WO2019181887A1 - Epilepsy determination device, epilepsy determination system, epilepsy determination method and epilepsy determination program - Google Patents

Epilepsy determination device, epilepsy determination system, epilepsy determination method and epilepsy determination program Download PDF

Info

Publication number
WO2019181887A1
WO2019181887A1 PCT/JP2019/011268 JP2019011268W WO2019181887A1 WO 2019181887 A1 WO2019181887 A1 WO 2019181887A1 JP 2019011268 W JP2019011268 W JP 2019011268W WO 2019181887 A1 WO2019181887 A1 WO 2019181887A1
Authority
WO
WIPO (PCT)
Prior art keywords
epilepsy
determination
images
epileptic seizure
learned model
Prior art date
Application number
PCT/JP2019/011268
Other languages
French (fr)
Japanese (ja)
Inventor
宏知 高橋
亜利 江間見
謙介 川合
Original Assignee
国立大学法人東京大学
学校法人自治医科大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 国立大学法人東京大学, 学校法人自治医科大学 filed Critical 国立大学法人東京大学
Publication of WO2019181887A1 publication Critical patent/WO2019181887A1/en

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B10/00—Other methods or instruments for diagnosis, e.g. instruments for taking a cell sample, for biopsy, for vaccination diagnosis; Sex determination; Ovulation-period determination; Throat striking implements
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/369—Electroencephalography [EEG]

Definitions

  • the present invention relates to an epilepsy determination device, an epilepsy determination system, an epilepsy determination method, and an epilepsy determination program.
  • an electroencephalogram is measured by placing an electrode on a patient's head, and an electroencephalogram pattern peculiar to epileptic seizures is observed. Observation of such an electroencephalogram pattern may be performed by an epilepsy specialist.
  • Non-Patent Document 1 described below describes a technique for classifying whether or not an epileptic seizure has occurred by using a neural network, using brain waves as input, extracting feature values by wavelet transform.
  • the present invention provides an epilepsy determination device, an epilepsy determination system, an epilepsy determination method, and an epilepsy determination program that can determine an epileptic seizure with higher accuracy.
  • An epilepsy determination device includes an acquisition unit that acquires brain wave data of a subject, a generation unit that generates brain images by cutting the brain wave data at a predetermined time width, and a plurality of images
  • a determination unit configured to input an image into the learned model and determine whether the brain wave of the epileptic seizure appears in any of the plurality of images by the learned model.
  • the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave.
  • the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
  • a plurality of test images representing brain waves are input to a learned model, and an evaluation unit that evaluates the determination accuracy of the learned model, and a predetermined time width is set based on the evaluation by the evaluation unit
  • the evaluation unit may evaluate the determination accuracy of the learned model based on a plurality of indices.
  • the determination accuracy of the learned model can be evaluated from various aspects, and the time width for extracting the electroencephalogram data can be adjusted more appropriately.
  • the plurality of indices are an index indicating whether or not an epileptic seizure appears in an image showing an epileptic seizure, and an epilepsy seizure does not appear in an image that does not show an epileptic seizure An index indicating whether it can be judged correctly, an image indicating that an epileptic seizure does not appear for an image showing an epileptic seizure, and an epileptic seizure appearing for an image not showing an epileptic seizure And an index indicating whether the determination is erroneously performed.
  • the plurality of images may include waveforms indicated by a plurality of colors corresponding to the plurality of electrodes for which the electroencephalogram data is measured.
  • a plurality of electrodes for which electroencephalogram data is measured can be distinguished, and determination can be performed under the same conditions as a specialist diagnoses an epileptic seizure, and an epileptic seizure can be determined with higher accuracy.
  • the learned model may be a learned convolutional neural network.
  • a learned model having a determination accuracy equivalent to or higher than that of a person can be used, and an epileptic seizure can be determined with higher accuracy.
  • the generation unit may generate a plurality of images by performing at least one of a low-pass filter, a high-pass filter, and a notch filter on the electroencephalogram data, and then cutting out the electroencephalogram data with a predetermined time width. Good.
  • An epilepsy determination system is an epilepsy determination system including a measurement apparatus that measures brain wave data of a subject and an epilepsy determination apparatus that determines epilepsy based on the electroencephalogram data.
  • the determination device cuts out the electroencephalogram data with a predetermined time width, generates a plurality of images representing the electroencephalogram, and inputs the plurality of images into the learned model.
  • one of the plurality of images A determination unit that determines whether an electroencephalogram of an epileptic seizure appears, and a notification unit that notifies a predetermined terminal when the determination unit determines that an electroencephalogram of the epileptic seizure appears in any of a plurality of images And having.
  • the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave.
  • the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
  • the burden of the electroencephalogram confirmation work can be reduced by notifying a predetermined terminal.
  • An epilepsy determination method includes acquiring a subject's brain wave data, cutting out the brain wave data with a predetermined time width, generating a plurality of images representing brain waves, and a plurality of images To the learned model, and to determine whether the brain wave of the epileptic seizure appears in any of the plurality of images by the learned model.
  • the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave.
  • the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
  • An epilepsy determination program includes a computer provided in an epilepsy determination apparatus, an acquisition unit that acquires brain wave data of a subject, a plurality of brain wave data cut out with a predetermined time width, and a plurality of brain waves It functions as a generation unit that generates an image and a determination unit that inputs a plurality of images to a learned model and determines whether the brain wave of an epileptic seizure appears in any of the plurality of images by the learned model.
  • the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave.
  • the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
  • an epilepsy determination device an epilepsy determination system, an epilepsy determination method, and an epilepsy determination program that can determine an epileptic seizure with higher accuracy can be provided.
  • FIG. 1 is a diagram showing functional blocks of an epilepsy determination system 1 according to an embodiment of the present invention.
  • the epilepsy determination system 1 includes an epilepsy determination device 10 and a measurement device 20.
  • the measuring device 20 is a device that measures the subject's brain waves, and measures the electrical activity of the subject's brain with a plurality of electrodes attached to the subject's head.
  • the measuring device 20 may have, for example, 21 electrodes arranged according to the international 10-20 method, but the number and arrangement of electrodes are arbitrary.
  • the epilepsy determination apparatus 10 includes an acquisition unit 11, a generation unit 12, a determination unit 13, an evaluation unit 14, a setting unit 15, a learning unit 16, a notification unit 17, and a storage unit 18.
  • the acquisition unit 11 acquires brain wave data of the subject.
  • the acquisition unit 11 may acquire the subject's brain wave data measured by the measurement device 20, but may acquire the subject's brain wave data stored in advance in the storage unit 18 or another storage device.
  • the generating unit 12 cuts out the electroencephalogram data with a predetermined time width and generates a plurality of images representing the electroencephalogram.
  • the predetermined time width can be arbitrarily set, but may be, for example, 0.5 seconds, 1 second, 2 seconds, 5 seconds, and 10 seconds.
  • the generation unit 12 cuts out the electroencephalogram data in the range of 0 to 10 seconds to express the first electroencephalogram.
  • An image is generated, brain wave data in the range of 1 to 11 seconds is cut out to generate a second image representing the brain wave, and this process is continued.
  • brain wave data of 3590 to 3600 seconds is cut out to represent the brain wave 3591
  • An image may be generated.
  • the generation unit 12 may shift the cut-out range at equal intervals so that the ranges of the electroencephalogram data cut out by the plurality of images overlap, but the range of the electroencephalogram data cut out by the plurality of images may not overlap. Good.
  • the generation unit 12 may generate a plurality of images by applying at least one of a low-pass filter, a high-pass filter, and a notch filter to the electroencephalogram data, and then cutting out the electroencephalogram data with a predetermined time width.
  • the low-pass filter may be a filter that passes data with a frequency of 60 Hz or less
  • the high-pass filter may be a filter that passes data with a frequency of 0.5 Hz or more
  • the notch filter is 50 Hz. It may be a filter that blocks data of a certain frequency.
  • the determination unit 13 inputs a plurality of images to the learned model 13a, and determines whether the brain wave of an epileptic seizure appears in any of the plurality of images by the learned model 13a.
  • the learned model 13a is a learned convolutional neural network (Convolutional Neural Network: CNN).
  • CNN Convolutional Neural Network
  • the evaluation unit 14 inputs a plurality of test images representing brain waves to the learned model 13a and evaluates the determination accuracy of the learned model 13a.
  • the evaluation unit 14 may evaluate the determination accuracy of the learned model 13a based on a plurality of indices. As a result, the determination accuracy of the learned model 13a can be evaluated in a multifaceted manner, and the time width for extracting the electroencephalogram data can be adjusted more appropriately.
  • an epileptic seizure there are two or more indicators that indicate whether an epileptic seizure appears correctly for an image showing an epileptic seizure, and an epileptic seizure for an image that does not show an epileptic seizure. If there is an index that indicates that the epileptic seizure does not appear, an index that indicates that the epileptic seizure does not appear, and an image that does not show the epileptic seizure appear. And an index indicating whether it is erroneously determined that it is present.
  • the determination accuracy of the learned model 13a in a multifaceted manner in the case of true positive, true negative, false negative and false positive, the determination accuracy can be evaluated in a multifaceted manner,
  • the time width for extracting the electroencephalogram data can be adjusted more appropriately.
  • the setting unit 15 sets a predetermined time width based on the evaluation by the evaluation unit 14.
  • the setting unit 15 may set a predetermined time width so that the evaluation by the evaluation unit 14 is improved.
  • the learning unit 16 includes a plurality of learning images representing brain waves, which are cut out with a predetermined time width set by the setting unit 15, and information regarding whether or not the plurality of learning images represent brain waves of epileptic seizures. , And learning data as learning data, a new learned model is generated.
  • the plurality of learning images representing the electroencephalograms may be images that have been measured by the measurement device 20 in the past and associated with information regarding whether or not they represent an epileptic seizure.
  • the learning model is a neural network such as a convolutional neural network
  • the learning unit 16 may perform learning processing of the weighting coefficient of the neural network by the error back propagation method.
  • the setting unit 15 adjusts the time width for extracting the electroencephalogram data, and the learning unit 16 can generate a new learned model so as to improve the determination accuracy.
  • the epilepsy seizure can be determined with higher accuracy by the epilepsy determination device 10.
  • the notification unit 17 notifies a predetermined terminal when the determination unit 13 determines that any one of the plurality of images shows an epileptic brain wave.
  • the predetermined terminal may be any information processing terminal.
  • the predetermined terminal may be a terminal used by a specialist, a terminal used by a person who cares for the subject, or a terminal used by the subject himself / herself. You may do it.
  • notification to a predetermined terminal can reduce the burden of confirming the electroencephalogram.
  • the storage unit 18 may store learning data used by the learning unit 16. But the learning part 16 and the memory
  • storage part 18 may not be provided in the epilepsy determination apparatus 10, and the learning process of a learning model may be performed by the other apparatus accessible via a communication network.
  • the epilepsy determination apparatus 10 may transmit the learning process condition to another apparatus that performs the learning process, and may receive information for configuring the learned model.
  • the information for configuring the learned model is information for identifying the learning model. In the case of a neural network, the number of layers, the type of layer, the connection of nodes between layers, and the weight coefficient of node connection (including threshold values) ) And the type of activation function.
  • FIG. 2 is a diagram illustrating a physical configuration of the epilepsy determination device 10 according to the present embodiment.
  • the epilepsy determination device 10 includes a CPU (Central Processing Unit) 10a corresponding to a calculation unit, a RAM (Random Access Memory) 10b corresponding to a storage unit, a ROM (Read only Memory) 10c corresponding to a storage unit, and a communication unit. 10d, an input unit 10e, and a display unit 10f.
  • Each of these components is connected to each other via a bus so that data can be transmitted and received.
  • the epilepsy determination apparatus 10 may be implement
  • the structure shown in FIG. 2 is an example, and the epilepsy determination apparatus 10 may have a structure other than these, and it is not necessary to have a part of these structures.
  • the CPU 10a is a control unit that performs control related to execution of a program stored in the RAM 10b or the ROM 10c, and calculates and processes data.
  • the CPU 10a is a calculation unit that executes a program for determining epilepsy (epilepsy determination program) based on an image representing an electroencephalogram.
  • the CPU 10a receives various data from the input unit 10e and the communication unit 10d, and displays a calculation result of the data on the display unit 10f or stores it in the RAM 10b or the ROM 10c.
  • the RAM 10b can rewrite data in the storage unit, and may be composed of, for example, a semiconductor storage element.
  • the RAM 10b may store an epilepsy determination program executed by the CPU 10a. These are examples, and the RAM 10b may store data other than these, or some of them may not be stored.
  • the ROM 10c is capable of reading data out of the storage unit, and may be composed of, for example, a semiconductor storage element.
  • the ROM 10c may store, for example, a search program and data that is not rewritten.
  • the communication unit 10d is an interface that connects the epilepsy determination device 10 to another device.
  • the communication unit 10d may be connected to a communication network such as the Internet or a LAN (Local Area Network).
  • the input unit 10e receives data input from the user, and may include, for example, a keyboard and a touch panel.
  • the display unit 10f visually displays the calculation result by the CPU 10a, and may be configured by, for example, an LCD (Liquid Crystal Display).
  • the display unit 10f may display, for example, an image representing an electroencephalogram or a determination result.
  • the epilepsy determination program may be provided by being stored in a computer-readable storage medium such as the RAM 10b or the ROM 10c, or may be provided via a communication network connected by the communication unit 10d.
  • various operations described with reference to FIG. 1 are realized by the CPU 10 a executing the epilepsy determination program.
  • these physical structures are illustrations, Comprising: It does not necessarily need to be an independent structure.
  • the epilepsy determination apparatus 10 may include an LSI (Large-Scale Integration) in which a CPU 10a, a RAM 10b, and a ROM 10c are integrated.
  • FIG. 3 is an example of the image P input to the learned model of the epilepsy determination apparatus 10 according to the present embodiment.
  • the image P is an example of an image cut out from the electroencephalogram data by the generation unit 12. In this example, the electroencephalogram data for 10 seconds is cut out.
  • the image P shows an electroencephalogram measured by 21 electrodes of the measuring device 20 and an electrocardiogram waveform.
  • the electroencephalogram measured by the 21 electrodes is the waveform from the top of the graph to the 21st, and the electrocardiogram waveform is the waveform shown at the bottom of the graph.
  • Image P includes waveforms indicated by a plurality of colors corresponding to a plurality of electrodes from which electroencephalogram data was measured.
  • waveforms drawn adjacent to each other in the vertical direction are shown in different colors, or the color of the waveform is changed so as to correspond to the arrangement of the electrodes, but the selection of the color is arbitrary.
  • color-coding in this way, it is possible to distinguish between multiple electrodes for which electroencephalogram data has been measured, and to make determinations under the same conditions as specialists diagnose epileptic seizures, and to determine epileptic seizures with higher accuracy Can do.
  • FIG. 4 is a first graph showing the determination accuracy of the epilepsy determination apparatus 10 according to the present embodiment.
  • the first graph shows the result of testing whether or not an epileptic seizure can be correctly determined by the epilepsy determination apparatus 10 according to the present embodiment by measuring brain wave data for several tens of hours for each of 24 subjects.
  • the horizontal axis indicates the time width (Time Window) set by the setting unit 15, and the vertical axis indicates the determination accuracy of the learned model in which the learning process is performed with the time width set.
  • the determination accuracy is a true positive value evaluated by the evaluation unit 14. That is, the determination accuracy shown in the first graph is a determination accuracy evaluated based on an index indicating whether or not an image showing an epileptic seizure can be correctly determined if an epileptic seizure appears.
  • the median represented by the box plot when the time width is 0.5 seconds, the median represented by the box plot is about 0, the upper quartile is slightly larger than 0.1, and the maximum value is 0.2. Degree.
  • the minimum value represented by the box plot is about 0
  • the lower quartile is about 0.5
  • the median is slightly larger than 0.7
  • the quartile is about 0.9
  • the maximum value is about 1.0.
  • the time width is 2 seconds
  • the minimum value represented by the box plot is about 0
  • the lower quartile is about 0.4
  • the median is about 0.7
  • the quartile is about 0.9
  • the maximum value is about 1.0.
  • the minimum value represented by the box plot is about 0
  • the lower quartile is about 0.5
  • the median is slightly smaller than 0.7
  • the upper side The quartile is slightly smaller than 0.9 and the maximum value is slightly smaller than 1.0.
  • the minimum value represented by the box plot is about 0
  • the lower quartile is about 0.2
  • the median is about 0.6
  • the quartile is slightly smaller than 0.8 and the maximum value is about 1.0.
  • the time width for cutting out the electroencephalogram data affects whether or not the learning process of the learning model can be performed appropriately, and may cause the determination accuracy to fluctuate.
  • the epilepsy determination apparatus 10 it is possible to generate a new learned model so as to improve the determination accuracy by adjusting the time width for extracting the electroencephalogram data, and to determine the epileptic seizure with higher accuracy. be able to.
  • FIG. 5 is a second graph showing the determination accuracy of the epilepsy determination apparatus 10 according to the present embodiment.
  • the second graph shows the results of testing the brain wave data for several tens of hours for 24 subjects and testing whether the normal state can be correctly determined by the epilepsy determination device 10 according to the present embodiment.
  • the horizontal axis indicates the time width (Time ⁇ Window) set by the setting unit 15, and the vertical axis indicates the determination accuracy of the learned model in which the learning process is performed with the time width set.
  • the determination accuracy is a true negative value evaluated by the evaluation unit 14. That is, the determination accuracy shown in the second graph is a determination accuracy evaluated based on an index indicating whether it is possible to correctly determine that an epileptic seizure does not appear for an image in which the epileptic seizure does not appear.
  • the minimum value represented by the box plot is about 0.97 to 0.99
  • the side quartile is about 0.98 to 0.99
  • the median is about 0.99
  • the upper quartile is about 0.995
  • the maximum value is about 1.0.
  • the first graph has a difference that the median value of determination accuracy is about 0 when the time width is 0.5 seconds, whereas the second graph is close to 1.0.
  • the median determination accuracy decreases as the time width is increased from 1 second to 10 seconds, whereas in the second graph, the time width is increased from 1 second to 10 seconds.
  • the median of the determination accuracy increases as the length increases.
  • the epilepsy determination apparatus 10 evaluates the determination accuracy of a learned model based on a plurality of indexes, and the electroencephalogram data is improved so as to improve the determination accuracy evaluated based on the plurality of indexes in a well-balanced manner.
  • index may be maximized.
  • FIG. 6 is a flowchart of epilepsy determination processing executed by the epilepsy determination system 1 according to the present embodiment.
  • the brain wave data of the subject is measured by the measuring device 20 (S10).
  • the subsequent processing may be executed after the electroencephalogram data is accumulated, or may be executed sequentially while measuring the electroencephalogram data.
  • the generation unit 12 of the epilepsy determination apparatus 10 performs preprocessing by filtering the acquired electroencephalogram data (S11). After that, the generation unit 12 cuts out the electroencephalogram data with a predetermined time width and generates a plurality of images (S12).
  • the determination unit 13 of the epilepsy determination apparatus 10 inputs a plurality of images to a learned model such as CNN, and determines whether an electroencephalogram of an epileptic seizure appears (S13). If it is determined that there is an epileptic seizure (S14: YES), the notification unit 17 of the epilepsy determination device 10 notifies that an epileptic seizure has been detected at a predetermined terminal (S15). On the other hand, when it is not determined that there is an epileptic seizure (S14: NO), the epilepsy determination process ends.
  • a learned model such as CNN
  • brain wave data is cut out with a predetermined time width, a plurality of images representing brain waves are generated, and brain waves of epileptic seizures appear in any of the plurality of images.
  • the determination can be made under the same conditions as the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
  • brain wave data is cut out with a predetermined time width to generate a plurality of images representing the brain waves, and the brain wave of an epileptic seizure appears in one of the plurality of images.
  • the learned model determine whether or not the epileptic seizure is detected, it is possible to perform the determination under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
  • the burden of the electroencephalogram confirmation work can be reduced by notifying a predetermined terminal.
  • FIG. 7 is a flowchart of the learning process executed by the epilepsy determination system 1 according to this embodiment.
  • the evaluation unit 14 of the epilepsy determination apparatus 10 inputs a test image to the learned model and evaluates the determination accuracy (S20).
  • the setting unit 15 sets a time width for cutting out the electroencephalogram data based on the evaluation by the evaluation unit 14 (S21).
  • the learning unit 16 cuts out the electroencephalogram data with the set time width, executes the learning process of the learning model, and generates a new learned model (S22). Then, the epilepsy determination device 10 mounts a new learned model in the determination unit 13 (S23). Thus, the learning process ends.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Pathology (AREA)
  • Engineering & Computer Science (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Physics & Mathematics (AREA)
  • Psychiatry (AREA)
  • Psychology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

Provided are an epilepsy determination device, an epilepsy determination system, an epilepsy determination method and an epilepsy determination program, with which it is possible to determine an epileptic seizure with high accuracy. An epilepsy determination device (10) comprises: an acquisition unit (11) that acquires brainwave data of a subject; a generation unit (12) that extracts the brainwave data in prescribed time widths to generate a plurality of images representing brainwaves; and a determination unit (13) that inputs the plurality of images to a learned model (13a) to cause the determination to be made as to whether an epileptic seizure brainwave appears in any of the plurality of images.

Description

てんかん判定装置、てんかん判定システム、てんかん判定方法及びてんかん判定プログラムEpilepsy determination device, epilepsy determination system, epilepsy determination method, and epilepsy determination program
　本発明は、てんかん判定装置、てんかん判定システム、てんかん判定方法及びてんかん判定プログラムに関する。 The present invention relates to an epilepsy determination device, an epilepsy determination system, an epilepsy determination method, and an epilepsy determination program.
　従来、患者の頭部に電極を配置して脳波（Electroencephalogram：EEG）を測定し、てんかん発作に特有の脳波パターンを観察することがある。そのような脳波パターンの観察は、てんかんの専門医によって行われることがある。 Conventionally, an electroencephalogram (EEG) is measured by placing an electrode on a patient's head, and an electroencephalogram pattern peculiar to epileptic seizures is observed. Observation of such an electroencephalogram pattern may be performed by an epilepsy specialist.
　下記非特許文献１には、脳波を入力として、ウェーブレット変換により特徴量を抽出し、ニューラルネットワークによりてんかん発作が生じているか否かを分類する技術が記載されている。 Non-Patent Document 1 described below describes a technique for classifying whether or not an epileptic seizure has occurred by using a neural network, using brain waves as input, extracting feature values by wavelet transform.
　しかしながら、てんかん発作には様々な種類があり、発作中の脳波パターンも多岐にわたることが知られている。そのため、従来の技術では、十分に高い精度でてんかん発作を判定することが難しかった。 However, there are various types of epileptic seizures, and it is known that the electroencephalogram patterns during the seizure are diverse. For this reason, it has been difficult for conventional techniques to determine epileptic seizures with sufficiently high accuracy.
　そこで、本発明は、てんかん発作をより高い精度で判定することができるてんかん判定装置、てんかん判定システム、てんかん判定方法及びてんかん判定プログラムを提供する。 Therefore, the present invention provides an epilepsy determination device, an epilepsy determination system, an epilepsy determination method, and an epilepsy determination program that can determine an epileptic seizure with higher accuracy.
　本発明の一態様に係るてんかん判定装置は、対象者の脳波データを取得する取得部と、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成する生成部と、複数の画像を学習済みモデルに入力し、学習済みモデルによって、複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させる判定部と、を備える。 An epilepsy determination device according to an aspect of the present invention includes an acquisition unit that acquires brain wave data of a subject, a generation unit that generates brain images by cutting the brain wave data at a predetermined time width, and a plurality of images A determination unit configured to input an image into the learned model and determine whether the brain wave of the epileptic seizure appears in any of the plurality of images by the learned model.
　この態様によれば、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成して、複数の画像のいずれかにてんかん発作の脳波が表れているかを学習済みモデルに判定させることで、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。 According to this aspect, the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave. Thus, the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
　上記態様において、脳波を表す複数のテスト用画像を学習済みモデルに入力して、学習済みモデルの判定精度を評価する評価部と、評価部による評価に基づいて、所定の時間幅の設定を行う設定部と、設定部により設定された所定の時間幅で切り出された、脳波を表す複数の学習用画像と、複数の学習用画像がてんかん発作の脳波を表しているか否かに関する情報と、を学習用データとして学習モデルを学習させ、新たな学習済みモデルを生成する学習部と、をさらに備えてもよい。 In the above aspect, a plurality of test images representing brain waves are input to a learned model, and an evaluation unit that evaluates the determination accuracy of the learned model, and a predetermined time width is set based on the evaluation by the evaluation unit A setting unit, a plurality of learning images representing brain waves cut out by a predetermined time width set by the setting unit, and information regarding whether or not the plurality of learning images represent brain waves of epileptic seizures, And a learning unit that learns a learning model as learning data and generates a new learned model.
　この態様によれば、脳波データを切り出す時間幅を調整して、判定精度が向上するように新たな学習済みモデルを生成することができ、より高い精度でてんかん発作を判定することができる。 According to this aspect, it is possible to generate a new learned model so as to improve the determination accuracy by adjusting the time width for extracting the electroencephalogram data, and to determine epileptic seizures with higher accuracy.
　上記態様において、評価部は、複数の指標に基づいて学習済みモデルの判定精度を評価してもよい。 In the above aspect, the evaluation unit may evaluate the determination accuracy of the learned model based on a plurality of indices.
　この態様によれば、学習済みモデルの判定精度を多面的に評価することができ、脳波データを切り出す時間幅をより適切に調整することができる。 According to this aspect, the determination accuracy of the learned model can be evaluated from various aspects, and the time width for extracting the electroencephalogram data can be adjusted more appropriately.
　上記態様において、複数の指標は、てんかん発作が表れている画像について、てんかん発作が表れていると正しく判定できるかを示す指標と、てんかん発作が表れていない画像について、てんかん発作が表れていないと正しく判定できるかを示す指標と、てんかん発作が表れている画像について、てんかん発作が表れていないと誤って判定するかを示す指標と、てんかん発作が表れていない画像について、てんかん発作が表れていると誤って判定するかを示す指標と、を含んでよい。 In the above aspect, the plurality of indices are an index indicating whether or not an epileptic seizure appears in an image showing an epileptic seizure, and an epilepsy seizure does not appear in an image that does not show an epileptic seizure An index indicating whether it can be judged correctly, an image indicating that an epileptic seizure does not appear for an image showing an epileptic seizure, and an epileptic seizure appearing for an image not showing an epileptic seizure And an index indicating whether the determination is erroneously performed.
　この態様によれば、トゥルー・ポジティブ、トゥルー・ネガティブ、フォールス・ネガティブ及びフォールス・ポジティブの場合について学習済みモデルの判定精度を多面的に評価することができ、脳波データを切り出す時間幅をより適切に調整することができる。 According to this aspect, it is possible to evaluate the determination accuracy of the learned model from the perspective of true positive, true negative, false negative, and false positive in a multifaceted manner, and more appropriately set the time width for extracting the electroencephalogram data. Can be adjusted.
　上記態様において、複数の画像は、脳波データを測定した複数の電極に対応する複数の色で示された波形を含んでもよい。 In the above aspect, the plurality of images may include waveforms indicated by a plurality of colors corresponding to the plurality of electrodes for which the electroencephalogram data is measured.
　この態様によれば、脳波データを測定した複数の電極を区別して、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。 According to this aspect, a plurality of electrodes for which electroencephalogram data is measured can be distinguished, and determination can be performed under the same conditions as a specialist diagnoses an epileptic seizure, and an epileptic seizure can be determined with higher accuracy. .
　上記態様において、学習済みモデルは、学習済みの畳み込みニューラルネットワークであってもよい。 In the above aspect, the learned model may be a learned convolutional neural network.
　この態様によれば、人と同程度かそれ以上の判定精度を有する学習済みモデルを用いることができ、より高い精度でてんかん発作を判定することができる。 According to this aspect, a learned model having a determination accuracy equivalent to or higher than that of a person can be used, and an epileptic seizure can be determined with higher accuracy.
　上記態様において、生成部は、脳波データに対してローパスフィルタ、ハイパスフィルタ及びノッチフィルタのうち少なくともいずれかを施した後、脳波データを所定の時間幅で切り出して、複数の画像を生成してもよい。 In the above aspect, the generation unit may generate a plurality of images by performing at least one of a low-pass filter, a high-pass filter, and a notch filter on the electroencephalogram data, and then cutting out the electroencephalogram data with a predetermined time width. Good.
　この態様によれば、脳波データに含まれることがあるノイズを除去した上で複数の画像を生成することができ、より高い精度でてんかん発作を判定することができる。 According to this aspect, it is possible to generate a plurality of images after removing noise that may be included in the electroencephalogram data, and to determine an epileptic seizure with higher accuracy.
　本発明の他の態様に係るてんかん判定システムは、対象者の脳波データを測定する測定装置と、脳波データに基づいて、てんかんを判定するてんかん判定装置と、を備えるてんかん判定システムであって、てんかん判定装置は、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成する生成部と、複数の画像を学習済みモデルに入力し、学習済みモデルによって、複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させる判定部と、判定部により、複数の画像のいずれかにてんかん発作の脳波が表れていると判定された場合に、所定の端末に通知する通知部と、を有する。 An epilepsy determination system according to another aspect of the present invention is an epilepsy determination system including a measurement apparatus that measures brain wave data of a subject and an epilepsy determination apparatus that determines epilepsy based on the electroencephalogram data. The determination device cuts out the electroencephalogram data with a predetermined time width, generates a plurality of images representing the electroencephalogram, and inputs the plurality of images into the learned model. Depending on the learned model, one of the plurality of images A determination unit that determines whether an electroencephalogram of an epileptic seizure appears, and a notification unit that notifies a predetermined terminal when the determination unit determines that an electroencephalogram of the epileptic seizure appears in any of a plurality of images And having.
　この態様によれば、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成して、複数の画像のいずれかにてんかん発作の脳波が表れているかを学習済みモデルに判定させることで、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。また、てんかん発作の脳波が表れていると判定された場合に、所定の端末に通知することで、脳波の確認作業の負担を軽減することができる。 According to this aspect, the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave. Thus, the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy. In addition, when it is determined that an electroencephalogram of an epileptic seizure appears, the burden of the electroencephalogram confirmation work can be reduced by notifying a predetermined terminal.
　本発明の他の態様に係るてんかん判定方法は、対象者の脳波データを取得することと、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成することと、複数の画像を学習済みモデルに入力し、学習済みモデルによって、複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させることと、を含む。 An epilepsy determination method according to another aspect of the present invention includes acquiring a subject's brain wave data, cutting out the brain wave data with a predetermined time width, generating a plurality of images representing brain waves, and a plurality of images To the learned model, and to determine whether the brain wave of the epileptic seizure appears in any of the plurality of images by the learned model.
　この態様によれば、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成して、複数の画像のいずれかにてんかん発作の脳波が表れているかを学習済みモデルに判定させることで、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。 According to this aspect, the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave. Thus, the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
　本発明の他の態様に係るてんかん判定プログラムは、てんかん判定装置に備えられたコンピュータを、対象者の脳波データを取得する取得部、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成する生成部、及び複数の画像を学習済みモデルに入力し、学習済みモデルによって、複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させる判定部、として機能させる。 An epilepsy determination program according to another aspect of the present invention includes a computer provided in an epilepsy determination apparatus, an acquisition unit that acquires brain wave data of a subject, a plurality of brain wave data cut out with a predetermined time width, and a plurality of brain waves It functions as a generation unit that generates an image and a determination unit that inputs a plurality of images to a learned model and determines whether the brain wave of an epileptic seizure appears in any of the plurality of images by the learned model.
　この態様によれば、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成して、複数の画像のいずれかにてんかん発作の脳波が表れているかを学習済みモデルに判定させることで、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。 According to this aspect, the brain wave data is cut out with a predetermined time width, a plurality of images representing the brain waves are generated, and the learned model determines whether any of the plurality of images shows an epileptic brain wave. Thus, the determination can be performed under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
　本発明によれば、てんかん発作をより高い精度で判定することができるてんかん判定装置、てんかん判定システム、てんかん判定方法及びてんかん判定プログラムを提供することができる。 According to the present invention, an epilepsy determination device, an epilepsy determination system, an epilepsy determination method, and an epilepsy determination program that can determine an epileptic seizure with higher accuracy can be provided.
本発明の実施形態に係るてんかん判定システムの機能ブロックを示す図である。It is a figure which shows the functional block of the epilepsy determination system which concerns on embodiment of this invention. 本実施形態に係るてんかん判定装置の物理的構成を示す図である。It is a figure which shows the physical structure of the epilepsy determination apparatus which concerns on this embodiment. 本実施形態に係るてんかん判定装置の学習済みモデルに入力される画像の一例である。It is an example of the image input into the learned model of the epilepsy determination apparatus which concerns on this embodiment. 本実施形態に係るてんかん判定装置の判定精度を示す第１グラフである。It is a 1st graph which shows the determination accuracy of the epilepsy determination apparatus which concerns on this embodiment. 本実施形態に係るてんかん判定装置の判定精度を示す第２グラフである。It is a 2nd graph which shows the determination precision of the epilepsy determination apparatus which concerns on this embodiment. 本実施形態に係るてんかん判定システムにより実行されるてんかん判定処理のフローチャートである。It is a flowchart of the epilepsy determination process performed by the epilepsy determination system which concerns on this embodiment. 本実施形態に係るてんかん判定システムにより実行される学習処理のフローチャートである。It is a flowchart of the learning process performed by the epilepsy determination system which concerns on this embodiment.
　添付図面を参照して、本発明の実施形態について説明する。なお、各図において、同一の符号を付したものは、同一又は同様の構成を有する。 Embodiments of the present invention will be described with reference to the accompanying drawings. In addition, in each figure, what attached | subjected the same code | symbol has the same or similar structure.
　図１は、本発明の実施形態に係るてんかん判定システム１の機能ブロックを示す図である。てんかん判定システム１は、てんかん判定装置１０と、測定装置２０とを備える。測定装置２０は、対象者の脳波を測定する装置であり、対象者の頭部に装着した複数の電極により、対象者の脳の電気活動を測定する。測定装置２０は、例えば国際１０－２０法に従って配置された２１個の電極を有してよいが、電極の数や配置は任意である。 FIG. 1 is a diagram showing functional blocks of an epilepsy determination system 1 according to an embodiment of the present invention. The epilepsy determination system 1 includes an epilepsy determination device 10 and a measurement device 20. The measuring device 20 is a device that measures the subject's brain waves, and measures the electrical activity of the subject's brain with a plurality of electrodes attached to the subject's head. The measuring device 20 may have, for example, 21 electrodes arranged according to the international 10-20 method, but the number and arrangement of electrodes are arbitrary.
　てんかん判定装置１０は、取得部１１、生成部１２、判定部１３、評価部１４、設定部１５、学習部１６、通知部１７及び記憶部１８を有する。取得部１１は、対象者の脳波データを取得する。取得部１１は、測定装置２０によって測定された対象者の脳波データを取得してよいが、予め記憶部１８又は他の記憶装置に記憶された対象者の脳波データを取得してもよい。 The epilepsy determination apparatus 10 includes an acquisition unit 11, a generation unit 12, a determination unit 13, an evaluation unit 14, a setting unit 15, a learning unit 16, a notification unit 17, and a storage unit 18. The acquisition unit 11 acquires brain wave data of the subject. The acquisition unit 11 may acquire the subject's brain wave data measured by the measurement device 20, but may acquire the subject's brain wave data stored in advance in the storage unit 18 or another storage device.
　生成部１２は、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成する。ここで、所定の時間幅は、任意に設定できるものであるが、例えば、０．５秒、１秒、２秒、５秒及び１０秒等であってよい。例えば、所定の時間幅が１０秒であり、脳波データの記録時間が１時間（３６００秒）である場合、生成部１２は、０～１０秒の範囲の脳波データを切り出して脳波を表す第１画像を生成し、１～１１秒の範囲の脳波データを切り出して脳波を表す第２画像を生成し、この処理を続けて、最後に３５９０～３６００秒の脳波データを切り出して脳波を表す第３５９１画像を生成してよい。生成部１２は、複数の画像により切り出される脳波データの範囲が重複するように、切り出す範囲を等間隔にずらすこととしてよいが、複数の画像により切り出される脳波データの範囲が重複しないようにしてもよい。 The generating unit 12 cuts out the electroencephalogram data with a predetermined time width and generates a plurality of images representing the electroencephalogram. Here, the predetermined time width can be arbitrarily set, but may be, for example, 0.5 seconds, 1 second, 2 seconds, 5 seconds, and 10 seconds. For example, when the predetermined time width is 10 seconds and the recording time of the electroencephalogram data is 1 hour (3600 seconds), the generation unit 12 cuts out the electroencephalogram data in the range of 0 to 10 seconds to express the first electroencephalogram. An image is generated, brain wave data in the range of 1 to 11 seconds is cut out to generate a second image representing the brain wave, and this process is continued. Finally, brain wave data of 3590 to 3600 seconds is cut out to represent the brain wave 3591 An image may be generated. The generation unit 12 may shift the cut-out range at equal intervals so that the ranges of the electroencephalogram data cut out by the plurality of images overlap, but the range of the electroencephalogram data cut out by the plurality of images may not overlap. Good.
　生成部１２は、脳波データに対してローパスフィルタ、ハイパスフィルタ及びノッチフィルタのうち少なくともいずれかを施した後、脳波データを所定の時間幅で切り出して、複数の画像を生成してよい。ここで、ローパスフィルタは、例えば６０Ｈｚ以下の周波数のデータを通過させるフィルタであってよく、ハイパスフィルタは、例えば０．５Ｈｚ以上の周波数のデータを通過させるフィルタであってよく、ノッチフィルタは、５０Ｈｚの周波数のデータを阻止するフィルタであってよい。これにより、脳波データに含まれることがあるノイズを除去した上で複数の画像を生成することができ、より高い精度でてんかん発作を判定することができる。 The generation unit 12 may generate a plurality of images by applying at least one of a low-pass filter, a high-pass filter, and a notch filter to the electroencephalogram data, and then cutting out the electroencephalogram data with a predetermined time width. Here, the low-pass filter may be a filter that passes data with a frequency of 60 Hz or less, for example, the high-pass filter may be a filter that passes data with a frequency of 0.5 Hz or more, and the notch filter is 50 Hz. It may be a filter that blocks data of a certain frequency. Thereby, it is possible to generate a plurality of images after removing noise that may be included in the electroencephalogram data, and to determine an epileptic seizure with higher accuracy.
　判定部１３は、複数の画像を学習済みモデル１３ａに入力し、学習済みモデル１３ａによって、複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させる。本実施形態に係るてんかん判定装置１０において、学習済みモデル１３ａは、学習済みの畳み込みニューラルネットワーク（Convolutional Neural Network:ＣＮＮ）である。学習済みモデル１３ａとして学習済みの畳み込みニューラルネットワークを用いることで、人と同程度かそれ以上の判定精度を達成することができ、より高い精度でてんかん発作を判定することができる。 The determination unit 13 inputs a plurality of images to the learned model 13a, and determines whether the brain wave of an epileptic seizure appears in any of the plurality of images by the learned model 13a. In the epilepsy determination apparatus 10 according to the present embodiment, the learned model 13a is a learned convolutional neural network (Convolutional Neural Network: CNN). By using a learned convolutional neural network as the learned model 13a, it is possible to achieve a determination accuracy equal to or higher than that of a human and to determine epileptic seizures with higher accuracy.
　評価部１４は、脳波を表す複数のテスト用画像を学習済みモデル１３ａに入力して、学習済みモデル１３ａの判定精度を評価する。ここで、評価部１４は、複数の指標に基づいて学習済みモデル１３ａの判定精度を評価してよい。これにより、学習済みモデル１３ａの判定精度を多面的に評価することができ、脳波データを切り出す時間幅をより適切に調整することができる。 The evaluation unit 14 inputs a plurality of test images representing brain waves to the learned model 13a and evaluates the determination accuracy of the learned model 13a. Here, the evaluation unit 14 may evaluate the determination accuracy of the learned model 13a based on a plurality of indices. As a result, the determination accuracy of the learned model 13a can be evaluated in a multifaceted manner, and the time width for extracting the electroencephalogram data can be adjusted more appropriately.
　より具体的には、複数の指標は、てんかん発作が表れている画像について、てんかん発作が表れていると正しく判定できるかを示す指標と、てんかん発作が表れていない画像について、てんかん発作が表れていないと正しく判定できるかを示す指標と、てんかん発作が表れている画像について、てんかん発作が表れていないと誤って判定するかを示す指標と、てんかん発作が表れていない画像について、てんかん発作が表れていると誤って判定するかを示す指標と、を含んでよい。このように、トゥルー・ポジティブ、トゥルー・ネガティブ、フォールス・ネガティブ及びフォールス・ポジティブの場合について学習済みモデル１３ａの判定精度を多面的に評価することで、判定精度を多面的に評価することができ、脳波データを切り出す時間幅をより適切に調整することができる。 More specifically, there are two or more indicators that indicate whether an epileptic seizure appears correctly for an image showing an epileptic seizure, and an epileptic seizure for an image that does not show an epileptic seizure. If there is an index that indicates that the epileptic seizure does not appear, an index that indicates that the epileptic seizure does not appear, and an image that does not show the epileptic seizure appear. And an index indicating whether it is erroneously determined that it is present. In this way, by evaluating the determination accuracy of the learned model 13a in a multifaceted manner in the case of true positive, true negative, false negative and false positive, the determination accuracy can be evaluated in a multifaceted manner, The time width for extracting the electroencephalogram data can be adjusted more appropriately.
　設定部１５は、評価部１４による評価に基づいて、所定の時間幅の設定を行う。設定部１５は、評価部１４による評価が向上するように、所定の時間幅を設定してよい。 The setting unit 15 sets a predetermined time width based on the evaluation by the evaluation unit 14. The setting unit 15 may set a predetermined time width so that the evaluation by the evaluation unit 14 is improved.
　学習部１６は、設定部１５により設定された所定の時間幅で切り出された、脳波を表す複数の学習用画像と、複数の学習用画像がてんかん発作の脳波を表しているか否かに関する情報と、を学習用データとして学習モデルを学習させ、新たな学習済みモデルを生成する。ここで、脳波を表す複数の学習用画像は、過去に測定装置２０により測定され、てんかん発作を表しているか否かに関する情報が関連付けられた画像であってよい。学習部１６は、学習モデルが畳み込みニューラルネットワーク等のニューラルネットワークの場合、誤差逆伝播法によってニューラルネットワークの重み係数の学習処理を行ってよい。このようにして、設定部１５により脳波データを切り出す時間幅を調整して、学習部１６によって判定精度が向上するように新たな学習済みモデルを生成することができ、新たな学習済みモデルを判定部１３に実装することで、てんかん判定装置１０によってより高い精度でてんかん発作を判定することができる。 The learning unit 16 includes a plurality of learning images representing brain waves, which are cut out with a predetermined time width set by the setting unit 15, and information regarding whether or not the plurality of learning images represent brain waves of epileptic seizures. , And learning data as learning data, a new learned model is generated. Here, the plurality of learning images representing the electroencephalograms may be images that have been measured by the measurement device 20 in the past and associated with information regarding whether or not they represent an epileptic seizure. When the learning model is a neural network such as a convolutional neural network, the learning unit 16 may perform learning processing of the weighting coefficient of the neural network by the error back propagation method. In this way, the setting unit 15 adjusts the time width for extracting the electroencephalogram data, and the learning unit 16 can generate a new learned model so as to improve the determination accuracy. By installing in the unit 13, the epilepsy seizure can be determined with higher accuracy by the epilepsy determination device 10.
　通知部１７は、判定部１３により、複数の画像のいずれかにてんかん発作の脳波が表れていると判定された場合に、所定の端末に通知する。ここで、所定の端末は、任意の情報処理端末であってよいが、例えば、専門医の用いる端末であったり、対象者を看護する者が用いる端末であったり、対象者自身が用いる端末であったりしてよい。てんかん発作の脳波が表れていると判定された場合に、所定の端末に通知することで、脳波の確認作業の負担を軽減することができる。 The notification unit 17 notifies a predetermined terminal when the determination unit 13 determines that any one of the plurality of images shows an epileptic brain wave. Here, the predetermined terminal may be any information processing terminal. For example, the predetermined terminal may be a terminal used by a specialist, a terminal used by a person who cares for the subject, or a terminal used by the subject himself / herself. You may do it. When it is determined that an electroencephalogram of an epileptic seizure appears, notification to a predetermined terminal can reduce the burden of confirming the electroencephalogram.
　記憶部１８は、学習部１６により用いられる学習用データを記憶してよい。もっとも、学習部１６及び記憶部１８は、てんかん判定装置１０に備えられていなくてもよく、学習モデルの学習処理は、通信ネットワークを介してアクセス可能な他の装置によって実行されてもよい。その場合、てんかん判定装置１０は、学習処理を行う他の装置に対して学習処理の条件を送信し、学習済みモデルを構成するための情報を受信してよい。学習済みモデルを構成するための情報は、学習モデルを特定する情報であり、ニューラルネットワークの場合、レイヤーの数、レイヤーの種類、レイヤー間のノードの接続、ノードの接続の重み係数（閾値を含む）及び活性化関数の種類等を含んでよい。 The storage unit 18 may store learning data used by the learning unit 16. But the learning part 16 and the memory | storage part 18 may not be provided in the epilepsy determination apparatus 10, and the learning process of a learning model may be performed by the other apparatus accessible via a communication network. In that case, the epilepsy determination apparatus 10 may transmit the learning process condition to another apparatus that performs the learning process, and may receive information for configuring the learned model. The information for configuring the learned model is information for identifying the learning model. In the case of a neural network, the number of layers, the type of layer, the connection of nodes between layers, and the weight coefficient of node connection (including threshold values) ) And the type of activation function.
　図２は、本実施形態に係るてんかん判定装置１０の物理的構成を示す図である。てんかん判定装置１０は、演算部に相当するＣＰＵ（Central Processing Unit）１０ａと、記憶部に相当するＲＡＭ（Random Access Memory）１０ｂと、記憶部に相当するＲＯＭ（Read only Memory）１０ｃと、通信部１０ｄと、入力部１０ｅと、表示部１０ｆと、を有する。これらの各構成は、バスを介して相互にデータ送受信可能に接続される。なお、本例ではてんかん判定装置１０が一台のコンピュータで構成される場合について説明するが、てんかん判定装置１０は、複数のコンピュータが組み合わされて実現されてもよい。また、図２で示す構成は一例であり、てんかん判定装置１０はこれら以外の構成を有してもよいし、これらの構成のうち一部を有さなくてもよい。 FIG. 2 is a diagram illustrating a physical configuration of the epilepsy determination device 10 according to the present embodiment. The epilepsy determination device 10 includes a CPU (Central Processing Unit) 10a corresponding to a calculation unit, a RAM (Random Access Memory) 10b corresponding to a storage unit, a ROM (Read only Memory) 10c corresponding to a storage unit, and a communication unit. 10d, an input unit 10e, and a display unit 10f. Each of these components is connected to each other via a bus so that data can be transmitted and received. In addition, although the case where the epilepsy determination apparatus 10 is comprised with one computer is demonstrated in this example, the epilepsy determination apparatus 10 may be implement | achieved combining a some computer. Moreover, the structure shown in FIG. 2 is an example, and the epilepsy determination apparatus 10 may have a structure other than these, and it is not necessary to have a part of these structures.
　ＣＰＵ１０ａは、ＲＡＭ１０ｂ又はＲＯＭ１０ｃに記憶されたプログラムの実行に関する制御やデータの演算、加工を行う制御部である。ＣＰＵ１０ａは、脳波を表す画像に基づいててんかんを判定するプログラム（てんかん判定プログラム）を実行する演算部である。ＣＰＵ１０ａは、入力部１０ｅや通信部１０ｄから種々のデータを受け取り、データの演算結果を表示部１０ｆに表示したり、ＲＡＭ１０ｂやＲＯＭ１０ｃに格納したりする。 The CPU 10a is a control unit that performs control related to execution of a program stored in the RAM 10b or the ROM 10c, and calculates and processes data. The CPU 10a is a calculation unit that executes a program for determining epilepsy (epilepsy determination program) based on an image representing an electroencephalogram. The CPU 10a receives various data from the input unit 10e and the communication unit 10d, and displays a calculation result of the data on the display unit 10f or stores it in the RAM 10b or the ROM 10c.
　ＲＡＭ１０ｂは、記憶部のうちデータの書き換えが可能なものであり、例えば半導体記憶素子で構成されてよい。ＲＡＭ１０ｂは、ＣＰＵ１０ａが実行するてんかん判定プログラムを記憶してよい。なお、これらは例示であって、ＲＡＭ１０ｂには、これら以外のデータが記憶されていてもよいし、これらの一部が記憶されていなくてもよい。 The RAM 10b can rewrite data in the storage unit, and may be composed of, for example, a semiconductor storage element. The RAM 10b may store an epilepsy determination program executed by the CPU 10a. These are examples, and the RAM 10b may store data other than these, or some of them may not be stored.
　ＲＯＭ１０ｃは、記憶部のうちデータの読み出しが可能なものであり、例えば半導体記憶素子で構成されてよい。ＲＯＭ１０ｃは、例えば探索プログラムや、書き換えが行われないデータを記憶してよい。 The ROM 10c is capable of reading data out of the storage unit, and may be composed of, for example, a semiconductor storage element. The ROM 10c may store, for example, a search program and data that is not rewritten.
　通信部１０ｄは、てんかん判定装置１０を他の機器に接続するインターフェースである。通信部１０ｄは、インターネットやＬＡＮ（Local Area Network）等の通信ネットワークに接続されてよい。 The communication unit 10d is an interface that connects the epilepsy determination device 10 to another device. The communication unit 10d may be connected to a communication network such as the Internet or a LAN (Local Area Network).
　入力部１０ｅは、ユーザからデータの入力を受け付けるものであり、例えば、キーボード及びタッチパネルを含んでよい。 The input unit 10e receives data input from the user, and may include, for example, a keyboard and a touch panel.
　表示部１０ｆは、ＣＰＵ１０ａによる演算結果を視覚的に表示するものであり、例えば、ＬＣＤ（Liquid Crystal Display）により構成されてよい。表示部１０ｆは、例えば脳波を表す画像や判定結果を表示してよい。 The display unit 10f visually displays the calculation result by the CPU 10a, and may be configured by, for example, an LCD (Liquid Crystal Display). The display unit 10f may display, for example, an image representing an electroencephalogram or a determination result.
　てんかん判定プログラムは、ＲＡＭ１０ｂやＲＯＭ１０ｃ等のコンピュータによって読み取り可能な記憶媒体に記憶されて提供されてもよいし、通信部１０ｄにより接続される通信ネットワークを介して提供されてもよい。てんかん判定装置１０では、ＣＰＵ１０ａがてんかん判定プログラムを実行することにより、図１を用いて説明した様々な動作が実現される。なお、これらの物理的な構成は例示であって、必ずしも独立した構成でなくてもよい。例えば、てんかん判定装置１０は、ＣＰＵ１０ａとＲＡＭ１０ｂやＲＯＭ１０ｃが一体化したＬＳＩ（Large-Scale Integration）を備えていてもよい。 The epilepsy determination program may be provided by being stored in a computer-readable storage medium such as the RAM 10b or the ROM 10c, or may be provided via a communication network connected by the communication unit 10d. In the epilepsy determination device 10, various operations described with reference to FIG. 1 are realized by the CPU 10 a executing the epilepsy determination program. In addition, these physical structures are illustrations, Comprising: It does not necessarily need to be an independent structure. For example, the epilepsy determination apparatus 10 may include an LSI (Large-Scale Integration) in which a CPU 10a, a RAM 10b, and a ROM 10c are integrated.
　図３は、本実施形態に係るてんかん判定装置１０の学習済みモデルに入力される画像Ｐの一例である。画像Ｐは、生成部１２によって脳波データから切り出された画像の一例であり、本例では１０秒間の脳波データを切り出している。 FIG. 3 is an example of the image P input to the learned model of the epilepsy determination apparatus 10 according to the present embodiment. The image P is an example of an image cut out from the electroencephalogram data by the generation unit 12. In this example, the electroencephalogram data for 10 seconds is cut out.
　画像Ｐは、測定装置２０が有する２１の電極により測定された脳波と、心電の波形とを示している。２１の電極により測定された脳波は、グラフの最上部から２１番目までの波形であり、心電の波形は、グラフの最下部に示された波形である。 The image P shows an electroencephalogram measured by 21 electrodes of the measuring device 20 and an electrocardiogram waveform. The electroencephalogram measured by the 21 electrodes is the waveform from the top of the graph to the 21st, and the electrocardiogram waveform is the waveform shown at the bottom of the graph.
　画像Ｐは、脳波データを測定した複数の電極に対応する複数の色で示された波形を含む。本例では、上下に隣り合って描画される波形が異なる色で示されたり、電極の配置に対応するように波形の色が変えられたりしているが、色の選択は任意である。このように色分けすることで、脳波データを測定した複数の電極を区別して、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。 Image P includes waveforms indicated by a plurality of colors corresponding to a plurality of electrodes from which electroencephalogram data was measured. In this example, waveforms drawn adjacent to each other in the vertical direction are shown in different colors, or the color of the waveform is changed so as to correspond to the arrangement of the electrodes, but the selection of the color is arbitrary. By color-coding in this way, it is possible to distinguish between multiple electrodes for which electroencephalogram data has been measured, and to make determinations under the same conditions as specialists diagnose epileptic seizures, and to determine epileptic seizures with higher accuracy Can do.
　図４は、本実施形態に係るてんかん判定装置１０の判定精度を示す第１グラフである。第１グラフは、２４人の対象者について、それぞれ数十時間にわたって脳波データを測定し、本実施形態に係るてんかん判定装置１０によって正しくてんかん発作を判定できるかテストした結果を示している。第１グラフは、横軸に設定部１５により設定した時間幅（Time Window）を示し、縦軸に設定された時間幅で学習処理が行われた学習済みモデルの判定精度を示している。判定精度は、評価部１４により評価されたトゥルー・ポジティブ（True Positive）の値である。すなわち、第１グラフで示す判定精度は、てんかん発作が表れている画像について、てんかん発作が表れていると正しく判定できるかを示す指標に基づき評価された判定精度である。 FIG. 4 is a first graph showing the determination accuracy of the epilepsy determination apparatus 10 according to the present embodiment. The first graph shows the result of testing whether or not an epileptic seizure can be correctly determined by the epilepsy determination apparatus 10 according to the present embodiment by measuring brain wave data for several tens of hours for each of 24 subjects. In the first graph, the horizontal axis indicates the time width (Time Window) set by the setting unit 15, and the vertical axis indicates the determination accuracy of the learned model in which the learning process is performed with the time width set. The determination accuracy is a true positive value evaluated by the evaluation unit 14. That is, the determination accuracy shown in the first graph is a determination accuracy evaluated based on an index indicating whether or not an image showing an epileptic seizure can be correctly determined if an epileptic seizure appears.
　第１グラフによると、時間幅が０．５秒の場合、箱ひげ図で表される中央値は０程度であり、上側四分位点は０．１よりやや大きく、最大値は０．２程度である。また、時間幅が１秒の場合、箱ひげ図で表される最小値は０程度であり、下側四分位点は０．５程度であり、中央値は０．７よりやや大きく、上側四分位点は０．９程度であり、最大値は１．０程度である。また、時間幅が２秒の場合、箱ひげ図で表される最小値は０程度であり、下側四分位点は０．４程度であり、中央値は０．７程度であり、上側四分位点は０．９程度であり、最大値は１．０程度である。また、時間幅が５秒の場合、箱ひげ図で表される最小値は０程度であり、下側四分位点は０．５程度であり、中央値は０．７よりやや小さく、上側四分位点は０．９よりやや小さく、最大値は１．０よりやや小さい程度である。また、時間幅が１０秒の場合、箱ひげ図で表される最小値は０程度であり、下側四分位点は０．２程度であり、中央値は０．６程度であり、上側四分位点は０．８よりやや小さく、最大値は１．０程度である。 According to the first graph, when the time width is 0.5 seconds, the median represented by the box plot is about 0, the upper quartile is slightly larger than 0.1, and the maximum value is 0.2. Degree. When the time width is 1 second, the minimum value represented by the box plot is about 0, the lower quartile is about 0.5, the median is slightly larger than 0.7, The quartile is about 0.9, and the maximum value is about 1.0. When the time width is 2 seconds, the minimum value represented by the box plot is about 0, the lower quartile is about 0.4, the median is about 0.7, The quartile is about 0.9, and the maximum value is about 1.0. Also, when the time width is 5 seconds, the minimum value represented by the box plot is about 0, the lower quartile is about 0.5, the median is slightly smaller than 0.7, and the upper side The quartile is slightly smaller than 0.9 and the maximum value is slightly smaller than 1.0. When the time width is 10 seconds, the minimum value represented by the box plot is about 0, the lower quartile is about 0.2, the median is about 0.6, The quartile is slightly smaller than 0.8 and the maximum value is about 1.0.
　このように、脳波データを切り出す時間幅は、学習モデルの学習処理が適切に行えるか否かに影響し、判定精度を変動させる要因となることがある。本実施形態に係るてんかん判定装置１０では、脳波データを切り出す時間幅を調整して、判定精度が向上するように新たな学習済みモデルを生成することができ、より高い精度でてんかん発作を判定することができる。 As described above, the time width for cutting out the electroencephalogram data affects whether or not the learning process of the learning model can be performed appropriately, and may cause the determination accuracy to fluctuate. In the epilepsy determination apparatus 10 according to the present embodiment, it is possible to generate a new learned model so as to improve the determination accuracy by adjusting the time width for extracting the electroencephalogram data, and to determine the epileptic seizure with higher accuracy. be able to.
　図５は、本実施形態に係るてんかん判定装置１０の判定精度を示す第２グラフである。第２グラフは、２４人の対象者について、それぞれ数十時間にわたって脳波データを測定し、本実施形態に係るてんかん判定装置１０によって正しく正常状態を判定できるかテストした結果を示している。第２グラフは、横軸に設定部１５により設定した時間幅（Time Window）を示し、縦軸に設定された時間幅で学習処理が行われた学習済みモデルの判定精度を示している。判定精度は、評価部１４により評価されたトゥルー・ネガティブ（True Negative）の値である。すなわち、第２グラフで示す判定精度は、てんかん発作が表れていない画像について、てんかん発作が表れていないと正しく判定できるかを示す指標に基づき評価された判定精度である。 FIG. 5 is a second graph showing the determination accuracy of the epilepsy determination apparatus 10 according to the present embodiment. The second graph shows the results of testing the brain wave data for several tens of hours for 24 subjects and testing whether the normal state can be correctly determined by the epilepsy determination device 10 according to the present embodiment. In the second graph, the horizontal axis indicates the time width (Time 設定 Window) set by the setting unit 15, and the vertical axis indicates the determination accuracy of the learned model in which the learning process is performed with the time width set. The determination accuracy is a true negative value evaluated by the evaluation unit 14. That is, the determination accuracy shown in the second graph is a determination accuracy evaluated based on an index indicating whether it is possible to correctly determine that an epileptic seizure does not appear for an image in which the epileptic seizure does not appear.
　第２グラフによると、時間幅が０．５秒、１秒、２秒、５秒及び１０秒の場合、箱ひげ図で表される最小値は０．９７～０．９９程度であり、下側四分位点は０．９８～０．９９程度であり、中央値は０．９９程度であり、上側四分位点は０．９９５程度であり、最大値は１．０程度である。 According to the second graph, when the time width is 0.5 second, 1 second, 2 seconds, 5 seconds and 10 seconds, the minimum value represented by the box plot is about 0.97 to 0.99, The side quartile is about 0.98 to 0.99, the median is about 0.99, the upper quartile is about 0.995, and the maximum value is about 1.0.
　図４の第１グラフと図５の第２グラフを比較すると、脳波データを切り出す時間幅と、判定精度との関係は、判定精度の評価に用いる指標によって異なることが確認できる。より具体的には、第１グラフでは、時間幅が０．５秒の場合に判定精度の中央値が０程度であるのに対して、第２グラフでは１．０に近いという違いがある。また、第１グラフでは、時間幅を１秒から１０秒まで長くしていくと判定精度の中央値が低下していくのに対して、第２グラフでは、時間幅を１秒から１０秒まで長くしていくと判定精度の中央値が上昇していくという違いがある。本実施形態に係るてんかん判定装置１０は、複数の指標に基づいて学習済みモデルの判定精度を評価して、複数の指標に基づいて評価された判定精度をバランス良く向上させるように、脳波データを切り出す時間幅を設定してよい。例えば、複数の指標に基づいて評価された判定精度の総和を最大化するように、脳波データを切り出す時間幅を設定してよい。 Comparing the first graph of FIG. 4 and the second graph of FIG. 5, it can be confirmed that the relationship between the time width for extracting the electroencephalogram data and the determination accuracy differs depending on the index used for the evaluation of the determination accuracy. More specifically, the first graph has a difference that the median value of determination accuracy is about 0 when the time width is 0.5 seconds, whereas the second graph is close to 1.0. In the first graph, the median determination accuracy decreases as the time width is increased from 1 second to 10 seconds, whereas in the second graph, the time width is increased from 1 second to 10 seconds. There is a difference that the median of the determination accuracy increases as the length increases. The epilepsy determination apparatus 10 according to the present embodiment evaluates the determination accuracy of a learned model based on a plurality of indexes, and the electroencephalogram data is improved so as to improve the determination accuracy evaluated based on the plurality of indexes in a well-balanced manner. You may set the time range to cut out. For example, you may set the time width which cuts out electroencephalogram data so that the sum total of the determination accuracy evaluated based on the some parameter | index may be maximized.
　図６は、本実施形態に係るてんかん判定システム１により実行されるてんかん判定処理のフローチャートである。はじめに、測定装置２０によって、対象者の脳波データを測定する（Ｓ１０）。以降の処理は、脳波データが蓄積された後に実行してもよいし、脳波データを測定しながら逐次実行してもよい。 FIG. 6 is a flowchart of epilepsy determination processing executed by the epilepsy determination system 1 according to the present embodiment. First, the brain wave data of the subject is measured by the measuring device 20 (S10). The subsequent processing may be executed after the electroencephalogram data is accumulated, or may be executed sequentially while measuring the electroencephalogram data.
　てんかん判定装置１０の生成部１２は、取得した脳波データにフィルタを施して前処理を実行する（Ｓ１１）。その後、生成部１２は、脳波データを所定の時間幅で切り出して、複数の画像を生成する（Ｓ１２）。 The generation unit 12 of the epilepsy determination apparatus 10 performs preprocessing by filtering the acquired electroencephalogram data (S11). After that, the generation unit 12 cuts out the electroencephalogram data with a predetermined time width and generates a plurality of images (S12).
　てんかん判定装置１０の判定部１３は、複数の画像をＣＮＮ等の学習済みモデルに入力し、てんかん発作の脳波が表れているか判定する（Ｓ１３）。ここで、てんかん発作有りと判定された場合（Ｓ１４：ＹＥＳ）、てんかん判定装置１０の通知部１７は、所定の端末にてんかん発作が検出されたことを通知する（Ｓ１５）。一方、てんかん発作有りと判定されない場合（Ｓ１４：ＮＯ）、てんかん判定処理は終了する。 The determination unit 13 of the epilepsy determination apparatus 10 inputs a plurality of images to a learned model such as CNN, and determines whether an electroencephalogram of an epileptic seizure appears (S13). If it is determined that there is an epileptic seizure (S14: YES), the notification unit 17 of the epilepsy determination device 10 notifies that an epileptic seizure has been detected at a predetermined terminal (S15). On the other hand, when it is not determined that there is an epileptic seizure (S14: NO), the epilepsy determination process ends.
　本実施形態に係るてんかん判定装置１０によれば、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成して、複数の画像のいずれかにてんかん発作の脳波が表れているかを学習済みモデルに判定させることで、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。 According to the epilepsy determination device 10 according to the present embodiment, brain wave data is cut out with a predetermined time width, a plurality of images representing brain waves are generated, and brain waves of epileptic seizures appear in any of the plurality of images. By making the learned model determine, the determination can be made under the same conditions as the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy.
　また、本実施形態に係るてんかん判定システム１によれば、脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成して、複数の画像のいずれかにてんかん発作の脳波が表れているかを学習済みモデルに判定させることで、専門医がてんかん発作を診断するのと同様の条件で判定を行うことができ、より高い精度でてんかん発作を判定することができる。また、てんかん発作の脳波が表れていると判定された場合に、所定の端末に通知することで、脳波の確認作業の負担を軽減することができる。 Further, according to the epilepsy determination system 1 according to the present embodiment, brain wave data is cut out with a predetermined time width to generate a plurality of images representing the brain waves, and the brain wave of an epileptic seizure appears in one of the plurality of images. By making the learned model determine whether or not the epileptic seizure is detected, it is possible to perform the determination under the same conditions as when the specialist diagnoses the epileptic seizure, and the epileptic seizure can be determined with higher accuracy. In addition, when it is determined that an electroencephalogram of an epileptic seizure appears, the burden of the electroencephalogram confirmation work can be reduced by notifying a predetermined terminal.
　図７は、本実施形態に係るてんかん判定システム１により実行される学習処理のフローチャートである。はじめに、てんかん判定装置１０の評価部１４は、テスト用画像を学習済みモデルに入力して、判定精度を評価する（Ｓ２０）。設定部１５は、評価部１４による評価に基づいて、脳波データを切り出す時間幅を設定する（Ｓ２１）。 FIG. 7 is a flowchart of the learning process executed by the epilepsy determination system 1 according to this embodiment. First, the evaluation unit 14 of the epilepsy determination apparatus 10 inputs a test image to the learned model and evaluates the determination accuracy (S20). The setting unit 15 sets a time width for cutting out the electroencephalogram data based on the evaluation by the evaluation unit 14 (S21).
　学習部１６は、設定された時間幅で脳波データを切り出して、学習モデルの学習処理を実行し、新たな学習済みモデルを生成する（Ｓ２２）。そして、てんかん判定装置１０は、新たな学習済みモデルを判定部１３に実装する（Ｓ２３）。以上により、学習処理が終了する。 The learning unit 16 cuts out the electroencephalogram data with the set time width, executes the learning process of the learning model, and generates a new learned model (S22). Then, the epilepsy determination device 10 mounts a new learned model in the determination unit 13 (S23). Thus, the learning process ends.
　このように、脳波データを切り出す時間幅を調整して、判定精度が向上するように新たな学習済みモデルを生成することができ、より高い精度でてんかん発作を判定することができる。また、学習用データが蓄積するに従って、より判定精度の高い学習済みモデルを生成して、判定部１３に実装する学習済みモデルを更新していくことができ、てんかん判定システム１の運用を続けるほど、より高い精度でてんかん発作を判定することができるようになる。 Thus, it is possible to generate a new learned model so as to improve the determination accuracy by adjusting the time width for extracting the electroencephalogram data, and to determine epileptic seizures with higher accuracy. Further, as the learning data accumulates, it is possible to generate a learned model with higher determination accuracy and update the learned model to be implemented in the determination unit 13, so that the operation of the epilepsy determination system 1 is continued. Will be able to determine epileptic seizures with higher accuracy.
　以上説明した実施形態は、本発明の理解を容易にするためのものであり、本発明を限定して解釈するためのものではない。実施形態が備える各要素並びにその配置、材料、条件、形状及びサイズ等は、例示したものに限定されるわけではなく適宜変更することができる。また、異なる実施形態で示した構成同士を部分的に置換し又は組み合わせることが可能である。 The embodiment described above is for facilitating the understanding of the present invention, and is not intended to limit the present invention. Each element included in the embodiment and its arrangement, material, condition, shape, size, and the like are not limited to those illustrated, and can be changed as appropriate. In addition, the structures shown in different embodiments can be partially replaced or combined.

Claims (10)

  1. 　対象者の脳波データを取得する取得部と、
    　前記脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成する生成部と、
    　前記複数の画像を学習済みモデルに入力し、前記学習済みモデルによって、前記複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させる判定部と、
    　を備えるてんかん判定装置。     An acquisition unit for acquiring the subject's brain wave data;
    Generating a plurality of images representing brain waves by cutting out the electroencephalogram data with a predetermined time width;
    A determination unit that inputs the plurality of images into a learned model, and determines whether the brain wave of an epileptic seizure appears in any of the plurality of images by the learned model;
    An epilepsy determination device comprising:
  2. 　脳波を表す複数のテスト用画像を前記学習済みモデルに入力して、前記学習済みモデルの判定精度を評価する評価部と、
    　前記評価部による評価に基づいて、前記所定の時間幅の設定を行う設定部と、
    　前記設定部により設定された前記所定の時間幅で切り出された、脳波を表す複数の学習用画像と、前記複数の学習用画像がてんかん発作の脳波を表しているか否かに関する情報と、を学習用データとして学習モデルを学習させ、新たな学習済みモデルを生成する学習部と、
    　をさらに備える請求項１に記載のてんかん判定装置。     An evaluation unit that inputs a plurality of test images representing an electroencephalogram into the learned model, and evaluates the determination accuracy of the learned model;
    A setting unit configured to set the predetermined time width based on the evaluation by the evaluation unit;
    Learning a plurality of learning images representing brain waves clipped by the predetermined time width set by the setting unit, and information regarding whether or not the plurality of learning images represent brain waves of epileptic seizures A learning unit that learns a learning model as data for generation and generates a new learned model;
    The epilepsy determination apparatus according to claim 1, further comprising:
  3. 　前記評価部は、複数の指標に基づいて前記学習済みモデルの判定精度を評価する、
    　請求項２に記載のてんかん判定装置。     The evaluation unit evaluates the determination accuracy of the learned model based on a plurality of indices.
    The epilepsy determination apparatus according to claim 2.
  4. 　前記複数の指標は、
    　てんかん発作が表れている画像について、てんかん発作が表れていると正しく判定できるかを示す指標と、
    　てんかん発作が表れていない画像について、てんかん発作が表れていないと正しく判定できるかを示す指標と、
    　てんかん発作が表れている画像について、てんかん発作が表れていないと誤って判定するかを示す指標と、
    　てんかん発作が表れていない画像について、てんかん発作が表れていると誤って判定するかを示す指標と、を含む、
    　請求項３に記載のてんかん判定装置。     The plurality of indicators are:
    For an image showing an epileptic seizure, an index indicating whether it can be correctly determined that an epileptic seizure appears,
    For an image that does not show an epileptic seizure, an index indicating whether it can be correctly determined that no epileptic seizure appears,
    An index indicating whether an image showing an epileptic seizure is erroneously determined not to show an epileptic seizure,
    An image indicating whether an image that does not show an epileptic seizure is erroneously determined to show an epileptic seizure,
    The epilepsy determination apparatus according to claim 3.
  5. 　前記複数の画像は、前記脳波データを測定した複数の電極に対応する複数の色で示された波形を含む、
    　請求項１から４のいずれか一項に記載のてんかん判定装置。     The plurality of images include waveforms shown in a plurality of colors corresponding to a plurality of electrodes from which the electroencephalogram data was measured,
    The epilepsy determination device according to any one of claims 1 to 4.
  6. 　前記学習済みモデルは、学習済みの畳み込みニューラルネットワークである、
    　請求項１から５のいずれか一項に記載のてんかん判定装置。     The learned model is a learned convolutional neural network,
    The epilepsy determination apparatus according to any one of claims 1 to 5.
  7. 　前記生成部は、前記脳波データに対してローパスフィルタ、ハイパスフィルタ及びノッチフィルタのうち少なくともいずれかを施した後、前記脳波データを前記所定の時間幅で切り出して、前記複数の画像を生成する、
    　請求項１から６のいずれか一項に記載のてんかん判定装置。     The generation unit performs at least one of a low-pass filter, a high-pass filter, and a notch filter on the brain wave data, and then cuts out the brain wave data with the predetermined time width to generate the plurality of images.
    The epilepsy determination apparatus according to any one of claims 1 to 6.
  8. 　対象者の脳波データを測定する測定装置と、前記脳波データに基づいて、てんかんを判定するてんかん判定装置と、を備えるてんかん判定システムであって、
    　前記てんかん判定装置は、
    　前記脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成する生成部と、
    　前記複数の画像を学習済みモデルに入力し、前記学習済みモデルによって、前記複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させる判定部と、
    　前記判定部により、前記複数の画像のいずれかにてんかん発作の脳波が表れていると判定された場合に、所定の端末に通知する通知部と、を有する、
    　てんかん判定システム。     An epilepsy determination system comprising a measurement device that measures brain wave data of a subject, and an epilepsy determination device that determines epilepsy based on the electroencephalogram data,
    The epilepsy determination device
    Generating a plurality of images representing brain waves by cutting out the electroencephalogram data with a predetermined time width;
    A determination unit that inputs the plurality of images into a learned model, and determines whether the brain wave of an epileptic seizure appears in any of the plurality of images by the learned model;
    A notification unit for notifying a predetermined terminal when it is determined by the determination unit that an electroencephalogram of an epileptic seizure appears in any of the plurality of images;
    Epilepsy determination system.
  9. 　対象者の脳波データを取得することと、
    　前記脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成することと、
    　前記複数の画像を学習済みモデルに入力し、前記学習済みモデルによって、前記複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させることと、
    　を含むてんかん判定方法。     Acquiring brain wave data of the subject,
    Cutting out the electroencephalogram data with a predetermined time width to generate a plurality of images representing electroencephalograms;
    Inputting the plurality of images into a learned model, and causing the learned model to determine whether an electroencephalogram of an epileptic seizure appears in any of the plurality of images;
    Epilepsy determination method including
  10. 　てんかん判定装置に備えられたコンピュータを、
    　対象者の脳波データを取得する取得部、
    　前記脳波データを所定の時間幅で切り出して、脳波を表す複数の画像を生成する生成部、及び
    　前記複数の画像を学習済みモデルに入力し、前記学習済みモデルによって、前記複数の画像のいずれかにてんかん発作の脳波が表れているかを判定させる判定部、
    　として機能させるてんかん判定プログラム。     A computer equipped with an epilepsy determination device
    An acquisition unit for acquiring brain wave data of the subject,
    A generation unit that cuts out the electroencephalogram data with a predetermined time width and generates a plurality of images representing electroencephalograms, and inputs the plurality of images into a learned model, and the trained model selects any one of the plurality of images. A determination unit that determines whether an electroencephalogram of an epileptic seizure appears,
    Epilepsy determination program to function as
PCT/JP2019/011268 2018-03-19 2019-03-18 Epilepsy determination device, epilepsy determination system, epilepsy determination method and epilepsy determination program WO2019181887A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2018051262A JP7072771B2 (en) 2018-03-19 2018-03-19 Epilepsy determination device, epilepsy determination system, epilepsy determination method and epilepsy determination program
JP2018-051262 2018-03-19

Publications (1)

Publication Number Publication Date
WO2019181887A1 true WO2019181887A1 (en) 2019-09-26

Family

ID=67986252

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2019/011268 WO2019181887A1 (en) 2018-03-19 2019-03-18 Epilepsy determination device, epilepsy determination system, epilepsy determination method and epilepsy determination program

Country Status (2)

Country Link
JP (1) JP7072771B2 (en)
WO (1) WO2019181887A1 (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001309898A (en) * 2000-04-28 2001-11-06 Takumi Ikuta Method of neural waveform diagnosis with wavelet function and its equipment
JP2003310564A (en) * 2002-04-22 2003-11-05 Fuji Xerox Co Ltd Automatic brain wave analyzing apparatus and method
CN106821376A (en) * 2017-03-28 2017-06-13 南京医科大学 A kind of epileptic attack early warning system and method based on deep learning algorithm

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108403111A (en) 2018-02-01 2018-08-17 华中科技大学 A kind of epileptic electroencephalogram (eeg) recognition methods and system based on convolutional neural networks

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001309898A (en) * 2000-04-28 2001-11-06 Takumi Ikuta Method of neural waveform diagnosis with wavelet function and its equipment
JP2003310564A (en) * 2002-04-22 2003-11-05 Fuji Xerox Co Ltd Automatic brain wave analyzing apparatus and method
CN106821376A (en) * 2017-03-28 2017-06-13 南京医科大学 A kind of epileptic attack early warning system and method based on deep learning algorithm

Also Published As

Publication number Publication date
JP2019162237A (en) 2019-09-26
JP7072771B2 (en) 2022-05-23

Similar Documents

Publication Publication Date Title
Bridwell et al. Moving beyond ERP components: a selective review of approaches to integrate EEG and behavior
CN105636515B (en) Computer program for assessing inattentive EEG automatically and assessment equipment
Bhardwaj et al. A novel genetic programming approach for epileptic seizure detection
Boubchir et al. A review of feature extraction for EEG epileptic seizure detection and classification
US20150134580A1 (en) Method And System For Training A Neural Network
US20180279939A1 (en) Method and system for locating seizure focus from interictal data
JP5476137B2 (en) Human interface based on biological and brain function measurement
JP7337788B2 (en) Genome discovery for use in machine learning methods
EP3594854A1 (en) Method and system for clustering users using cognitive stress report for classifying stress levels
JP2022536552A (en) Seizure detection system and method based on changes in electroencephalogram (EEG) nonlinearity
Azami et al. Entropy analysis of univariate biomedical signals: Review and comparison of methods
Portoles et al. Characterizing synchrony patterns across cognitive task stages of associative recognition memory
Guerrero et al. Conex–Connect: Learning patterns in extremal brain connectivity from MultiChannel EEG data
CN115281685A (en) Sleep stage identification method and device based on anomaly detection and computer readable storage medium
WO2019181887A1 (en) Epilepsy determination device, epilepsy determination system, epilepsy determination method and epilepsy determination program
JP5207172B2 (en) Waveform analysis apparatus and waveform analysis program
Goovaerts et al. Advanced EEG Processing for the Detection of Drowsiness in Drivers.
US20230248295A1 (en) Method for selecting features from electroencephalogram signals
Scarciglia et al. Physiological noise: Definition, estimation, and characterization in complex biomedical signals
US11950919B2 (en) System for real-time measurement of the activity of a cognitive function and method for calibrating such a system
Wȩsierski et al. Rating by detection: an artifact detection protocol for rating EEG quality with average event duration
Grivel et al. Benefits of Zero-Phase or Linear Phase Filters to Design Multiscale Entropy: Theory and Application
Roldán et al. Characterization of entropy measures against data loss: application to EEG records
CN110840433B (en) Workload evaluation method weakly coupled with job task scene
US20230277118A1 (en) Methods and systems for detecting and assessing cognitive impairment

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: 19771277

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19771277

Country of ref document: EP

Kind code of ref document: A1