WO2022264757A1 - 医用画像診断システム、医用画像診断方法及びプログラム - Google Patents
医用画像診断システム、医用画像診断方法及びプログラム Download PDFInfo
- Publication number
- WO2022264757A1 WO2022264757A1 PCT/JP2022/021224 JP2022021224W WO2022264757A1 WO 2022264757 A1 WO2022264757 A1 WO 2022264757A1 JP 2022021224 W JP2022021224 W JP 2022021224W WO 2022264757 A1 WO2022264757 A1 WO 2022264757A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- medical image
- determination
- normal
- processor
- image
- Prior art date
Links
- 238000002405 diagnostic procedure Methods 0.000 title claims abstract description 9
- 230000005856 abnormality Effects 0.000 claims abstract description 52
- 238000000034 method Methods 0.000 claims description 41
- 230000002159 abnormal effect Effects 0.000 claims description 27
- 238000012805 post-processing Methods 0.000 claims description 16
- 230000035945 sensitivity Effects 0.000 claims description 13
- 210000000056 organ Anatomy 0.000 claims description 7
- 238000002059 diagnostic imaging Methods 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 abstract description 23
- 238000002591 computed tomography Methods 0.000 description 75
- 238000012545 processing Methods 0.000 description 55
- 230000003902 lesion Effects 0.000 description 46
- 238000013473 artificial intelligence Methods 0.000 description 35
- 238000001514 detection method Methods 0.000 description 35
- 238000004195 computer-aided diagnosis Methods 0.000 description 31
- 201000010099 disease Diseases 0.000 description 28
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 28
- 238000010586 diagram Methods 0.000 description 12
- 239000003550 marker Substances 0.000 description 10
- 230000010354 integration Effects 0.000 description 9
- 210000004072 lung Anatomy 0.000 description 6
- 206010058467 Lung neoplasm malignant Diseases 0.000 description 4
- 201000005202 lung cancer Diseases 0.000 description 4
- 208000020816 lung neoplasm Diseases 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 238000012790 confirmation Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000002595 magnetic resonance imaging Methods 0.000 description 2
- 238000002600 positron emission tomography Methods 0.000 description 2
- 208000004930 Fatty Liver Diseases 0.000 description 1
- 206010016654 Fibrosis Diseases 0.000 description 1
- 206010019708 Hepatic steatosis Diseases 0.000 description 1
- 206010035664 Pneumonia Diseases 0.000 description 1
- 208000035977 Rare disease Diseases 0.000 description 1
- 238000012062 charged aerosol detection Methods 0.000 description 1
- 230000007882 cirrhosis Effects 0.000 description 1
- 208000019425 cirrhosis of liver Diseases 0.000 description 1
- 238000001360 collision-induced dissociation Methods 0.000 description 1
- 238000011960 computer-aided design Methods 0.000 description 1
- 208000031513 cyst Diseases 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 208000010706 fatty liver disease Diseases 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 210000004185 liver Anatomy 0.000 description 1
- 201000007270 liver cancer Diseases 0.000 description 1
- 208000014018 liver neoplasm Diseases 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 201000003144 pneumothorax Diseases 0.000 description 1
- 238000002601 radiography Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 231100000240 steatosis hepatitis Toxicity 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
- G06T7/0014—Biomedical image inspection using an image reference approach
- G06T7/0016—Biomedical image inspection using an image reference approach involving temporal comparison
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
- A61B6/5211—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data
- A61B6/5217—Devices using data or image processing specially adapted for radiation diagnosis involving processing of medical diagnostic data extracting a diagnostic or physiological parameter from medical diagnostic data
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
- A61B6/02—Arrangements for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
- A61B6/03—Computed tomography [CT]
- A61B6/032—Transmission computed tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/24—Indexing scheme for image data processing or generation, in general involving graphical user interfaces [GUIs]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20076—Probabilistic image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Definitions
- the present invention relates to a medical image diagnostic system, a medical image diagnostic method, and a program, and more particularly to technology for supporting medical image diagnosis.
- AI Artificial Intelligence
- Patent Document 1 regarding abnormal shadow candidates detected from a medical image, only detection information of abnormal shadow candidates suspected of being true positive abnormal shadows and/or abnormal shadow candidates with low visibility is provided, Disclosed is a medical image processing system that prevents doctors from overlooking and improves the efficiency of interpretation work.
- a medical checkup is being conducted to check the health status of the subject.
- examinees are mainly healthy persons, and the purpose of diagnosis, target organs, and target diseases are limited.
- Physicians need to judge whether there is a real abnormality, diagnose the possible disease name, and prepare a detailed report for areas that may be abnormal in the medical images obtained in the medical examination. These tasks are important.
- the doctor must also confirm the image without abnormal areas. For example, in the case of health checkups mainly for young people, it is unlikely that there will be patients with abnormalities, so the doctor must check many "images with no abnormalities", which is a heavy burden on the doctor.
- lesion detection AI In contrast, it is conceivable to support image diagnosis with lesion detection AI.
- a lesion detection AI is usually created for each disease, and there are a huge number of types of diseases, making it difficult to create an AI that corresponds to all diseases.
- the amount of learning data for rare diseases is small, making it difficult to create a lesion detection AI.
- lesion detection AI cannot be created for unknown diseases, doctors are also responsible for diagnosing unknown diseases.
- the number of lesion detection AIs increases and the accuracy of each increases, there is a problem that there is a limit to the final output accuracy of lesion detection AIs.
- the present invention has been made in view of such circumstances, and provides a medical image diagnostic system, a medical image diagnostic method, and a medical image diagnostic method that reduce the burden on a doctor when image diagnosis is performed on a large number of medical images, such as in a medical examination.
- the purpose is to provide a program.
- One aspect of a medical imaging diagnostic system for achieving the above object comprises at least one processor and at least one memory storing instructions for causing the at least one processor to execute, the at least one processor comprising: performing a first determination for determining whether or not a medical image obtained by photographing a subject is normal, and if it is determined in the first determination that the medical image is not normal; It is a medical image diagnostic system that performs a second determination to determine whether or not there is an abnormality.
- a normal medical image is, for example, a case where the medical image can be said to be an image of a healthy person.
- a healthy person is a healthy person, eg, a person who is free from disease, disease, and lesions. According to this aspect, it is possible to reduce the burden on a doctor when performing image diagnosis on a large number of medical images.
- At least one processor performs a third determination to determine the presence or absence of abnormality from the medical image when the medical image is determined to be normal in the first determination.
- At least one processor preferably makes the third determination at a different timing than the second determination.
- At least one processor performs a first determination using the probability that the medical image is normal, and if the first determination determines that the medical image is normal with a probability lower than a first threshold, It is preferable to make the determination of 3.
- the at least one processor performs a first determination using a first trained model that outputs whether or not the medical image is normal when a medical image is input, and performs a first determination of a medical image determined to be abnormal in the third determination.
- the images are used to retrain the first trained model.
- the first trained model is preferably a trained model trained using a combination of abnormal medical images and normal medical images, and labels indicating the presence or absence of an abnormality as a learning data set.
- At least one processor performs a fourth determination for determining the presence or absence of an abnormality from the medical image when the second determination determines that there is no abnormality, and in the fourth determination, the sensitivity is higher than the sensitivity in the second determination. It is preferable to determine the presence or absence of an abnormality from a medical image with relatively high sensitivity.
- the at least one processor determines, in a first case, that the medical image is normal in the first determination, that the medical image is not normal in the first determination, and that there is an abnormality in the second determination.
- the first case and the second case It is preferable to display the discrimination result of the medical image on the display in different manners for the case and the third case.
- the at least one processor causes the display to display the medical image discrimination results in different manners in the second case and the third case.
- the at least one processor preferably performs different post-processing on the medical image in the first case, the second case and the third case.
- At least one processor preferably performs the first discrimination and the second discrimination for each organ of the subject from the medical image.
- the at least one processor makes the second determination using a second trained model that outputs abnormalities in the medical image when the medical image is input.
- One aspect of the medical image diagnostic method for achieving the above object comprises: a first determination step of determining whether or not a medical image obtained by photographing a subject is normal; and a second determination step of determining whether or not there is an abnormality in the medical image if the step determines that the medical image is not normal. According to this aspect, it is possible to reduce the burden on a doctor when performing image diagnosis on a large number of medical images.
- One aspect of the program for achieving the above object is a program for causing a computer to execute the above medical image diagnosis method.
- a computer-readable non-transitory storage medium in which this program is recorded may also be included in this embodiment.
- FIG. 1 is a block diagram of a medical image diagnostic system according to this embodiment.
- FIG. 2 is a flow chart showing a medical image diagnosis method.
- FIG. 3 is a process diagram showing a medical image diagnosis method.
- FIG. 4 is a diagram showing a display form.
- FIG. 5 is a diagram showing a display form.
- FIG. 6 is a diagram showing a display form.
- FIG. 7 is a diagram showing a display form.
- the medical image diagnostic system determines whether there is an abnormality in a large number of medical images such as a health checkup, and determines whether the medical image is normal (whether the image is of a healthy person). This reduces the burden on the doctor when performing the determination of
- FIG. 1 is a block diagram of a medical image diagnostic system 10 according to this embodiment.
- the medical image diagnostic system 10 includes a modality 12, an image storage server 14, each company's CAD (Computer-Aided Diagnosis) processing server 18, a results integration CAD processing server 16, a PACS (Picture Archiving and Communication System) viewer 20;
- CAD Computer-Aided Diagnosis
- PACS Physical Archiving and Communication System
- the modality 12, the image storage server 14, each company's CAD processing server 18, the results integration CAD processing server 16, and the PACS viewer 20 are each connected to a communication network such as the Internet so that data can be sent and received.
- the modality 12 is an imaging device that captures an image of the inspection target site of the subject and generates a medical image.
- the modality 12 includes, for example, an X-ray imaging device, a CT (Computed Tomography) device, an MRI (Magnetic Resonance Imaging) device, a PET (Positron Emission Tomography) device, an ultrasonic device, and a CR (Computed Radiography) device using a plane X-ray detector. ) devices.
- the image storage server 14 is a server that manages medical images captured by the modality 12 .
- a computer having a large-capacity storage device is applied to the image storage server 14 .
- the computer incorporates software that provides the functionality of the data storage system.
- the image storage server 14 acquires medical images captured by the modality 12 and stores them in a large-capacity storage device.
- DICOM Digital Imaging and COmmunications in Medicine
- a medical image may be added with DICOM tag information defined in the DICOM standard.
- image in this specification can include the meaning of image data, which is a signal representing an image, in addition to the meaning of the image itself such as a photograph.
- the result integration CAD processing server 16 includes a first determination unit 16A.
- the first determination unit 16A performs a first determination of whether or not the medical image acquired from the image storage server 14 is normal for each organ. Whether or not the medical image is normal is, for example, whether or not the medical image can be said to be an image of a healthy person.
- a healthy person is a healthy person, eg, a person who is free from disease, disease, and lesions.
- the first determination result of the first determination unit 16A is linked to the medical image in the image storage server 14 and stored in a large-capacity storage device.
- the result integration CAD processing server 16 comprises a processor 16B and a memory 16C.
- Processor 16B executes instructions stored in memory 16C.
- the hardware structure of the processor 16B is various processors as shown below.
- Various processors include a CPU (Central Processing Unit), which is a general-purpose processor that executes software (programs) and acts as various functional units, a GPU (Graphics Processing Unit), which is a processor specialized for image processing, A circuit specially designed to execute specific processing such as PLD (Programmable Logic Device), which is a processor whose circuit configuration can be changed after manufacturing such as FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), etc. Also included are dedicated electrical circuits, which are processors with configuration, and the like.
- One processing unit may be composed of one of these various processors, or two or more processors of the same or different type (for example, a plurality of FPGAs, a combination of CPU and FPGA, or a combination of CPU and GPU).
- a plurality of functional units may be configured by one processor.
- a single processor is configured by combining one or more CPUs and software.
- a processor acts as a plurality of functional units.
- SoC System On Chip
- various functional units are configured using one or more of the above various processors as a hardware structure.
- the hardware structure of these various processors is, more specifically, an electrical circuit that combines circuit elements such as semiconductor elements.
- the memory 16C stores instructions to be executed by the processor 16B.
- the memory 16C includes RAM (Random Access Memory) and ROM (Read Only Memory), not shown.
- the processor 16B uses the RAM as a work area, executes software using various programs and parameters including a medical image processing program stored in the ROM, and uses the parameters stored in the ROM or the like to obtain the result Various processes of the integrated CAD processing server 16 are executed.
- Each company's CAD processing server 18 is composed of a plurality of CAD processing servers owned by a plurality of companies. Each company's CAD processing server 18 may be a single CAD processing server. Each company's CAD processing server 18 includes a second determination unit 18A.
- the second determination unit 18A is a medical image acquired from the image storage server 14, and performs an abnormality detection process for each organ on a medical image determined to be abnormal by at least the first determination unit 16A, It includes a program for performing a second determination for determining the presence or absence of one or more abnormalities from a medical image. Abnormalities include, for example, at least one of diseases, illnesses, and lesions.
- the second determination result of the second determination unit 18A is linked to the medical image in the image storage server 14 and stored in the large-capacity storage device.
- the second determination unit 18A may be provided in the result integration CAD processing server 16.
- the PACS viewer 20 is a terminal device used by a user such as a doctor, and for example, a known image viewer for interpretation is applied.
- PACS viewer 20 may be a personal computer, workstation, or tablet terminal.
- the PACS viewer 20 comprises an input device 20A and a display 20B.
- the input device 20A includes a pointing device such as a mouse and an input device such as a keyboard.
- a user can input instructions to the medical image diagnostic system 10 using the input device 20A.
- the display 20B displays screens required for operations on the input device 20A and functions as a part that implements a GUI (Graphical User Interface).
- a medical image captured by the modality 12 is displayed on the display 20B.
- the first determination result and the second determination result are displayed as CAD results on the display 20B.
- a touch panel display in which the input device 20A and the display 20B are integrated may be applied to the PACS viewer 20 .
- FIG. 2 is a flow chart showing a medical image diagnostic method using the medical image diagnostic system 10.
- FIG. 3 is a process diagram showing the medical image diagnosis method.
- the medical image diagnosis method is implemented by the processor 16B executing a medical image diagnosis program (medical image diagnosis support program) stored in the memory 16C.
- a processor not shown
- the medical image diagnostic program may be provided by a computer-readable non-transitory storage medium.
- the result integration CAD processing server 16 may read the medical image diagnosis program from the non-temporary storage medium and store it in the memory 16C.
- the medical image diagnostic method is performed for each organ of the subject.
- a case of diagnosing a lung CT image will be described.
- step S1 the processor 16B of the result integration CAD processing server 16 causes the image storage server 14 to acquire a CT image of the subject's lungs captured by the modality 12.
- the image storage server 14 acquires CT images captured by the modality 12 .
- step S2 the processor 16B inputs the CT image acquired by the image storage server 14 to the first determination unit 16A.
- the first determination unit 16A includes a normality determination AI 16D that determines whether the medical image is normal.
- the normality determination AI 16D is a trained model (an example of a "first trained model") that determines whether a CT image of the lung is normal when it is input, and includes a convolutional neural network.
- the normal discrimination AI 16D is generated by deep learning using a learning data set of medical images and labels indicating the presence or absence of abnormality.
- the normal discrimination AI 16D is generated by deep learning using a learning data set of normal CT images and normal labels and a learning data set of abnormal CT images and abnormal labels.
- a normal CT image is a CT image of a healthy subject.
- a non-normal CT image is a CT image with some abnormality, such as a CT image of a person having at least one of disease, disease, and lesion.
- the normality determination AI 16D is generated to output the degree of normality of the input CT image as a numerical value (score, an example of "probability").
- the normality determination AI 16D outputs that the CT image is not normal when the degree of normality of the CT image is less than a predetermined threshold, and outputs that the CT image is normal when the degree of normality of the CT image is greater than or equal to the threshold.
- the first determination unit 16A inputs the CT image to the normal determination AI 16D and performs the first determination (process P1).
- step S3 the processor 16B acquires the first determination result from the first determination unit 16A, and determines whether the CT image is normal from the first determination result. That is, the processor 16B determines whether or not there is an abnormality in the CT image. If the CT image is normal (an example of "first case", process P2), the process proceeds to step S4, and if the CT image is not normal (process P3), the process proceeds to step S5.
- step S4 the processor 16B causes the display 20B of the PACS viewer 20 to display the normal CT image in the display mode A, and performs post-processing in the processing mode A (process P4). Further, the processor 16B adds the “type A” incidental information to the CT image, stores it in the image storage server 14, and ends the processing of this flowchart.
- the post-processing of processing mode A is an example of post-processing in the first case where the medical image is determined to be normal in the first determination.
- step S5 an example of a "second determination step”
- the processor 16B inputs an abnormal CT image to the CAD processing server 18 of each company.
- Each company's CAD processing server 18 inputs the CT image to the second determination unit 18A and acquires the second determination result (process P5).
- the second discrimination unit 18A includes a lesion detection AI 18B manufactured by Company A that detects disease ⁇ , a lesion detection AI 18C manufactured by Company A that detects disease ⁇ , and a lesion detection AI 18C manufactured by Company A that detects disease ⁇ .
- disease ⁇ is lung cancer
- disease ⁇ is pneumonia
- disease ⁇ is pneumothorax.
- Each of the lesion detection AIs 18B to 18F outputs a trained model (“second trained model”) that outputs a diseased area (lesion area, an example of “abnormality”) in the CT image when a lung CT image is input. ), each containing a convolutional neural network.
- Each of the lesion detection AIs 18B to 18F is generated by performing deep learning on CT images of the lungs, using label images obtained by labeling each diseased region by a doctor as learning data.
- Each of the lesion detection AIs 18B to 18F is set to obtain the probability of each disease for each pixel of the CT image, and pixels exceeding a predetermined threshold are regarded as the region of the disease.
- Each of the lesion detection AIs 18B to 18F has a higher specificity than when performing lesion detection independently, that is, the threshold is set relatively high. As a result, each of the lesion detection AIs 18B-18F detects locations with a higher possibility of lesions. This is because there are few abnormalities in physical examinations and the like.
- the second discrimination unit 18A inputs the CT image to each lesion detection AI 18B-18F.
- Each of the lesion detection AIs 18B to 18F performs lesion detection processing from the CT image and outputs the result as the second determination result.
- step S6 the processor 16B acquires and integrates the second determination results from the second determination unit 18A (process P6).
- step S7 the processor 16B determines whether or not there is an abnormality (here, lesion) in the CT image from the integrated second determination result. If there is an abnormality in the CT image (an example of "second case", process P7), the process proceeds to step S8, and if there is no abnormality in the CT image (an example of "third case", process P8) goes to step S9.
- an abnormality here, lesion
- step S8 the processor 16B causes the display 20B of the PACS viewer 20 to display the abnormal CT image in the display mode B, and performs post-processing in the processing mode B (process P9).
- the post-processing of processing mode B is an example of post-processing in the second case where the medical image is determined to be abnormal in the first determination and abnormal in the second determination.
- the processor 16B adds the “type B” incidental information to the CT image, stores it in the image storage server 14, and ends the processing of this flowchart.
- step S9 the processor 16B causes the display 20B of the PACS viewer 20 to display the CT image with no abnormality in the display mode C, and performs post-processing in the processing mode C (process P10).
- the post-processing of processing mode C is an example of post-processing in the third case where the medical image is determined to be abnormal in the first determination and not abnormal in the second determination.
- the processor 16B adds the "type C" incidental information to the CT image, stores it in the image storage server 14, and terminates the processing of this flowchart.
- the processor 16B causes the display 20B to display the determination result in different modes for the display mode A, the display mode B, and the display mode C. That is, the processor 16B causes the display 20B to display the determination result in the display mode A in a manner different from the display modes B and C. FIG. The processor 16B may cause the display 20B to display the determination results in different modes for the display mode B and the display mode C, respectively.
- FIG. 4 is a diagram showing display form A. As shown in FIG. As shown in FIG. 4, in the display form A, a CT image I1 is displayed on the display 20B. Further, in the display mode A , an explanatory text T1 for the CT image I1 is displayed in the area on the right side of the CT image I1. In this case, the explanation T1 "Determined as normal by CAD" is displayed on the display 20B. Since there is a high possibility that the CT image I1 is not abnormal, only a display indicating that the CT image I1 is normal may be performed without displaying the CT image I1, and confirmation by the doctor may be skipped.
- FIG. 5 is a diagram showing a display form B.
- FIG. 5 in the display mode B, the CT image I2 is displayed on the display 20B, and the marker M1 surrounding the lesion area of the CT image I2 is superimposed on the CT image I2 .
- an explanation T2 for the CT image I2 which is an explanation T2 for the lesion area surrounded by the markers M1, is displayed in the area on the right side of the CT image I2 .
- the lesion area is detected by a lesion detection AI 18B manufactured by Company A that detects disease ⁇ (lung cancer), and the explanatory text T2 "Detected by lung cancer detection CAD manufactured by Company A " is displayed on the display 20B. .
- FIG. 6 is a diagram showing display form C.
- a CT image I3 is displayed on the display 20B, and a marker M2 surrounding the entire CT image I3 is superimposed on the CT image I3 .
- an explanatory text T3 related to the marker M2 which is the explanatory text T3 for the CT image I3 , is displayed in the area on the right side of the CT image I3 .
- an explanation T3 is displayed on the display 20B, stating "Cannot be determined to be normal by the normality determination AI. However, no abnormality is reported by any of the CADs." In this way, by using a display form different from the display form A and the display form B, it is possible to correct the judgment that depends too much on the lesion detection AI.
- processor 16B performs post-processing in different modes for the processing mode A and the processing modes B and C, respectively.
- processing mode A a flag is set so that the check by the doctor is simple, and in processing mode B and processing mode C, the flag is not set.
- the display order of the interpretation/examination list for the doctor to check the medical images may be changed so that the CT images of the processing mode B and the CT images of the processing mode C are prioritized over the CT images of the processing mode A. good.
- the processor 16B may perform post-processing in different modes for the processing mode B and the processing mode C, respectively.
- the medical image diagnosis method it is possible to determine whether the medical image is normal in the first determination unit 16A. Further, when the first determination unit 16A determines that the medical image is not normal, the second determination unit 18A can determine the presence or absence of an abnormality from the medical image. Therefore, it is possible to reduce the burden on doctors when performing image diagnosis on a large number of medical images.
- the second determination unit 18A determines whether there is an abnormality from the medical image. may be performed.
- the third determination process may be performed when the medical image is determined to be normal with a probability lower than a predetermined first threshold in the first determination.
- the third determination process is also an example of post-processing in the first case where the medical image is determined to be normal in the first determination.
- the normal determination AI 16D may be re-learned using the learning data set of the CT image and the abnormal label. As a result, it is possible to determine that the CT image is not normal in the normal determination AI 16D.
- the sensitivity of the normal determination AI 16D may be changed to a value at which the CT image is considered abnormal.
- the second determination unit 18A may again perform the fourth determination process of determining the presence or absence of an abnormality from the medical image.
- the fourth determination process is performed with a higher sensitivity than the second determination process (lower specificity than the second determination process) until it is determined that an abnormality exists, that is, the threshold is set relatively low. You can do it.
- the fourth determination process is also an example of post-processing in the third case where the first determination determined that the medical image was not normal and the second determination determined that there was no abnormality.
- the fourth discrimination process extracts lesions based on evaluation criteria that are acceptable even if a non-disease is discriminated as a disease, and reports it to the doctor. Assuming that the display form in this case is display form D, it is desirable to present to the doctor that the lesion is extracted in display form D with increased sensitivity.
- FIG. 7 is a diagram showing a display form D.
- FIG. 7 in the display mode D, the CT image I3 is displayed on the display 20B, the marker M2 surrounding the entire CT image I3 is superimposed on the CT image I3 , and the lesion of the CT image I3 is displayed.
- a marker M3 surrounding the region is superimposed on the CT image I3 .
- the marker M3 is displayed with a dashed line to indicate that it is a lesion detected with increased sensitivity.
- an explanatory text T4 related to the marker M3 which is the explanatory text T4 for the CT image I2 , is displayed in the area on the right side of the CT image I2 .
- an explanation T4 is displayed on the display 20B, stating, "The lung cancer was detected by the detection CAD made by company A as a result of the CAD processing again with increased sensitivity because the normal determination CAD cannot be said to be normal.”
- a slider bar SB for setting the sensitivity of the second discrimination section 18A may be displayed in the display mode C.
- the third determination process may be performed with the set sensitivity, and the display form D shown in FIG. 7 may be entered.
- the medical image and the determination result are associated with each other as usual without performing the above-described third and fourth determination processes. and save it in the image saving server 14 .
- each of the lesion detection AIs of the second discrimination unit 18A extracts liver cancer, multiple cysts, cirrhosis, and fatty liver from a CT image including the liver, and each of the above-described processes is performed.
- each of the above-described processes may be performed on a medical image including other organs.
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Quality & Reliability (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- High Energy & Nuclear Physics (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Molecular Biology (AREA)
- Veterinary Medicine (AREA)
- Heart & Thoracic Surgery (AREA)
- Optics & Photonics (AREA)
- Biophysics (AREA)
- Physiology (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
Description
本実施形態に係る医用画像診断システムは、健康診断等のように大量の医用画像に対して異常の有無の判別と医用画像が正常であるか否か(健常者の画像であるか否か)の判別とを行う場合に医師の負担を軽減するものである。
図2は、医用画像診断システム10を用いた医用画像診断方法を示すフローチャートである。また、図3は、医用画像診断方法を示すプロセス図である。医用画像診断方法は、プロセッサ16Bがメモリ16Cに記憶された医用画像診断プログラム(医用画像診断支援プログラム)を実行することで実現される。なお、第2の判別部18Aが各社CAD処理サーバ18に備えられる場合、各社CAD処理サーバ18に備えられた図示しないプロセッサが各社CAD処理サーバ18に備えられた図示しないメモリに記憶された医用画像診断プログラムを実行し、かつ、プロセッサ16Bがメモリ16Cに記憶された医用画像診断プログラムを実行することで実現される。医用画像診断プログラムは、コンピュータが読み取り可能な非一時的記憶媒体によって提供されてもよい。この場合、結果統合CAD処理サーバ16は、非一時的記憶媒体から医用画像診断プログラムを読み取り、メモリ16Cに記憶させてもよい。
第1の判別部16Aにおいて正常であると判別された医用画像について、例えば病院の閉院時間後(異なるタイミングの一例)に、第2の判別部18Aにおいて医用画像から異常の有無を判別する第3の判別処理を行ってもよい。第3の判別処理は、第1の判別において医用画像が予め定められた第1の閾値より低い確率で正常であると判別された場合に、行ってもよい。第3の判別処理も、第1の判別において医用画像が正常であると判別された第1の場合における後処理の一例である。
12…モダリティ
14…画像保存サーバ
16…結果統合CAD処理サーバ
16A…第1の判別部
16B…プロセッサ
16C…メモリ
16D…正常判別AI
18…各社CAD処理サーバ
18A…第2の判別部
18B…A社製病変検出AI
18C…A社製病変検出AI
18D…A社製病変検出AI
18E…B社製病変検出AI
18F…C社製病変検出AI
20…PACSビューワ
20A…入力装置
20B…ディスプレイ
I1…CT画像
I2…CT画像
I3…CT画像
M1…マーカ
M2…マーカ
M3…マーカ
P1~P10…医用画像診断の各プロセス
SB…スライダーバー
S1~S9…医用画像診断の各ステップ
T1…説明文
T2…説明文
T3…説明文
T4…説明文
Claims (14)
- 少なくとも1つのプロセッサと、
前記少なくとも1つのプロセッサに実行させるための命令を記憶する少なくとも1つのメモリと、
を備え、
前記少なくとも1つのプロセッサは、
被検者を撮影して得られた医用画像が正常であるか否かを判別する第1の判別を行い、
前記第1の判別において前記医用画像が正常でないと判別された場合に、前記医用画像から異常の有無を判別する第2の判別を行う、
医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記第1の判別において前記医用画像が正常であると判別された場合に、前記医用画像から異常の有無を判別する第3の判別を行う、
請求項1に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記第3の判別を、前記第2の判別とは異なるタイミングで行う、
請求項2に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記医用画像が正常である確率を用いて前記第1の判別を行い、
前記第1の判別において前記医用画像が第1の閾値より低い確率で正常であると判別された場合に、前記第3の判別を行う、
請求項2又は3に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記医用画像を入力すると前記医用画像が正常であるか否かを出力する第1の学習済みモデルによって前記第1の判別を行い、
前記第3の判別において前記異常があると判別された医用画像を用いて前記第1の学習済みモデルを再学習させる、
請求項2から4のいずれか1項に記載の医用画像診断システム。 - 前記第1の学習済みモデルは、異常のある医用画像及び異常のない医用画像と、異常の有無を示すラベルとの組み合わせを学習データセットとして学習された学習済みモデルである、
請求項5に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記第2の判別において前記異常がないと判別された場合に、前記医用画像から異常の有無を判別する第4の判別を行い、
前記第4の判別において、前記第2の判別における感度よりも相対的に高い感度で前記医用画像から異常の有無を判別する、
請求項1から6のいずれか1項に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記医用画像を入力すると前記医用画像の異常を出力する第2の学習済みモデルによって前記第2の判別を行う、
請求項1から7のいずれか1項に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記第1の判別において前記医用画像が正常であると判別された第1の場合、前記第1の判別において前記医用画像が正常ではないと判別されかつ前記第2の判別において前記異常があると判別された第2の場合、及び前記第1の判別において前記医用画像が正常ではないと判別されかつ前記第2の判別において前記異常がないと判別された第3の場合について、前記第1の場合と前記第2の場合及び前記第3の場合とで異なる態様で前記医用画像の判別結果をディスプレイに表示させる、
請求項1から8のいずれか1項に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記第2の場合と前記第3の場合とで異なる態様で前記医用画像の前記判別結果を前記ディスプレイに表示させる、
請求項9に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記第1の場合と第2の場合及び前記第3の場合とで前記医用画像に関してそれぞれ異なる後処理を行う、
請求項9又は10に記載の医用画像診断システム。 - 前記少なくとも1つのプロセッサは、
前記医用画像から前記被検者の臓器ごとに前記第1の判別及び前記第2の判別を行う、
請求項1から11のいずれか1項に記載の医用画像診断システム。 - 被検者を撮影して得られた医用画像が正常であるか否かを判別する第1の判別工程と、
前記第1の判別工程において前記医用画像が正常でないと判別された場合に、前記医用画像から異常の有無を判別する第2の判別工程と、
を備える医用画像診断方法。 - 請求項13に記載の医用画像診断方法をコンピュータに実行させるためのプログラム。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP22824751.6A EP4358022A1 (en) | 2021-06-17 | 2022-05-24 | Medical image diagnostic system, medical image diagnostic method, and program |
JP2023529713A JPWO2022264757A1 (ja) | 2021-06-17 | 2022-05-24 | |
US18/533,069 US20240112345A1 (en) | 2021-06-17 | 2023-12-07 | Medical image diagnosis system, medical image diagnosis method, and program |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2021100611 | 2021-06-17 | ||
JP2021-100611 | 2021-06-17 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US18/533,069 Continuation US20240112345A1 (en) | 2021-06-17 | 2023-12-07 | Medical image diagnosis system, medical image diagnosis method, and program |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022264757A1 true WO2022264757A1 (ja) | 2022-12-22 |
Family
ID=84526168
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2022/021224 WO2022264757A1 (ja) | 2021-06-17 | 2022-05-24 | 医用画像診断システム、医用画像診断方法及びプログラム |
Country Status (4)
Country | Link |
---|---|
US (1) | US20240112345A1 (ja) |
EP (1) | EP4358022A1 (ja) |
JP (1) | JPWO2022264757A1 (ja) |
WO (1) | WO2022264757A1 (ja) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006340835A (ja) | 2005-06-08 | 2006-12-21 | Konica Minolta Medical & Graphic Inc | 異常陰影候補の表示方法及び医用画像処理システム |
JP2012026982A (ja) * | 2010-07-27 | 2012-02-09 | Panasonic Electric Works Sunx Co Ltd | 検査装置 |
JP2019067069A (ja) * | 2017-09-29 | 2019-04-25 | アンリツ株式会社 | 異常検知装置及び異常検知方法並びに異常検知プログラム |
US20200202103A1 (en) * | 2017-06-09 | 2020-06-25 | University Of Surrey | Method and Apparatus for Processing Retinal Images |
JP6885517B1 (ja) * | 2020-03-17 | 2021-06-16 | 株式会社村田製作所 | 診断支援装置及びモデル生成装置 |
-
2022
- 2022-05-24 WO PCT/JP2022/021224 patent/WO2022264757A1/ja active Application Filing
- 2022-05-24 EP EP22824751.6A patent/EP4358022A1/en active Pending
- 2022-05-24 JP JP2023529713A patent/JPWO2022264757A1/ja active Pending
-
2023
- 2023-12-07 US US18/533,069 patent/US20240112345A1/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006340835A (ja) | 2005-06-08 | 2006-12-21 | Konica Minolta Medical & Graphic Inc | 異常陰影候補の表示方法及び医用画像処理システム |
JP2012026982A (ja) * | 2010-07-27 | 2012-02-09 | Panasonic Electric Works Sunx Co Ltd | 検査装置 |
US20200202103A1 (en) * | 2017-06-09 | 2020-06-25 | University Of Surrey | Method and Apparatus for Processing Retinal Images |
JP2019067069A (ja) * | 2017-09-29 | 2019-04-25 | アンリツ株式会社 | 異常検知装置及び異常検知方法並びに異常検知プログラム |
JP6885517B1 (ja) * | 2020-03-17 | 2021-06-16 | 株式会社村田製作所 | 診断支援装置及びモデル生成装置 |
Also Published As
Publication number | Publication date |
---|---|
JPWO2022264757A1 (ja) | 2022-12-22 |
US20240112345A1 (en) | 2024-04-04 |
EP4358022A1 (en) | 2024-04-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7756314B2 (en) | Methods and systems for computer aided targeting | |
JP2019033966A (ja) | 画像処理装置、画像処理方法、及び画像処理プログラム | |
US20190267132A1 (en) | Medical image display device, method, and program | |
US20190189270A1 (en) | Hospital information apparatus, hospital information system, and hospital information processing method | |
JP7525248B2 (ja) | 医用情報処理装置及び医用情報処理プログラム | |
Wang et al. | Automatic creation of annotations for chest radiographs based on the positional information extracted from radiographic image reports | |
EP4316378A1 (en) | Medical image processing device, method for operating medical image processing device, and operation program for medical image processing device | |
WO2022264757A1 (ja) | 医用画像診断システム、医用画像診断方法及びプログラム | |
US12062447B2 (en) | Medical image diagnosis support device, method, and program | |
WO2021187483A1 (ja) | 文書作成支援装置、方法およびプログラム | |
WO2021177156A1 (ja) | 画像処理装置、画像表示システム、画像処理方法及びプログラム | |
WO2022264755A1 (ja) | 医用画像診断システム、医用画像診断方法及びプログラム | |
JP2021175454A (ja) | 医用画像処理装置、方法およびプログラム | |
JP7524334B2 (ja) | 画像表示装置、方法およびプログラム | |
JP7361930B2 (ja) | 医用画像処理装置、方法およびプログラム | |
US20230360213A1 (en) | Information processing apparatus, method, and program | |
JP7376715B2 (ja) | 経過予測装置、経過予測装置の作動方法および経過予測プログラム | |
JP7483018B2 (ja) | 画像処理装置、画像処理方法及びプログラム、画像処理システム | |
WO2020241857A1 (ja) | 医療文書作成装置、方法およびプログラム、学習装置、方法およびプログラム、並びに学習済みモデル | |
US20230281810A1 (en) | Image display apparatus, method, and program | |
WO2023199957A1 (ja) | 情報処理装置、情報処理方法及び情報処理プログラム | |
US20230197253A1 (en) | Medical image processing apparatus, method, and program | |
US20230289534A1 (en) | Information processing apparatus, information processing method, and information processing program | |
WO2022264608A1 (ja) | 医用画像診断システム、医用画像診断システム評価方法及びプログラム | |
WO2022209500A1 (ja) | 機械学習モデル作成支援装置、機械学習モデル作成支援装置の作動方法、機械学習モデル作成支援装置の作動プログラム |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22824751 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2023529713 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022824751 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2022824751 Country of ref document: EP Effective date: 20240117 |