in the present disclosure.

[0063] Second Embodiment In the first embodiment, the likelihood of belonging to each of multiple contrast phases was used as an index value for the contrast state, but in the second embodiment, an example is described in which the elapsed time since the injection of contrast agent is used as an index value for the contrast state.

[0064] The hardware configuration of the contrast condition determination device 10 according to the second embodiment may be the same as that of the first embodiment. The second embodiment will be described with respect to the differences from the first embodiment.

[0065] Fig. 5 is a conceptual diagram showing an outline of the processing by the contrast condition determination device 10 according to the second embodiment, and Fig. 6 is a functional block diagram showing an outline of the processing function of the contrast condition determination device 10 according to the second embodiment.

[0066] Here, an example of a contrast state determination device 10 is described, which uses as input a plurality of slice images sampled at equal intervals from 3D CT data of a patient who has undergone liver angiography using a CT device, estimates the number of seconds since the injection of a contrast agent for the image series of the input plurality of slice images, and determines the contrast state based on the estimated number of seconds. Hereinafter, the term "number of seconds" in this specification includes the number of seconds indicating the elapsed time since the injection of a contrast agent, unless otherwise explicitly stated.

[0067] 5 and 6, the contrast state determination device 10 includes a second estimation unit 34 that receives input of an image IM and estimates the number of seconds as the elapsed time from the injection of the contrast agent for the image series of image IM, an integration unit 36 ​​that derives an integrated elapsed time from the input multiple elapsed times, and a determination unit 38 that determines the contrast phase from the derived elapsed time. The contrast phase determined by the determination unit 38 is output from an output unit 19 as a final result.

[0068] In Figure 5, three second estimation units 34 are shown to show the processing flow when three different images IM are input, but the second estimation units 34 to which each image IM is input are the same (single) processing unit.

[0069] The plurality of slice images input to the contrast state determination device 10 are the same as those in the first embodiment. The number of seconds estimation unit 34 estimates the number of seconds from the injection of the contrast agent for the image series of the input images.

[0070] The second estimation unit 34 includes a trained model 34A trained by machine learning. When a two-dimensional image based on a three-dimensional image obtained by contrast imaging is input, the trained model 34A outputs the number of seconds since the injection of the contrast agent for the input two-dimensional image series. The trained model 34A is configured using, for example, a convolutional neural network. The trained model 34A is trained by a well-known method using pairs of two-dimensional images based on an image series acquired before or after the injection of a contrast agent into the subject and the number of seconds since the injection of the contrast agent for the image series as training data. The second estimation unit 34 inputs the input two-dimensional image into the trained model 34A and estimates the elapsed time since the injection of the contrast agent for the two-dimensional image series.

[0071] In the example shown in Figure 5, the second estimation unit 34 outputs an estimation result PS11 (an example of a "first elapsed time") of 70 seconds as the index value for image IM1, outputs an estimation result PS12 (an example of a "second elapsed time") of 75 seconds as the index value for image IM2, and outputs an estimation result PS13 of 80 seconds as the index value for image IM3.

[0072] The integration unit 36 ​​integrates the input estimation results using any statistical method to estimate the number of seconds since the injection of the contrast agent in the image series. The integration unit 36 ​​may integrate the estimation results by, for example, simple averaging, weighted averaging, or selecting one of the multiple estimation results to integrate. In the example shown in FIG. 5, the integration unit 36 ​​integrates estimation results PS11, PS12, and PS13 and outputs 72 seconds as the final estimation result PS14.

[0073] The discrimination unit 38 discriminates the enhancement state from the value of the number of seconds integrated by the integration unit 36. The discrimination unit 38 includes a conversion table 38A. The conversion table 38A may be stored in the computer-readable medium 104. The conversion table 38A associates the elapsed time from the start of contrast agent injection with the enhancement phase. For example, in the case of liver contrast imaging, the conversion table 38A associates less than 50 seconds from the start of contrast agent injection with the arterial phase, 50 seconds to less than 120 seconds with the portal vein phase, and 120 seconds or more with the equilibrium phase. The discrimination unit 38 discriminates the enhancement state using the conversion table 38A. In the example shown in FIG. 5, the final estimation result PS14 is 72 seconds, so the enhancement state is determined to be the portal vein phase.

[0074] Here, the elapsed times estimated from images IM1, IM2, and IM3 are integrated in the integration unit 36, and the final contrast state is determined from the integrated elapsed time in the discrimination unit 38. However, the second estimation unit 34 may estimate the elapsed time of the image series of image IM1 (an example of the "first elapsed time"), the elapsed time of the image series of image IM2 (an example of the "second elapsed time"), and the elapsed time of the image series of image IM3, and the discrimination unit 38 may discriminate the contrast state of the image series of image IM1 (an example of the "first contrast state"), the contrast state of the image series of image IM2 (an example of the "second contrast state"), and the contrast state of the image series of image IM3, and the integration unit 36 ​​may integrate the discriminated contrast state of the image series of images IM1, the contrast state of the image series of images IM2, and the contrast state of the image series of image IM3 to determine the final contrast state.

[0075] Third Embodiment In the third embodiment, an example will be described in which a probability distribution of the elapsed time from the injection of a contrast agent is used as an index value related to the contrast state. The hardware configuration of the contrast state determination device 10 according to the third embodiment may be the same as that of the first embodiment.

[0076] FIG. 7 is a conceptual diagram showing an overview of the processing by the contrast state determination device 10 according to the third embodiment. Here, an example of the contrast state determination device 10 that estimates the distribution of the number of seconds from the injection of the contrast agent based on a plurality of input slice images and determines the contrast state based on the estimated distribution of the number of seconds will be described.

[0077] As shown in FIG. 7, the contrast state determination device 10 includes a number-of-seconds distribution estimation unit 14 that receives an input of the image IM and estimates a probability distribution of the number of seconds (hereinafter referred to as the "number-of-seconds distribution"), an integration unit 16 that integrates a plurality of number-of-seconds distributions PD estimated from a plurality of inputs, a maximum point identification unit 18 that identifies the number of seconds at which the probability is maximized from a new distribution (hereinafter referred to as the "integrated distribution") obtained by the integration process, and a determination unit 38 that determines the contrast phase by associating the identified number of seconds with the contrast phase. The contrast phase converted by the determination unit 38 is output as the final result.

[0078] In FIG. 7, three number-of-seconds distribution estimation units 14 are shown to illustrate the processing flow when three different images IM are input. However, the number-of-seconds distribution estimation unit 14 for each input image IM is the same (single) processing unit.

[0079] FIG. 8 is an explanatory diagram showing an example 1 of the processing in the number-of-seconds distribution estimation unit 14. The number-of-seconds distribution estimation unit 14 includes a regression estimation unit 22 and a variable conversion unit 24. The regression estimation unit 22 includes a learned model trained by machine learning so as to receive an input of the image IM and output an estimated value Oa of the number of seconds and a score value Ob indicating the certainty (confidence) of the estimated value Oa. The learned model as the regression model applied to the regression estimation unit 22 is configured using, for example, a convolutional neural network. The numerical range of the estimated value Oa of the number of seconds output from the regression estimation unit 22 may be "-∞ < Oa < ∞", and the numerical range of the score value Ob of the certainty may be "-∞ < Ob < ∞". Note that the regression model is not limited to a CNN, and various machine learning models can be applied.

[0080] The variable transformation unit 24 transforms the estimated value Oa of the number of seconds and the score value Ob of its likelihood in accordance with the following equations (1) and (2), respectively, to generate parameters μ and b of the probability distribution model. μ = Oa (1) b=1 / log(1+exp(-Ob)) (2)

[0081] The function in equation (2) is an example of a mapping that converts the likelihood score value Ob to a value b in the positive domain. FIG. 9 is a graph of the function y=1 / log(1+exp(-x)) used in the variable conversion in equation (2). The parameter μ is an example of a "first parameter" in this disclosure. The parameter b is an example of a "second parameter" in this disclosure.

[0082] In the third embodiment, a Laplace distribution is applied as a probability distribution model for the number of seconds distribution. The Laplace distribution is expressed by the function of the following equation (3).

[0083]

number

[0084] The reason for converting the likelihood score Ob to a positive value b is related to the application of the Laplace distribution as a probability distribution model for the number of seconds distribution. If the parameter b is a negative value (b<0), the Laplace distribution does not hold as a probability distribution, so it is necessary to ensure that the parameter b is a positive value (b>0).

[0085] FIG. 10 shows an example graph of the number of seconds distribution estimated by the parameters μ and b estimated by the number of seconds distribution estimation unit 14. Note that the position indicated by the dashed line GT in the figure corresponds to the correct number of seconds (correct number of seconds). Estimating a pair of an estimated value Oa and its likelihood score Ob from an input image IM essentially corresponds to estimating the number of seconds distribution. The estimated number of seconds Oa is an example of a "random variable" in this disclosure.

[0086] 11 is an explanatory diagram showing an example of processing in the integrating unit 16 and maximum point identifying unit 18. For simplicity of explanation, an example is shown in which two second distributions estimated by the second distribution estimating unit 14 are integrated, but the same applies when three or more second distributions are integrated.

[0087] Graph GD1 shown in the upper left of FIG. 11 is an example of a second distribution (probability distribution P1) represented by parameters μ1 and b1 estimated by the second distribution estimation unit 14 for input image IM1 (not shown in FIG. 11). Parameter μ1 is an example of a "first elapsed time," parameter b1 is an example of a "first likelihood of the first elapsed time," and probability distribution P1 is an example of a "first probability distribution model." The integration unit 16 takes the logarithm of the estimated second distribution, converts it into a logarithmic probability density, and sums and integrates multiple logarithmic probability densities. This corresponds to calculating the product of probabilities over the same number of seconds.

[0088] Graph GL1 in FIG. 11 is an example of a logarithmic probability density logP1 obtained by taking the logarithm of the probability distribution P1. Graph GD2 shown in the lower left of FIG. 11 is an example of a second distribution (probability distribution P2) represented by parameters μ2 and b2 estimated by the second distribution estimation unit 14 for the input of image IM2 (not shown in FIG. 11). Parameter μ2 is an example of a "second elapsed time," parameter b2 is an example of a "second likelihood of the second elapsed time," and probability distribution P2 is an example of a "second probability distribution model." Furthermore, parameters μ1 and μ2 are examples of "multiple elapsed times," parameters b1 and b2 are examples of "multiple likelihoods," and probability distributions P1 and P2 are examples of "multiple probability distribution models." Graph GL2 in FIG. 11 is an example of a logarithmic probability density logP1 obtained by taking the logarithm of the probability distribution P2. logP2 This is an example.

[0089] 11 is an example of a joint logarithmic probability density obtained by integrating the logarithmic probability density logP1 and the logarithmic probability density logP2, and is an example of the "product of the first probability distribution model and the second probability distribution model." The distribution shown in the graph GLS is an example of the "integrated distribution" in the present disclosure.

[0090] The maximum point identification unit 18 identifies the value x of the parameter μ that maximizes the logarithmic probability from the integrated logarithmic probability density. The processing in the maximum point identification unit 18 can be expressed by the following equation (4).

[0091]

number

[0092] The target function of arg min (the part after Σ) shown on the right side of the equal sign in the second line of equation (4) corresponds to the loss function during training in machine learning, which will be described later. The right side of the equal sign in the third line corresponds to the weighted median equation. The parameter bi, which corresponds to the weight during integration, dynamically changes depending on the output of the regression estimation unit 22.

[0093] In the case of the integrated logarithmic probability density shown in graph GLS in Figure 11, the input value (maximum point) at which the joint logarithmic probability is maximized is μ1, and μ1 is selected as the final estimation result (final result). Note that μ1 is the estimation result for image IM1 among the multiple slice images input. In Figure 11, the calculation is performed by converting from the second distribution to the logarithmic probability density, but in essence, the process considers the joint probability of multiple second distributions (probability distributions) estimated from multiple different inputs, and derives the value at which the joint probability is maximized as the final result.

[0094] By adopting the Laplace distribution as the probability distribution model, the integrated distribution (joint probability distribution) takes the form of a weighted median. Therefore, even if some of the multiple estimation results are significantly out of sync due to artifacts or other factors, the influence of the outliers can be suppressed, making it possible to obtain highly accurate estimations.

[0095] {Machine learning method example 1} 12 is a diagram illustrating an example of a machine learning method for generating a regression model to be applied to the seconds distribution estimation unit 14. The training data used for machine learning includes an image TIM as input data and correct answer data (teacher signal t) corresponding to the input. The image TIM may be a slice image constituting an image series of 3D CT data, and the teacher signal t may be a value indicating the number of seconds (ground truth) from the injection of a contrast agent when the series to which the slice image belongs was captured.

[0096] For example, multiple training data sets are generated by linking each corresponding teacher signal t to all slices in an image series. "Linking" can also be referred to as "association" or "connection." "Training" is synonymous with "learning." The same teacher signal t may be linked to slices in the same image series. In other words, teacher signals t may be linked on an image series-by-image series basis. Similarly, multiple training data sets are generated by linking each corresponding teacher signal t to each slice in multiple image series. The collection of multiple training data sets generated in this way is used as a training dataset.

[0097] The learning model 20 (an example of a "second learning model") is configured using CNN. The learning model 20 is used in combination with a variable transformation unit 24. Note that the variable transformation unit 24 may be integrated into the learning model 20.

[0098] When an image TIM read from the training data set is input to the learning model 20, an estimate Oa of the number of seconds and a score Ob of its likelihood are output from the learning model 20. The estimate Oa and score Ob are transformed by the variable transformation unit 24 into parameters μ and b of a probability distribution model.

[0099] The loss function L used during training is defined by the following equation (5).

[0100]

number

[0101] As shown in the lower part of FIG. 12, the sum of the losses for all slices in the same image series is given by the following equation (6):

[0102]

number

[0103] The subscript i is an index that identifies each slice. The sum of losses expressed by Equation (6) is used to apply backpropagation, and the learning model 20 is trained using stochastic gradient descent, similar to normal CNN training (the parameters of the learning model 20 are updated). The sum of losses calculated by Equation (6) is an example of the "calculation result of the loss function" in this disclosure. By training the learning model 20 using multiple training data including multiple image series, the parameters of the learning model 20 are optimized, and a trained model is obtained. The trained model thus obtained is applied as a regression model to the number-of-seconds distribution estimation unit 14.

[0104] Figure 13 is an explanatory diagram of the loss function used during training. The loss function is a negative logarithmic likelihood, and the formula used for regression estimation is directly optimized through learning. Learning maximizes the logarithmic likelihood of the training signal t in seconds. The graph GRμ in Figure 13 shows the graph for the parameter μ of the loss function shown in equation (5). The gradient of graph GRμ with respect to the parameter μ is stable.

[0105] On the other hand, the graph for parameter b of the loss function shown in equation (5) is graph GRb in Figure 13. Graph GRb has an unstable gradient with respect to parameter b. In the region where the value of b is small, 1 / b is dominant, while in the region where the value of b is large, logb is dominant.

[0106] A graph GRb with an unstable gradient is transformed into a graph GROb by performing a variable transformation on the parameter b using a function such as b=1 / softplus(-Ob). The softplus function is defined as softplus(x)=log(1+exp(x)). The function used for the variable transformation of the parameter b is a function that asymptotically approaches -1 / x as x→-∞ and asymptotically approaches exp(x) as x→∞, and the gradient instability can be canceled using such a function.

[0107] The machine learning method for the learning model 20 described using Figures 12 and 13 is an example of a "method for generating a trained model" in the present disclosure.

[0108] <Functional Configuration of the Contrast Condition Determination Device 10> 14 is a functional block diagram showing an outline of the processing functions of the contrast enhancement state determination device 10 according to the third embodiment. The processor 102 of the contrast enhancement state determination device 10 executes a contrast enhancement state determination program 130 stored in a memory 104A, thereby functioning as a data acquisition unit 12, a number-of-seconds distribution estimation unit 14, an integration unit 16, a maximum point identification unit 18, a determination unit 38, and an output unit 19.

[0109] The data acquisition unit 12 receives input of data to be processed. In the example of FIG. 14, the data acquisition unit 12 acquires an image IMi, which is a slice image sampled from CT data. The subscript i represents an index number for identifying multiple images, and in FIG. 14, it is indicated that n different images, i=1 to n, may be input. n may be an integer equal to or greater than 2. The data acquisition unit 12 may execute a process of extracting slice images at equal intervals from the CT data, or may acquire slice images sampled in advance by a processing unit (not shown) or the like.

[0110] The images IMi captured via the data acquisition unit 12 are input to the regression estimation unit 22 of the seconds distribution estimation unit 14. The regression estimation unit 22 outputs a pair of an estimated value Oa of the number of seconds from each input image IMi and a score value Ob indicating its likelihood.

[0111] The estimated value Oa output from the regression estimation unit 22 is converted into a parameter μi of a probability distribution model in the variable transformation unit 24, and the likelihood score value Ob output from the regression estimation unit 22 is converted into a parameter bi of the probability distribution model in the variable transformation unit 24. The probability distribution Pi of the number of seconds is estimated using these two parameters μi and bi.

[0112] By inputting multiple images IMi (i = 1 to n) in the same series, a set of an estimated value Oa and a score value Ob is estimated for each image IMi, and converted into a set of parameters μi, bi, to estimate a probability distribution Pi of the number of seconds. The multiple sets of estimated values ​​Oa and score values ​​Ob estimated from each image IMi are an example of "multiple sets of estimation results" in the present disclosure.

[0113] The integration unit 16 performs processing to integrate multiple probability distributions Pi obtained by inputting multiple images IMi. In Fig. 14, a logarithm conversion unit 26 takes the logarithm of the probability distribution Pi and converts it into a logarithmic probability density logPi, and an integrated distribution generation unit 28 calculates the sum of the logarithmic probability densities logPi to obtain an integrated distribution.

[0114] The maximum point identification unit 18 identifies the value of the number of seconds (maximum point) at which the probability is maximum from the integrated distribution, and outputs the identified value of the number of seconds. Note that the maximum point identification unit 18 may be configured to be incorporated into the integration unit 16.

[0115] The discrimination unit 38 discriminates the enhancement state from the value of the number of seconds identified by the maximum point identification unit 18. The discrimination unit 38 includes a conversion table 38A. The conversion table 38A is stored in the computer-readable medium 104. The conversion table 38A associates the elapsed time from the start of contrast agent injection with the enhancement phase. For example, in the case of liver contrast imaging, the conversion table 38A associates less than 50 seconds from the start of contrast agent injection with the arterial phase, 50 seconds to less than 120 seconds with the portal venous phase, and 120 seconds or more with the equilibrium phase. The discrimination unit 38 discriminates the enhancement state using the conversion table 38A.

[0116] When estimating the number of seconds from multiple images and averaging them as the final result, the result may be poor if the number of seconds is output from images that are not suitable for estimation. According to the third embodiment, by outputting a confidence level along with the number of seconds and weighting the result according to the confidence level, the estimation is less susceptible to outliers and becomes more robust.

[0117] Fourth Embodiment In the third embodiment, a Laplace distribution is used as the probability distribution model of the number of seconds distribution, but other probability distribution models may be applied. In the fourth embodiment, an example will be described in which a Gaussian distribution is used instead of the Laplace distribution. In the fourth embodiment, the processing contents of each processing unit, that is, the number of seconds distribution estimation unit 14, the integration unit 16, and the maximum point identification unit 18, differ from those in the third embodiment.

[0118] 15 is an explanatory diagram showing a second example of the processing in the second number distribution estimation unit 14 of the contrast condition determination device 10 according to the fourth embodiment. The processing in FIG. 15 is applied instead of the processing described in FIG. 8.

[0119] The variable transformation unit 24 in the fourth embodiment transforms the likelihood score Ob into a parameter σ using the following equation (7) instead of equation (2). 2 Convert to. σ 2 =1 / log(1+exp(-Ob)) (7)

[0120] σ 2 plays the role of certainty. σ 2 corresponds to the variance and σ corresponds to the standard deviation.

[0121] The Gaussian distribution is expressed by the function of the following equation (8).

[0122]

number

[0123] The score value Ob is a positive value (σ 2 ) is the same as in the third embodiment. 2If is a negative value, the Gaussian distribution does not hold as a probability distribution, so the parameter σ 2 is a positive value (σ 2 > 0).

[0124] FIG. 16 shows the parameters μ and σ estimated by the second distribution estimation unit 14. 2 1 shows an example of a graph of the distribution of seconds estimated by

[0125] 17 is an explanatory diagram showing an example of processing in the integrating unit 16 and maximum point identifying unit 18 of the contrast condition determining device 10 according to the fourth embodiment. Here, an example is shown in which two second distributions estimated by the second distribution estimating unit 14 are integrated.

[0126] The graph GD1g shown in the upper left of FIG. 17 is a graph of the parameters μ1 and σ estimated by the second distribution estimation unit 14 in FIG. 2 This is an example of a second distribution (probability distribution P1) represented by 1. The integration unit 16 takes the logarithm of the estimated second distribution, converts it into a logarithmic probability density, and integrates the sum of multiple logarithmic probability densities. This corresponds to finding the product of probabilities over the same number of seconds.

[0127] 17 is an example of a logarithmic probability density logP1 obtained by taking the logarithm of the probability distribution P1. A graph GD2g shown in the lower left of FIG. ... 2 17 is an example of a logarithmic probability density function (probability distribution P2) represented by 2. Graph GL2g in FIG. 17 is a logarithmic probability density function obtained by taking the logarithm of probability distribution P2. logP2 This is an example.

[0128] The graph GLSg shown on the far right in FIG. 17 is an example of a joint logarithmic probability density obtained by integrating the logarithmic probability density logP1 and the logarithmic probability density logP2.

[0129] The maximum point identification unit 18 identifies the value x at which the logarithmic probability is maximized from the integrated simultaneous logarithmic probability density. The processing in the maximum point identification unit 18 can be expressed by the following equation (9).

[0130]

number

[0131] The target function of arg min (the part after Σ) shown on the right side of the equal sign in the second line of equation (9) corresponds to the loss function during training in machine learning, which will be described later. Also, the right side of the equal sign in the third line corresponds to the weighted average equation.

[0132] In the case of the integrated logarithmic probability density shown in graph GLSg of FIG. 17, the input value (maximum point) x at which the logarithmic probability is maximized is selected as the final estimation result (final result).

[0133] {Machine learning method example 2} Fig. 18 is an explanatory diagram that shows an outline of an example of a machine learning method for generating a regression model that is applied to the number-of-seconds distribution estimation unit 14 in the fourth embodiment. The training data used for learning may be the same as in the third embodiment. Regarding Fig. 18, differences from Fig. 12 will be described.

[0134] When an image TIM read from the training data set is input to the learning model 20, an estimated value Oa of the number of seconds and a score value Ob of its likelihood are output from the learning model 20. The estimated value Oa and the score value Ob of the likelihood are converted by the variable transformation unit 24 into parameters μ and σ of the probability distribution model. 2 The variables are transformed into

[0135] The loss function L during training is defined by the following equation (10):

[0136]

number

[0137] Figure 1 8 As shown in the bottom row of the figure, if we take the sum of the losses for all slices in the same image series, we get the following equation (11):

[0138]

number

[0139] The error backpropagation method is applied using the sum of losses expressed by Equation (11), and the learning model 20 is trained using the stochastic gradient descent method, similar to normal CNN training. By training the learning model 20 using multiple training data including multiple image series, the parameters of the learning model 20 are optimized and a trained model is obtained. The trained model thus obtained is applied to the number-of-seconds distribution estimation unit 14.

[0140] Variation 1 In the first to fourth embodiments, images IM1, IM2, and IM3, which are slice images (tomographic images) obtained by cutting slices at equal intervals from three-dimensional CT data, are used as input, but the images to be processed are not limited to these. For example, as shown in Fig. 19, two-dimensional image IM11 (an example of a "first two-dimensional image") containing information on image IM1 (an example of a "first slice image"), two-dimensional image IM12 (an example of a "second two-dimensional image") containing information on image IM2 (an example of a "second slice image"), and two-dimensional image IM13 containing information on image IM3 may be used. The two-dimensional image may be the tomographic image TGimg itself, a Maximum Intensity Projection (MIP) image MIPimg formed at equal intervals, or an average image AVEimg generated from multiple slice images.

[0141] Variation 2 The input to the time phase estimation unit 30, the number of seconds estimation unit 34, and the number of seconds distribution estimation unit 14 may be a combination of multiple types of data elements. For example, as shown in Fig. 20, at least one of a slice image, an MIP image, and an average image, which are partial images of CT data from the same series, can be used as input, and a combination of these multiple image types may be input to the time phase estimation unit 30, the number of seconds estimation unit 34, and the number of seconds distribution estimation unit 14. For example, a combination of an average image and an MIP image may be input to the number of seconds distribution estimation unit 14 to estimate the number of seconds distribution. The MIP image and the average image are examples of generated images generated from partial images of three-dimensional CT data.

[0142] <<Example of medical information system configuration>> FIG. 21 is a block diagram showing an example of the configuration of a medical information system 200 including a medical image processing device 220. The contrast condition determination device 10 described as the first to fourth embodiments is incorporated into the medical image processing device 220, for example. The medical information system 200 is a computer network established in a medical institution such as a hospital. The medical information system 200 includes a modality 230 that captures medical images, a DICOM server 240, the medical image processing device 220, an electronic medical record system 244, and a viewer terminal 246, and these elements are connected via a communication line 248. The communication line 248 may be an in-house communication line within the medical institution. Furthermore, a part of the communication line 248 may be a wide-area communication line.

[0143] Specific examples of the modality 230 include a CT device 231, an MRI (Magnetic Resonance Imaging) device 232, an ultrasound diagnostic device 233, a PET (Positron Emission Tomography) device 234, an X-ray diagnostic device 235, an X-ray fluoroscopic diagnostic device 236, and an endoscope device 237. There may be various combinations of types of modality 230 connected to the communication line 248 depending on the medical institution.

[0144] The DICOM server 240 is a server that operates in accordance with the DICOM specifications. The DICOM server 240 is a computer that stores and manages various data, including images captured using the modality 230, and is equipped with a large-capacity external storage device and a database management program. The DICOM server 240 communicates with other devices via a communication line 248, sending and receiving various data, including image data. The DICOM server 240 receives various data, including image data generated by the modality 230, via the communication line 248, and stores and manages the data on a recording medium, such as a large-capacity external storage device. The storage format of the image data and communication between devices via the communication line 248 are based on the DICOM protocol.

[0145] The medical image processing device 220 can acquire data from the DICOM server 240 or the like via a communication line 248. The medical image processing device 220 performs image analysis and other various processes on medical images captured by the modality 230. In addition to the processing functions of the contrast enhancement state determination device 10, the medical image processing device 220 may be configured to perform various analytical processes such as computer-aided diagnosis (CAD), for example, a process for recognizing a lesion area from an image, a process for identifying a disease name or other classification, or a segmentation process for recognizing an organ or other area. The medical image processing device 220 can also send processing results to the DICOM server 240 and the viewer terminal 246. The processing functions of the medical image processing device 220 may be installed in the DICOM server 240 or the viewer terminal 246.

[0146] Various types of data stored in the database of the DICOM server 240 and various types of information including the processing results generated by the medical image processing device 220 can be displayed on the viewer terminal 246.

[0147] The viewer terminal 246 is a PAC S(The viewer terminal 246 is a terminal for viewing images called a DICOM (Digital Image Archiving and Communication Systems) viewer or a DICOM viewer. A plurality of viewer terminals 246 can be connected to the communication line 248. The form of the viewer terminal 246 is not particularly limited, and may be a personal computer, a workstation, a tablet terminal, or the like.

[0148] About the programs that run computers A program that causes a computer to realize the processing functions of the contrast state discrimination device 10 can be recorded on a computer-readable medium such as an optical disk, a magnetic disk, a semiconductor memory, or other tangible non-transitory information storage medium, and the program can be provided through this information storage medium.

[0149] In addition, instead of providing the program by storing it on such a tangible, non-transitory computer-readable medium, it is also possible to provide the program signal as a download service using a telecommunications line such as the Internet.

[0150] Furthermore, some or all of the processing functions of the contrast condition determination device 10 may be realized by cloud computing, and may also be provided as a SasS (Software as a Service) service.

[0151] <<Hardware configuration of each processing unit>> The hardware structure of the processing units that execute various processes in the contrast state discrimination device 10, such as the data acquisition unit 12, the number-of-seconds distribution estimation unit 14, the integration unit 16, the maximum point identification unit 18, the output unit 19, the regression estimation unit 22, the variable transformation unit 24, the logarithmic transformation unit 26, the integrated distribution generation unit 28, the time phase estimation unit 30, and the integration unit 32, is, for example, various processors as shown below.

[0152] Various types of processors include CPUs, which are general-purpose processors that execute programs and function as various processing units, GPUs, which are processors specialized for image processing, programmable logic devices (PLDs), such as FPGAs (Field Programmable Gate Arrays), which are processors whose circuit configuration can be changed after manufacturing, and dedicated electrical circuits, such as ASICs (Application Specific Integrated Circuits), which are processors with circuit configurations designed specifically to execute specific processes.

[0153] A single processing unit may be configured with one of these various processors, or may be configured with two or more processors of the same or different types. For example, a single processing unit may be configured with multiple FPGAs, or a combination of a CPU and an FPGA, or a combination of a CPU and a GPU. Alternatively, multiple processing units may be configured with a single processor. Examples of multiple processing units configured with a single processor include, first, a configuration in which a single processor is configured with a combination of one or more CPUs and software, as typified by client and server computers, and this processor functions as multiple processing units. Second, a configuration in which a processor is used to realize the functions of an entire system including multiple processing units on a single IC (Integrated Circuit) chip, as typified by a system-on-chip (SoC). In this way, the various processing units are configured with one or more of the above-mentioned various processors as a hardware structure.

[0154] Furthermore, the hardware structure of these various processors is, more specifically, an electric circuit made up of a combination of circuit elements such as semiconductor elements.

[0155] "others" The present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the gist of the technical idea of ​​the present disclosure. [Explanation of symbols]

[0156] 10. Contrast state determination device 12 Data Acquisition Section 14. Number of seconds distribution estimation unit 16 Integration Department 18 Maximum point identification part 19 Output section 20 Learning Model 22 Regression estimation section 24 Variable transformation section 26 Logarithmic conversion section 28 Integrated distribution generator 30 Time phase estimation section 30A Pre-trained model 32 Integration Department 34 Seconds Estimation Section 34A Pre-trained model 36 Integration Department 38 Discrimination part 38A Conversion Table 102 processors 104 Computer-readable medium 104A Memory 104B Storage 106 Communication Interface 108 Input / Output Interface 110 Bus 114 Input Device 116 Display device 130 Contrast Status Discrimination Program 200 Medical Information Systems 220 Medical image processing equipment 230 Modalities 231 CT device 232 MRI machine 233 Ultrasound diagnostic equipment 234 PET equipment 235 X-ray diagnostic equipment 236 X-ray fluoroscopy equipment 237 Endoscopic Devices 240 DICOM Server 244 Electronic Medical Record System 246 viewer terminal 248 communication lines GD1 graph GD1g graph GD2 graph GD2g graph GL1 graph GL1g graph GL2 graph GL2g graph GLS graph GLSg graph GRb graph GRμ graph GROb graph IM Image IM1, IM2, IM3, IMi, IMn images TIM Images IM11, IM12, IM13 2D images Oa estimate Ob score value P1, P2, Pi probability distributions PD seconds distribution PS1, PS2, PS3, PS11, PS12, PS13, PS14 Estimation results SB patient SBP1, SBP2, SBP3 position S1~S5 Steps of contrast state determination method

Claims

1. at least one processor; at least one memory storing instructions for execution by said at least one processor; Equipped with The at least one processor acquiring a plurality of two-dimensional images each including information on slice images at different positions of the subject from a first image series taken before or after injection of a contrast agent into the subject; an index value relating to a contrast state from each of the plurality of two-dimensional images; determining an enhancement state of the first image series based on the estimated plurality of index values; Contrast state determination device.

2. the index value is a likelihood that the index value belongs to each of a plurality of contrast states, The at least one processor deriving an index value that integrates the plurality of index values; determining an enhancement state of the first image series based on the integrated index value; The imaging condition determination device according to claim 1 .

3. a first learning model that, when a two-dimensional image based on an image series captured before or after injection of a contrast agent into a subject is input, outputs a likelihood that the image belongs to each of a plurality of contrast states; The at least one processor inputting the plurality of two-dimensional images into the first learning model to estimate the plurality of index values; The imaging condition determination device according to claim 2 .

4. the index value is the elapsed time since the contrast agent was injected into the subject; The at least one processor deriving an elapsed time by integrating the elapsed times that are the plurality of index values; determining a contrast state of the first image series based on the integrated elapsed time; The imaging condition determination device according to claim 1 .

5. the index value is an elapsed time from the injection of the contrast agent into the subject and a likelihood of the elapsed time; The at least one processor deriving an elapsed time by integrating the plurality of elapsed times that are the plurality of index values ​​based on the plurality of probabilities that are the plurality of index values; determining a contrast state of the first image series based on the integrated elapsed time; The imaging condition determination device according to claim 1 .

6. The at least one processor deriving the integrated elapsed time based on the product of a plurality of probability distribution models each having the elapsed time and the likelihood as parameters; The imaging condition determination device according to claim 5 .

7. a second learning model that, when a two-dimensional image based on an image series captured before or after injection of a contrast agent into a subject is input, outputs an elapsed time since injection of the contrast agent into the subject and a likelihood of the elapsed time; The at least one processor inputting the plurality of two-dimensional images into the second learning model to estimate the plurality of elapsed times and the plurality of likelihoods; The imaging condition determination device according to claim 5 .

8. a conversion table in which the elapsed time since the injection of the contrast agent into the subject is associated with the contrast state; The at least one processor determining a contrast state of the first image series based on the integrated elapsed time and the conversion table; The imaging condition determination device according to claim 4.

9. a conversion table in which the elapsed time since the injection of the contrast agent into the subject is associated with the contrast state; The at least one processor estimating a plurality of elapsed times from the injection of the contrast agent into the subject as the plurality of index values; determining a plurality of contrast states of the first image series based on the plurality of elapsed times and the conversion table; determining an enhancement state of the first image series based on the plurality of enhancement states; The imaging condition determination device according to claim 1 .

10. the two-dimensional image is at least one of the slice image, a maximum intensity projection (MIP) image of the plurality of slice images, and an average image of the plurality of slice images; The imaging condition determination device according to any one of claims 1 to 9.

11. the first image series are three-dimensional images including a liver of the subject; the enhancement state includes at least one of a non-enhanced, arterial phase, portal venous phase, and equilibrium phase; The imaging condition determination device according to claim 1 .

12. the first image series are three-dimensional images including the kidneys of the subject; The contrast-enhanced state includes at least one of a non-contrast, corticomedullary phase, parenchymal phase, and excretory phase. The imaging condition determination device according to claim 1 .

13. acquiring a plurality of two-dimensional images from a first image series taken before or after injection of a contrast agent into the subject, each of the two-dimensional images including information on slice images at different positions of the subject; estimating an index value relating to a contrast enhancement state from each of the plurality of two-dimensional images; determining an enhancement state of the first image series based on the estimated index values; A contrast state determination method comprising:

14. A program for causing a computer to execute the contrast state determination method according to claim 13.

15. A non-transitory computer-readable recording medium on which the program according to claim 14 is recorded.

Images