Computer program, information processing device, and information processing method

Abstract

Provided is a computer program with which it is possible to provide brain diagnosis information using a minimally-invasive MRI image, and without performing a PET scan. Also provided are an information processing device and an information processing method.　This computer program causes a computer to execute a process for acquiring an MRI image of a subject, specifying prescribed brain diagnosis information on the basis of the acquired MRI image, and newly generating and displaying the specified brain diagnosis information.

Description

Computer program, information processing device, and information processing method

[0001] The present invention relates to a computer program, an information processing device, and an information processing method.

[0002] In recent years, the aging population has led to an increase in the number of dementia patients and those at risk of developing dementia (mild cognitive impairment). The four major diseases that cause dementia are Alzheimer's disease (AD), dementia with Lewy bodies (DLB), vascular dementia, and frontotemporal lobar degeneration (FTLD). Other diseases, such as multiple sclerosis (MS), also present with various neurological symptoms, including dementia. While the cause of Alzheimer's disease remains unknown, specific brain lesions are observed as the disease progresses. For example, in Alzheimer's disease and frontotemporal lobar degeneration, the deposition of senile plaques caused by amyloid beta and the accumulation of abnormal proteins, primarily tau, are known to occur outside of neurons. It has been shown that the deposition of senile plaques occurs in the earliest stages of the onset of Alzheimer's disease, beginning well before clinical symptoms appear (for example, several decades ago). Furthermore, a protein of unknown function, α-synuclein, is expressed in brain neurons and is believed to be a cause of neurodegenerative diseases such as Parkinson's disease.

[0003] Brain examinations use MRI (Magnetic Resonance Imaging) and PET (Positron Emission Tomography). MRI detects signals using a magnetic field, so there is no radiation exposure, making it possible to measure the shape and size of tissues and detect even small lesions. PET, on the other hand, can examine the activity of tissues by measuring the uptake of radioactive glucose. Patent Document 1 discloses an apparatus that uses PET images to show the concentration distribution of the drug on a cross-section of the brain after injecting a drug that binds to amyloid beta in brain tissue into the subject.

[0004] International Publication No. WO2014 / 034724

[0005] As mentioned above, it is desirable to examine the shape and function of the brain using both MRI and PET. However, taking PET images using drugs requires the administration of radioactive materials, even in small amounts, which can lead to radiation exposure and is highly invasive and burdensome for patients. Furthermore, some patients with certain diseases, such as kidney disease, may not be able to receive drugs. Another problem is the high cost of PET scans. The time and effort required to undergo both modalities can also be problematic.

[0006] The present invention has been made in view of the above circumstances, and aims to provide a computer program, an information processing device, and an information processing method that can provide brain diagnostic information using minimally invasive MRI images without performing a PET examination.

[0007] The present application includes multiple means for solving the above-mentioned problems. As an example, a computer program causes a computer to execute the following process: acquire an MRI image of a subject, identify specific brain diagnostic information based on the acquired MRI image, and newly generate and display the identified brain diagnostic information.

[0008] According to the present invention, brain diagnostic information can be provided using minimally invasive MRI images.

[0009] 1 is a diagram illustrating an example of the configuration of an information processing system according to this embodiment. FIG. 2 is a diagram illustrating an example of the configuration of an information processing device. FIG. 3 is a diagram illustrating an example of processing of a first learning model. FIG. 4 is a diagram illustrating a first example of processing of a second learning model. FIG. 5 is a diagram illustrating a second example of processing of the second learning model. FIG. 6 is a diagram illustrating an example of processing of a third learning model. FIG. 7 is a diagram illustrating an example of a processing procedure when using the first learning model. FIG. 8 is a diagram illustrating an example of a processing procedure when using the second learning model. FIG. 9 is a diagram illustrating an example of a processing procedure when using the third learning model. FIG. 10 is a diagram illustrating an example of a processing procedure when calculating iron accumulation level or amyloid accumulation level. FIG. 11 is a diagram illustrating a first display example of brain diagnostic information. FIG. 12 is a diagram illustrating a second display example of brain diagnostic information. FIG. 13 is a diagram illustrating a third display example of brain diagnostic information. FIG. 14 is a diagram illustrating a fourth display example of brain diagnostic information. FIG. 15 is a diagram illustrating a fifth display example of brain diagnostic information.

[0010] An embodiment of the present invention will now be described with reference to the drawings. Fig. 1 is a diagram showing an example of the configuration of an information processing system according to this embodiment. The information processing system includes an information processing device 50. An input device 20 and a display device 30 are connected to the information processing device 50. An MRI device 10 and an image data server 100 are also connected to the information processing device 50 via a communication network 1.

[0011] The MRI device 10 is a device capable of capturing cross-sectional images using the magnetic resonance phenomenon, and can obtain MRI images (also referred to as MR images). By selecting imaging conditions, MRI images can be obtained that reflect tissue density, relaxation time (longitudinal relaxation time T1, transverse relaxation time T2), blood flow, and the amount of hydrogen atoms (proton density). MRI images can be generated by performing reconstruction processing on MRI signals containing position information.

[0012] As used herein, MRI images include, for example, T1 weighted images, T2 weighted images, and T2 ＊ This includes T2 weighted images, Fluid-Attenuated Inversion Recovery (FLAIR) images, Susceptibility-weighted imaging (SWI) images, and Phase Difference Enhanced Imaging (PADRE) images. SWI images are T2 ＊ It can be created from an MRI phase-enhanced image, and is an image in which magnetic susceptibility is qualitatively enhanced. PADRE images use the magnetic information of tissue contained in MRI phase image information, and employ a technique (phase-enhanced imaging) that emphasizes the phase difference to enhance tissue contrast within the brain. By selecting and enhancing phase information so as to selectively depict lesions and tissues to be observed, it is possible to image the nerve pathways within the brain, enabling image diagnosis of neurodegenerative diseases, etc. QSM (Quantitative Susceptibility Mapping) images are created using T2 ＊ The enhanced image can be generated by performing a predetermined calculation on the enhanced image. ＊ A predetermined calculation is performed on the enhanced image to obtain R2 ＊(R2 star) images can also be generated. QSM images and R2* images are collectively referred to as magnetic susceptibility images. In a broad sense, magnetic susceptibility images may also be included in MRI images. The predetermined arithmetic processing function required for converting an MRI image into a magnetic susceptibility image may be provided in the MRI device 10, the information processing device 50, the image data server 100, or another device (not shown) connected to the communication network 1. The MRI device 10 is installed, for example, in a medical institution such as a hospital. MRI images obtained by the MRI device 10 are stored in the image data server 100. MRI images are also referred to as MR images.

[0013] The image data server 100 records MRI images for each patient. For example, for each patient, the following information is recorded in association with the MRI image: the date on which the MRI image was taken (examination date), the imaging conditions, medication history such as whether or not medication was administered at the time of imaging and the number of times medication was administered, the name and amount of the therapeutic drug administered, and the patient's medical history.

[0014] QSM is a method for calculating local magnetic susceptibility from MRI phase images. The QSM imaging method uses multi-echo capture of 3D-GRE (Gradient Echo) intensity images and phase images. Since magnetic susceptibility is a material-specific physical property, material information within a voxel can be estimated from the magnetic susceptibility. QSM images quantitatively represent magnetic susceptibility. Paramagnetic materials with high magnetic susceptibility (e.g., hemosiderin and deoxyhemoglobin in blood) are displayed in white, while diamagnetic materials with low magnetic susceptibility are displayed in black.

[0015] The display device 30 includes a liquid crystal display panel, an organic EL display panel, or the like, and can display the results of processing by the information processing device 50. For example, the display device 30 can display brain diagnostic information obtained from MRI images. Here, brain diagnostic information means information that a doctor may use to diagnose the state of the brain, and includes not only amyloid beta and the like, but also information that may be used to diagnose specific diseases or disorders.

[0016] The input device 20 is an input interface such as a keyboard or a mouse that accepts operations of the information processing device 50. The input device 20 may be a touch panel, soft keys, hard keys, or the like provided on the display device 30. A medical professional such as a doctor can operate the input device 20 to input information such as a reference region and a region of interest (ROI) for obtaining brain diagnostic information. In addition, the medical professional such as a doctor can operate the input device 20 to display the processing results by the information processing device 50 on the display device 30. The input device 20 and the display device 30 may be incorporated into the information processing device 50.

[0017] Although not shown, client devices (personal computers, etc.) used by medical professionals such as doctors are connected to the communication network 1, and the client devices are set to be able to access the information processing device 50, which serves as a server. The medical professionals import MRI images from the MRI apparatus 10 into the client devices and upload them to the information processing device 50. The information processing device 50 may perform the processing described below, transmit brain diagnosis information (processing results) to the client devices, and display the brain diagnosis information on the client devices.

[0018] 2 is a diagram showing an example of the configuration of an information processing device 50. The information processing device 50 includes a control unit 51 that controls the entire device, a communication unit 52, a memory 53, an interface unit 54, an image processing unit 55, a display control unit 56, a diagnostic information generation unit 57, a storage unit 58, and a recording medium reading unit 64. The storage unit 58 stores a computer program 59, a first learning model 61, a second learning model 62, a third learning model 63, and required information.

[0019] The image processing unit 55, the display control unit 56, and the diagnostic information generating unit 57 may be configured as hardware, may be realized as software (computer program), or may be configured as both hardware and software. The information processing device 50 may be configured by distributing functions among multiple information processing devices.

[0020] The control unit 51 can be configured with a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), a GPU (Graphics Processing Unit), etc. The control unit 51 can execute processing defined by a computer program 59. In other words, the processing by the control unit 51 is also processing by the computer program 59.

[0021] The communication unit 52 includes, for example, a communication module, and has a communication function with the MRI apparatus 10 and the image data server 100 via the communication network 1. The communication unit 52 can acquire MRI images or magnetic susceptibility images from the MRI apparatus 10 or the image data server 100. The communication unit 52 can acquire information such as a reference region and a region of interest for obtaining brain diagnostic information.

[0022] The memory 53 can be configured with semiconductor memory such as SRAM (Static Random Access Memory), DRAM (Dynamic Random Access Memory), flash memory, etc. A computer program 59 can be loaded into the memory 53, and the control unit 51 can execute the computer program 59. The recording medium M on which the computer program 59 is recorded can be read by the recording medium reading unit 64.

[0023] The interface unit 54 provides an interface function between the input device 20 and the display device 30. The information processing device 50 (control unit 51) can exchange data and information with the input device 20 and the display device 30 via the interface unit 54.

[0024] The storage unit 58 can be configured, for example, by a hard disk or a semiconductor memory.

[0025] When the image processing unit 55 acquires an MRI image via the communication unit 52, it converts the acquired MRI image into a magnetic susceptibility image. The image processing unit 55 standardizes the magnetic susceptibility image to standardize the magnetic susceptibility value of the whole brain. The standardization process is a scaling method in which the minimum value of magnetic susceptibility is set to 0 and the maximum value is set to 1. The value x' after scaling is calculated as follows: x' = (x - xmin ) / (x max -x min ) where x is the value of magnetic susceptibility before scaling, and x max indicates the maximum value that x can take, and x min indicates the minimum value that x can take, which reduces the variability.

[0026] The display control unit 56 controls the information to be displayed on the display device 30 via the interface unit 54. The display control unit 56 can, for example, cause the display device 30 to display brain diagnosis information (processing results) by the information processing device 50.

[0027] The diagnostic information generation unit 57 generates brain diagnostic information based on the processing result by the control unit 51 and the processing result using the first learning model 61, the second learning model 62, and the third learning model 63. The brain diagnostic information will be described in detail later.

[0028] The control unit 51 acquires an MRI image or a magnetic susceptibility image of a patient (subject), identifies predetermined brain diagnostic information based on the acquired MRI image or magnetic susceptibility image, and displays the identified brain diagnostic information on the display device 30. This makes it possible to provide brain diagnostic information using a minimally invasive MRI image or a magnetic susceptibility image that can be converted from an MRI image without performing a PET examination. A specific description will be given below. Furthermore, although a QSM image is used as an example of a magnetic susceptibility image, the magnetic susceptibility image is not limited to a QSM image.

[0029] In a first example of identifying brain diagnostic information, a first learning model 61 predicts a predicted PET image based on an MRI image or a QSM image, and identifies brain diagnostic information indicating the accumulation of aggregates such as amyloid beta, tau protein, or alpha-synuclein based on the predicted PET image. In a second example, a second learning model 62 identifies brain diagnostic information indicating the accumulation of aggregates such as amyloid beta, tau protein, or alpha-synuclein based on an MRI image or a QSM image. In a third example, a third learning model 63 identifies brain diagnostic information such as amyloid beta positive / negative or tau protein positive / negative based on an MRI image or a QSM image. In a fourth example, brain diagnostic information indicating the accumulation or magnetic susceptibility of iron based on an MRI image or a QSM image is identified. Since a large number of fibrous aggregates is believed to inhibit the function of brain neurons and lead to cognitive impairment and motor disorders, measuring the presence, shape, size, etc. of these aggregates can estimate the degree or risk of brain neurodegeneration, providing valuable diagnostic information. The fibrillar aggregates in the present invention are mainly composed of amyloid beta, tau, and alpha-synuclein, but also include aggregates that have a fibrillar aggregate morphology, such as complexes with other brain proteins and aggregations of heterogeneous molecular species. Below, methods for identifying brain diagnostic information will be explained in order.

[0030] (First Example) FIG. 3 is a diagram illustrating an example of processing by the first learning model 61. The control unit 51 reads the first learning model 61 from the storage unit 58 and can perform the processing illustrated in FIG. 3 using the read first learning model 61. The first learning model 61 can be configured, for example, with a neural network (e.g., a convolutional neural network (CNN)). U-net or a generative adversarial network (GAN) may also be used, or a combination of these may be used. When an MRI image or QSM image of a patient is input, the first learning model 61 outputs a predicted PET image corresponding to the input MRI image or QSM image.

[0031] The predicted PET images output by the first learning model 61 are not PET images obtained by a PET examination using a drug, but rather predicted PET images (predicted images) predicted from MRI images or QSM images, and are less invasive. The predicted PET images output by the first learning model 61 may include amyloid predicted PET images, tau predicted PET images, α-synuclein aggregate images, and CL (centiloid) values. In amyloid PET, even with the same level of amyloid accumulation, PET values ​​vary depending on the PET drug, so centiloid is a standardized value independent of the drug. For example, the degree of amyloid accumulation in the brain can be expressed as a number ranging from 0 to 100, with the average for young healthy individuals being 0 and the average for confirmed AD being 100. Similar to PET images using a drug, the predicted PET images output by the first learning model 61 visualize, for example, the distribution of amyloid beta, tau protein, or α-synuclein in the brain. In this specification, unless otherwise specified, the predicted PET image is the predicted PET image output by the first learning model 61.

[0032] The first learning model 61 can be generated, for example, as follows. First, in a first step, the control unit 51 acquires first training data including an MRI image and voxel values ​​of the MRI image. The first training data may be acquired from the image data server 100. The control unit 51 trains the first learning model 61 based on the acquired first training data so that, when an MRI image is input, the first learning model 61 outputs the voxel values ​​of the MRI image. Next, in a second step, the control unit 51 acquires second training data including a magnetic susceptibility image (or an MRI image) and a predicted PET image corresponding to the magnetic susceptibility image. The second training data may be acquired from the image data server 100. The control unit 51 generates the first learning model 61 based on the acquired second training data so that, when a magnetic susceptibility image (or an MRI image) is input, the first learning model 61 outputs a predicted PET image.

[0033] As described above, by training the first learning model 61 using first training data including MRI images in the first stage, it becomes possible to train the first learning model 61 in the second stage with a small amount of data and in a short time, and the prediction accuracy of the predicted PET image can be improved even with a small amount of second training data.

[0034] The control unit 51 (diagnostic information generation unit 57) acquires the predicted PET image output by the first learning model 61, and based on the acquired predicted PET image, can identify brain diagnostic information indicating the degree of accumulation of at least one of the fibrous aggregates (e.g., at least one of amyloid beta, tau protein, and alpha-synuclein).

[0035] SUVR (Standardized Uptake Value Ratio) can be used as an index of accumulation. SUVR can be calculated using the formula: SUVR = SUV in the region of interest / SUV in the reference region. SUV (Standardized Uptake Value) can be expressed using the formula: SUV = PET value / administered radioactivity / body weight. Regions of interest include, but are not limited to, diagnostic regions such as the frontal lobe, occipital lobe, parietal lobe, posterior cingulate gyrus, and striatum. Reference regions include, but are not limited to, the entire cerebellum, the entire cerebellum and brainstem, cerebellar gray matter, and pons. The SUV of amyloid beta, tau protein, or alpha-synuclein can be determined, for example, by counting the number of voxels whose brightness values ​​in the region of interest and reference region (required region) are equal to or greater than a predetermined threshold. The accumulation degree (e.g., 00%) can be calculated based on the ratio of the count value to the total number of voxels in the required region.

[0036] Alternatively, CL (centiloid) may be used as an index of accumulation. CL can be calculated for each image based on SUVR. However, the CL value must be calculated using an appropriate conversion formula corresponding to the drug administered to the patient (subject).

[0037] (Second Example) FIG. 4 is a diagram illustrating a first example of processing by the second learning model 62. The control unit 51 can read the second learning model 62 from the storage unit 58 and perform the processing illustrated in FIG. 4 using the read second learning model 62. The second learning model 62 can be configured, for example, as a neural network. When the patient's MRI image or QSM image and information regarding the reference region are input, the second learning model 62 outputs an SUVR value for each voxel (or a CL value for the MRI image or QSM image). Note that the CL value may be calculated for a single image, such as an MRI image or QSM image, and calculated using a required conversion formula from the SUVR value for each voxel output by the second learning model 62. The control unit 51 identifies the SUVR value (or the CL value for the MRI image or QSM image) output by the second learning model 62 as brain diagnosis information.

[0038] FIG. 5 illustrates a second example of processing by the second learning model 62. When the second learning model 62 receives the patient's MRI or QSM image and information about the reference region and the region of interest, it outputs the SUVR value of each voxel in the region of interest (or the CL value for the MRI or QSM image). The CL value may be calculated for a single image, such as an MRI or QSM image, and calculated from the SUVR value of the region of interest output by the second learning model 62 using a required conversion formula. The control unit 51 identifies the SUVR value (or the CL value for the MRI or QSM image) output by the second learning model 62 as brain diagnosis information.

[0039] The second learning model 62 can be generated, for example, as follows, similarly to the first learning model 61. First, in a first step, the control unit 51 acquires first training data including an MRI image and voxel values ​​of the MRI image. The first training data may be acquired from the image data server 100. When an MRI image is input, the control unit 51 trains the second learning model 62 based on the acquired first training data so that the second learning model 62 outputs the voxel values ​​of the MRI image. Next, in a second step, the control unit 51 acquires second training data including a magnetic susceptibility image (or an MRI image) and brain diagnosis information indicating the degree of accumulation of at least one of fibrous aggregates (e.g., at least one of amyloid beta and tau protein). The second training data may be acquired from the image data server 100. Based on the acquired second training data, the control unit 51 generates a second learning model 62 so that when a magnetic susceptibility image (or MRI image) is input, it outputs brain diagnostic information indicating the degree of accumulation of at least one of fibrous aggregates (e.g., at least one of amyloid beta and tau protein).

[0040] As described above, by training the second learning model 62 using first training data including MRI images in the first stage, it becomes possible to train the second learning model 62 in the second stage with a small amount of data and in a short time, and the prediction accuracy of brain diagnostic information can be improved even with a small amount of second training data.

[0041] (Third Example) FIG. 6 is a diagram illustrating an example of processing by the third learning model 63. The control unit 51 can read the third learning model 63 from the memory unit 58 and perform the processing shown in FIG. 6 using the read third learning model 63. The third learning model 63 can be configured, for example, as a neural network. When a patient's MRI image or QSM image and information about a region of interest are input, the third learning model 63 outputs brain diagnosis information indicating at least one of amyloid beta positive, amyloid beta negative, tau protein positive, and tau protein negative in the region of interest. Amyloid beta positive means that abnormal accumulation of amyloid beta is present, and amyloid beta negative means that amyloid beta accumulation is present but not abnormal. The same applies to tau protein positive / negative. Note that if region of interest information is not input, amyloid beta positive / negative and tau protein positive / negative are predicted based on the entire input QSM image or MRI image.

[0042] The third learning model 63 can be generated, for example, as follows, similarly to the first learning model 61. First, in a first step, the control unit 51 acquires first training data including an MRI image and voxel values ​​of the MRI image. The first training data may be acquired from the image data server 100. The control unit 51 trains the third learning model 63 based on the acquired first training data so that, when an MRI image is input, the third learning model 63 outputs the voxel values ​​of the MRI image. Next, in a second step, the control unit 51 acquires second training data including a magnetic susceptibility image (or an MRI image) and brain diagnosis information indicating at least one of amyloid beta positive / negative and tau protein positive / negative. The second training data may be acquired from the image data server 100. The control unit 51 generates the third learning model 63 based on the acquired second training data so that, when a magnetic susceptibility image (or an MRI image) is input, the third learning model 63 outputs brain diagnosis information indicating at least one of amyloid beta positive / negative and tau protein positive / negative.

[0043] As described above, by training the third learning model 63 using first training data including MRI images in the first stage, it becomes possible to train the third learning model 63 in the second stage with a small amount of data and in a short time, and the prediction accuracy of brain diagnosis information can be improved even if the second training data is small.

[0044] (Fourth Example) The control unit 51 acquires a QSM image of a patient and, based on the acquired QSM image, identifies brain diagnostic information indicating the degree of iron accumulation compared to that of healthy subjects. Pathological studies have shown a progression of amyloid accumulation → phosphorylated tau accumulation → inflammation in the brain → neuronal displacement → disease. Iron accumulation may occur in inflamed areas of the brain. Brain diagnostic information indicating the degree of iron accumulation can be used to estimate amyloid accumulation and evaluate brain atrophy. Furthermore, when the control unit 51 acquires an MRI image of a patient, the control unit 51 may convert the acquired MRI image into a QSM image. Furthermore, the control unit 51 may also acquire information regarding the region of interest and identify brain diagnostic information indicating the degree of iron accumulation or amyloid accumulation in the region of interest compared to that of healthy subjects. This will be described in detail below.

[0045] First, data from healthy individuals is collected in advance, and a database of magnetic susceptibility distributions in the healthy individual data is created. Then, QSM images of standard brains from healthy individuals are created. A database of healthy individuals is constructed, and z-scores (brain diagnostic information) are calculated for each patient's region of interest based on the distribution of magnetic susceptibility for each region of interest. In other words, the patient's QSM images are subjected to image processing in units of voxels, which are three-dimensional pixels (VBM: Voxel-Based Morphometry). A typical statistical processing method is to generate a z-score map.

[0046] The z-score can be calculated as follows: The mean and standard deviation of magnetic susceptibility are calculated for each voxel from the QSM image of a standard brain of a healthy subject, and the z-score is calculated based on the calculated mean and standard deviation and the magnetic susceptibility of the patient's QSM image. The z-score can be calculated using the following formula: z-score = (M(x,y,z) - I(x,y,z)) / SD(x,y,z). M represents the mean magnetic susceptibility of healthy subjects, SD represents the standard deviation of the magnetic susceptibility of healthy subjects, and I represents the patient's magnetic susceptibility. The z-score indicates how many times the standard deviation of the magnetic susceptibility distribution of a standard brain of a healthy subject is different. Using the z-score map, it is possible to quantitatively analyze which parts of the patient's QSM image are changing in comparison with those of a healthy subject (normal standard brain). For example, voxels with positive z-score values ​​indicate areas of atrophy compared to normal brain regions, and the larger the value, the greater the statistical deviation. For example, a z-score of "2" indicates a difference exceeding two standard deviations from the mean, which is evaluated as having a statistically significant difference with a risk of approximately 5%, allowing for quantitative evaluation of atrophy in the region of interest.

[0047] Next, the processing of the information processing device 50 will be described.

[0048] 7 is a diagram showing an example of a processing procedure when the first learning model 61 is used. For convenience, the following description will be given assuming that the control unit 51 is the subject of the processing. The control unit 51 acquires an MRI image of the subject (patient) (S11) and accepts the setting of a region of interest and a reference region (S12). The control unit 51 converts the acquired MRI image into a QSM image (S13) and standardizes the converted QSM image (S14). Note that if a QSM image is acquired directly, the processing of step S13 is not necessary.

[0049] The control unit 51 inputs the standardized QSM image into the first learning model 61 and acquires a predicted PET image output by the first learning model 61 (S15). Based on the acquired predicted PET image, the control unit 51 calculates the SUVR, which indicates the degree of accumulation of amyloid beta and tau protein in the region of interest, and the CL (centiloid) for the QSM image (S16). The control unit 51 outputs brain diagnosis information (S17) and ends the process.

[0050] 8 is a diagram showing an example of a processing procedure when the second learning model 62 is used. The control unit 51 acquires an MRI image of the subject (patient) (S21) and accepts the setting of the region of interest and the reference region (S22). The control unit 51 converts the acquired MRI image into a QSM image (S23) and standardizes the converted QSM image (S24). Note that if the QSM image is acquired directly, the processing of step S23 is not necessary.

[0051] The control unit 51 inputs the set region of interest, reference region, and standardized QSM image to the second learning model 62, and obtains the SUVR indicating the degree of accumulation of amyloid beta and tau protein in the region of interest and the CL (centiloid) for the QSM image output by the second learning model 62 (S25). The control unit 51 outputs brain diagnosis information (S26), and ends the process.

[0052] 9 is a diagram showing an example of a processing procedure when the third learning model 63 is used. The control unit 51 acquires an MRI image of the subject (patient) (S31) and accepts the setting of a region of interest (S32). The control unit 51 converts the acquired MRI image into a QSM image (S33) and standardizes the converted QSM image (S34). Note that if the QSM image is acquired directly, the processing of step S33 is not necessary.

[0053] The control unit 51 inputs the standardized QSM image to the third learning model 63, and obtains the amyloid β positive / negative (+ / -) and tau protein positive / negative (+ / -) in the region of interest output by the third learning model 63 (S35). The control unit 51 outputs brain diagnosis information (S36), and ends the process.

[0054] 10 is a diagram showing an example of a processing procedure for calculating the degree of iron accumulation or the degree of amyloid accumulation. The control unit 51 acquires an MRI image of the subject (patient) (S41) and accepts the setting of a region of interest and a reference region (S42). The control unit 51 converts the acquired MRI image into a QSM image (S43) and standardizes the converted QSM image (S44). Note that if a QSM image is acquired directly, the processing of step S43 is not necessary.

[0055] The control unit 51 generates a QSM image of a standard brain of a healthy subject group by referring to the healthy subject DB (S45). The control unit 51 calculates a Z-score indicating the degree of iron accumulation or amyloid accumulation in the region of interest based on the QSM image of the subject and the QSM image of the healthy subject group. The control unit 51 calculates an SUVR indicating the degree of iron accumulation or amyloid accumulation in the region of interest and a centiloid (CL) for the QSM image based on the QSM image of the subject (S47). The control unit 51 outputs brain diagnosis information (S48) and ends the process.

[0056] The control unit 51 may perform all of the processes in the first to fourth examples described above and output brain diagnosis information, or may perform required processes among the first to fourth examples. For example, the control unit 51 may perform only the first and fourth examples and output brain diagnosis information.

[0057] The diagnostic information generating unit 57 generates brain diagnostic information to be output, and the display control unit 56 performs control processing for displaying the brain diagnostic information on the display device 30. An example of displaying the brain diagnostic information will be described below.

[0058] 11 is a diagram showing a first display example of brain diagnostic information. A diagnostic information screen 210 displays a patient information area 211 that displays patient information, an image area 214 that displays an image of the patient, a numerical value area 217 that displays indicators and numerical values ​​of the brain diagnostic information, a similarity score area 216 that displays similarity scores, and a recommended examination area 218 that displays recommended examination items.

[0059] The patient information area 211 displays information such as patient ID (which may include name) for selecting a patient, date of birth, age, sex, examination date such as MRI examination, medication history, and medical history. The patient ID may be selected from among multiple patients. If there are multiple examination dates, the examination date may be selected.

[0060] By selecting the "Predicted Image" tab 212, the predicted PET image output by the first learning model 61 is displayed in the image area 214. The predicted PET image displays cross-sectional images in the form of axial, sagittal, and coronal planes. A horizontal bar and cursor 215 are displayed below (or to the side of) each cross-sectional image to specify which slice image among multiple tomographic images to display. By moving the cursor 215, the desired cross-sectional image can be displayed. The patient's QSM image may also be displayed simultaneously in the image area 214. By selecting the "Input Image" tab 213, the input QSM image can be displayed in the image area 214.

[0061] The numerical value area 217 displays a reference region setting window for selecting a reference region (in the figure, the entire cerebellum is set), and an SUVR value indicating the degree of accumulation of amyloid beta or tau protein for each region of interest. The SUVR value may be calculated using information output by the second learning model 62 or based on the predicted PET image output by the first learning model 61. As shown in the figure, the centiloid (CL) value for the entire image may be displayed, and an overall assessment of amyloid beta positive / negative (+ / -) or tau protein positive / negative (+ / -) may be displayed.

[0062] The similarity score area 216 displays the similarity with the positive image in the range of 0 to 1 (in the example shown in the figure, the similarity is displayed as 0.35). The recommended examination area 218 displays examination items that can be recommended to the patient based on the brain diagnosis information. The recommended examination items may be output by the diagnostic information generation unit 57 on a rule basis, for example.

[0063] This allows doctors to make their own diagnosis by observing QSM images, and automatically provides brain diagnostic information, which can assist doctors in their diagnoses and reduce the burden on doctors during diagnosis. Furthermore, because brain diagnostic information is provided automatically, it can reduce the variation in diagnoses due to the experience of individual doctors.

[0064] 12 is a diagram showing a second display example of brain diagnostic information. A diagnostic information screen 220 displays a patient information area 211 that displays patient information, an image area 221 that displays an image of the patient, and a numerical value area 222 that displays indicators and numerical values ​​of the brain diagnostic information. The patient information area 211 is the same as in the first display example shown in FIG. 11.

[0065] The image area 221 displays the predicted PET image output by the first learning model 61. The cross-sectional image to be displayed for the predicted PET image can be selected from an axial plane (Axial), a sagittal plane (Sagittal), and a coronal plane (Coronal). In the illustrated example, an axial plane (Axial) is selected. The image area 221 displays the input QSM image and the predicted PET image for comparison. As with the first display example, the desired cross-sectional image can be displayed by moving the cursor 215.

[0066] As in the first display example, the numerical value area 222 displays a reference region setting window for selecting a reference region (in the figure, the entire cerebellum is set), an SUVR value indicating the degree of amyloid-beta or tau protein accumulation for each region of interest, and a CL value indicating the degree of amyloid-beta or tau protein accumulation for each region of interest. The numerical value area 222 displays the determination result of amyloid-beta positive / negative (+ / -) or tau protein positive / negative (+ / -) output by the third learning model 63.

[0067] This allows doctors to make their own diagnosis by observing QSM images, and automatically provides brain diagnostic information, which can assist doctors in their diagnoses and reduce the burden on doctors during diagnosis. Furthermore, because brain diagnostic information is provided automatically, it can reduce the variation in diagnoses due to the experience of individual doctors.

[0068] 13 is a diagram showing a third display example of brain diagnostic information. A diagnostic information screen 230 displays a patient information area 211 that displays patient information, an image area 233 that displays an image of the patient, and a numerical value area 234 that displays indicators and numerical values ​​of the brain diagnostic information. The patient information area 211 is the same as in the first display example shown in FIG. 11.

[0069] By selecting the "Slice" tab 231, the required number of slice images are selected from the cross-sectional images of the input QSM image in the image area 233, and a Z-score map is displayed, with Z-score values ​​indicating the degree of iron accumulation superimposed on each slice image. The cross-sectional images of the QSM image can be selected from axial, sagittal, and coronal planes. In the illustrated example, the axial plane is selected. Furthermore, the degree of iron accumulation can be visualized by applying a gradation to the QSM image according to the Z-score value. In the illustrated example, the Z-score is visualized in a range from 0 to 4. This allows easy determination of the extent of iron accumulation in which parts of the brain. The degree of amyloid accumulation may also be estimated and displayed based on the degree of iron accumulation.

[0070] The magnetic susceptibility value is displayed for each region of interest in the numerical value area 234. By selecting the "results summary" tab 232, a diagnosis information screen 240 of a fourth display example, which will be described later, is displayed.

[0071] As described above, when a doctor observes QSM images and makes his / her own judgment, brain diagnostic information is automatically provided, which can assist the doctor in making a diagnosis and reduce the burden on the doctor during the diagnosis. In addition, because brain diagnostic information is automatically provided, it is possible to reduce the variation in diagnosis due to the experience of individual doctors.

[0072] 14 is a diagram showing a fourth display example of brain diagnostic information. A diagnostic information screen 240 displays a patient information area 211 that displays patient information, and a first numerical area 241 and a second numerical area 242 that display indicators and numerical values ​​of the brain diagnostic information. The patient information area 211 is the same as in the first display example shown in FIG. 11.

[0073] The first numerical value area 241 displays Z-score values ​​for the whole brain gray matter and the whole brain white matter, as well as the status of iron deposition. In the illustrated example, Z-scores are visualized in a range of 0 to 4. The second numerical value area 242 displays Z-score values ​​for other regions than the whole brain gray matter and the whole brain white matter, as well as the status of iron deposition. In the illustrated example, the frontal lobe, temporal lobe, occipital lobe, and parietal lobe are displayed, but this is not limiting. The degree of amyloid accumulation may also be estimated from the degree of iron accumulation and displayed.

[0074] As described above, the Z-score indicating the degree of iron accumulation and the status of iron deposition are displayed for the whole brain gray matter, whole brain white matter, and other regions of interest, which reduces the burden on doctors during diagnosis and also reduces the variability in diagnoses due to the experience of individual doctors.

[0075] 15 is a diagram showing a fifth display example of brain diagnostic information. A diagnostic information screen 250 displays a patient information area 211 that displays patient information, an image area 253 that displays an image of the patient, and a numerical value area 254 that displays indicators and numerical values ​​of the brain diagnostic information. The patient information area 211 is the same as in the first display example shown in FIG. 11.

[0076] By selecting the "Magnetic susceptibility distribution image" tag 251, a magnetic susceptibility distribution image is displayed in the image area 253. The magnetic susceptibility distribution image can visualize magnetic susceptibility (iron accumulation level) by applying a gradation to the QSM image according to the magnetic susceptibility value. The cross-sectional image of the QSM image can be selected from axial, sagittal, and coronal planes. In the illustrated example, the axial plane is selected. As in the first display example, the desired cross-sectional image can be displayed by moving the cursor 215. Note that the amyloid accumulation level may be estimated from the iron accumulation level and displayed.

[0077] In the numerical value area 254, the magnetic susceptibility value and the Z score value are displayed for each region of interest.

[0078] As described above, when a doctor observes QSM images and makes his / her own judgment, brain diagnostic information is automatically provided, which can assist the doctor in making a diagnosis and reduce the burden on the doctor during the diagnosis. In addition, because brain diagnostic information is automatically provided, it is possible to reduce the variation in diagnosis due to the experience of individual doctors.

[0079] According to this embodiment, the use of less invasive MRI images can reduce the burden on patients compared to when an actual PET scan is performed. Furthermore, since MRI scans are more prevalent than PET scans, screening of more patients becomes possible.

[0080] In this embodiment, the information processing device 50 may acquire information other than the MRI image of the subject, for example, information including at least one of test results regarding the subject's cognitive ability (e.g., cognitive ability test results, cognitive function test results, etc.) and biomarkers (e.g., blood test results, genetic information, etc.), identify specific brain diagnostic information based on the acquired information, and newly generate and display the identified brain diagnostic information.

[0081] The computer program of this embodiment causes a computer to execute the following processes: acquire an MRI image of a subject, identify specific brain diagnostic information based on the acquired MRI image, and newly generate and display the identified brain diagnostic information.

[0082] The computer program of this embodiment causes a computer to execute a process of acquiring a magnetic susceptibility image generated based on an MRI image of a subject, inputting the acquired magnetic susceptibility image into a first learning model that outputs a predicted PET image when the magnetic susceptibility image is input, acquiring a predicted PET image, and identifying the brain diagnostic information indicating the degree of accumulation of at least one of the fibrous aggregates based on the acquired predicted PET image.

[0083] The computer program of this embodiment causes a computer to execute a process of acquiring a magnetic susceptibility image generated based on an MRI image of a subject, inputting the acquired magnetic susceptibility image into a second learning model that outputs brain diagnostic information indicating the accumulation level of at least one of the fibrous aggregates when the magnetic susceptibility image is input, and acquiring the brain diagnostic information indicating the accumulation level of at least one of the fibrous aggregates, thereby identifying the brain diagnostic information.

[0084] The computer program of this embodiment causes a computer to acquire a magnetic susceptibility image generated based on an MRI image of a subject, input the acquired magnetic susceptibility image into a third learning model that outputs brain diagnostic information indicating at least one of amyloid beta positive / negative and tau protein positive / negative when the magnetic susceptibility image is input, and acquire the brain diagnostic information indicating at least one of amyloid beta positive / negative and tau protein positive / negative, thereby identifying the brain diagnostic information.

[0085] The computer program of this embodiment causes a computer to execute a process of acquiring a magnetic susceptibility image generated based on an MRI image of the subject, and identifying the brain diagnostic information indicating the degree of iron accumulation compared to that of healthy subjects based on the acquired magnetic susceptibility image.

[0086] The computer program of this embodiment causes a computer to execute the following processes: acquire first training data including an MRI image and voxel values ​​of the MRI image; train the first learning model based on the acquired first training data so that when an MRI image is input, the voxel values ​​of the MRI image are output; acquire second training data including a magnetic susceptibility image and a predicted PET image; and generate the first learning model based on the acquired second training data so that when a magnetic susceptibility image is input, the predicted PET image is output.

[0087] The computer program of this embodiment causes a computer to execute the following processes: acquire first training data including an MRI image and voxel values ​​of the MRI image; train the second learning model based on the acquired first training data so that when an MRI image is input, the voxel values ​​of the MRI image are output; acquire second training data including a magnetic susceptibility image and brain diagnostic information indicating the accumulation level of at least one of fibrous aggregates; and generate the second learning model based on the acquired second training data so that when a magnetic susceptibility image is input, the brain diagnostic information indicating the accumulation level of at least one of fibrous aggregates is output.

[0088] The computer program of this embodiment causes a computer to execute the following processes: acquire first training data including an MRI image and voxel values ​​of the MRI image; train the third learning model based on the acquired first training data so that when an MRI image is input, the voxel values ​​of the MRI image are output; acquire second training data including a magnetic susceptibility image and brain diagnostic information indicating at least one of amyloid beta positive / negative and tau protein positive / negative; and generate the third learning model based on the acquired second training data so that when a magnetic susceptibility image is input, the brain diagnostic information indicating at least one of amyloid beta positive / negative and tau protein positive / negative is output.

[0089] The computer program of this embodiment causes a computer to execute the following process: acquire information including test results regarding the subject's cognitive ability and at least one of biomarkers; identify specific brain diagnostic information based on the acquired information; and newly generate and display the identified brain diagnostic information.

[0090] The information processing device of this embodiment includes an acquisition unit that acquires MRI images of a subject, an identification unit that identifies specific brain diagnostic information based on the acquired MRI images, and a display unit that newly generates and displays the identified brain diagnostic information.

[0091] The information processing method of this embodiment acquires an MRI image of a subject, identifies predetermined brain diagnostic information based on the acquired MRI image, and newly generates and displays the identified brain diagnostic information.

[0092] The present embodiment can be used to diagnose dementia, multiple sclerosis, mild cognitive impairment (MCI), mild cognitive impairment due to Alzheimer's disease (MCI due to AD), prodromal Alzheimer's disease (prodromal AD), the pre-symptomatic stage of Alzheimer's disease / preclinical AD, Parkinson's disease, insomnia, sleep disorders, cognitive decline, cognitive impairment, amyloid-positive / negative diseases, movement disorders, movement function disorders, movement disorder diseases, Alzheimer's disease, synucleinopathy, multiple system atrophy, vascular dementia, cerebrovascular disorders, dementia with Lewy bodies, other neurodegenerative diseases, and the like.

[0093] REFERENCE SIGNS LIST 1 Communication network 10 MRI device 20 Input device 30 Display device 50 Information processing device 51 Control unit 52 Communication unit 53 Memory 54 Interface unit 55 Image processing unit 56 Display control unit 57 Diagnostic information generation unit 58 Storage unit 59 Computer program 61 First learning model 62 Second learning model 63 Third learning model 64 Recording medium reading unit 100 Image data server

Claims

1. On the computer, obtaining a magnetic susceptibility image generated based on the MRI image of the subject; inputting the magnetic susceptibility image into a learning model, and identifying brain diagnostic information indicating the degree of accumulation of at least one of fibrous aggregates based on information output from the learning model; Controlling the identified brain diagnostic information to be displayed on a screen; A computer program that executes a process.

2. The information output from the learning model is a predicted PET image.

2. The computer program of claim 1.

3. The index indicating the degree of accumulation is SUVR (Standardized Update Value Ratio) or CL (Centiloid), 2. The computer program of claim 1.

4. The information output from the learning model is at least one of amyloid beta positive / negative and tau protein positive / negative.

2. The computer program of claim 1.

5. Identifying the brain diagnostic information indicating the degree of iron accumulation compared to a healthy subject based on the magnetic susceptibility image.

5. A computer program product according to claim 1, which causes a process to be executed.

6. The magnetic susceptibility image is a QSM image. A computer program according to any one of claims 1 to 4.

7. an acquisition unit that acquires a magnetic susceptibility image generated based on an MRI image of the subject; an identification unit that inputs the magnetic susceptibility image into a learning model and identifies brain diagnostic information indicating the accumulation level of at least one of fibrous aggregates based on information output from the learning model; a display unit that controls the specified brain diagnostic information to be displayed on a screen; Equipped with Information processing device.

8. obtaining a magnetic susceptibility image generated based on the MRI image of the subject; inputting the magnetic susceptibility image into a learning model, and identifying brain diagnostic information indicating the degree of accumulation of at least one of fibrous aggregates based on information output from the learning model; Controlling the identified brain diagnostic information to be displayed on a screen; Information processing methods.