US-12616389-B2 - Estimation system, estimation method, program, estimation model, brain activity training apparatus, brain activity training method, and brain activity training program

US12616389B2US 12616389 B2US12616389 B2US 12616389B2US-12616389-B2

Abstract

An estimation system obtains brain wave measurement data and functional magnetic resonance imaging measurement data simultaneously measured from a subject, calculates first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data, calculates second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data, calculates a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated with the use of a plurality of second functional connectivities, and determines an estimation model for estimating disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label.

Inventors

Takeshi Ogawa
RYUTA TAMANO
Motoaki Kawanabe
Mitsuo Kawato

Assignees

ADVANCED TELECOMMUNICATIONS RESEARCH INSTITUTE INTERNATIONAL
SHIONOGI & CO., LTD.

Dates

Publication Date: 20260505
Application Date: 20210701
Priority Date: 20200702

Claims (18)

1 . An estimation system with one or more processors configured to: obtain, from an electroencephalograph, brain wave measurement data and, from a functional magnetic resonance imaging device, functional magnetic resonance imaging measurement data simultaneously measured from a subject, the brain wave measurement data including time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject; calculate first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data; calculate second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data; calculate a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities; and determine an estimation model for estimating the disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label; and outputting a signal including the disorder-likelihood to a presentation apparatus.
2 . The estimation system according to claim 1 , further configured to estimate the disorder-likelihood of the subject by inputting the brain wave measurement data measured from the subject into the estimation model.
3 . The estimation system according to claim 2 , further configured to calculate a second score in accordance with the estimated disorder-likelihood of the subject and present to the subject, information in accordance with the calculated second score.
4 . The estimation system according to claim 3 , wherein the estimation model is prepared for each disorder, and an estimation model corresponding to a disorder that manifests in the subject is applied to the subject.
5 . The estimation system according to claim 1 , wherein change in symptom of the subject is assessed based on a second score in accordance with an estimated disorder-likelihood of the subject.
6 . The estimation system according to claim 1 , wherein calculating the disorder-likelihood label comprises calculating the score representing the disorder-likelihood based on a sum of results of multiplication of the plurality of second functional connectivities brought in correspondence with disorder-likelihood to be estimated by respective corresponding weight parameters.
7 . The estimation system according to claim 6 , wherein calculating the disorder-likelihood label comprises normalizing the score representing the disorder-likelihood and subject the normalized score to threshold processing.
8 . The estimation system according to claim 1 , wherein the estimation model includes information for selecting first functional connectivity to be used for estimation among first functional connectivities for each channel combination and a weight parameter brought in correspondence with the selected first functional connectivity.
9 . The estimation system according to claim 1 , wherein calculating the first functional connectivity is based on a correlation value between time waveforms in a section included in a window set in common for time waveforms of brain waves in two channels of interest.
10 . The estimation system according to claim 1 , wherein calculating the first functional connectivity comprises calculating the first functional connectivity for each frequency band included in the brain wave measurement data and/or for each window size of a set window.
11 . The estimation system according to claim 10 , further configured to determine in advance in accordance with the subject, the frequency band included in the brain wave measurement data to be inputted to the estimation model and/or the window size.
12 . The estimation system according to claim 1 , wherein calculating the second functional connectivity is based on a correlation value between time waveforms in a section included in a window set in common for time waveforms indicating amounts of activities in two regions of interest.
13 . An estimation method comprising: obtaining, from an electroencephalograph, brain wave measurement data and, from a functional magnetic resonance imaging device, functional magnetic resonance imaging measurement data simultaneously measured from a subject, the brain wave measurement data including time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject; calculating first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data; calculating second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data; calculating a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities; determining an estimation model for estimating the disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label; and outputting a signal including the disorder-likelihood to a presentation apparatus.
14 . A non-transitory storage medium storing a program thereon, when executed by one or more processors, causing the one or more processors to perform: obtaining, from an electroencephalograph, brain wave measurement data and, from a functional magnetic resonance imaging device, functional magnetic resonance imaging measurement data simultaneously measured from a subject, the brain wave measurement data including time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject; calculating first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data; calculating second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data; calculating a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities; determining an estimation model for estimating the disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label; and outputting a signal including the disorder-likelihood to a presentation apparatus.
15 . A trained estimation model for estimating disorder-likelihood of a subject based on brain wave measurement data measured from the subject, processing for constructing the estimation model comprising: obtaining, from an electroencephalograph, brain wave measurement data and, from a functional magnetic resonance imaging device, functional magnetic resonance imaging measurement data simultaneously measured from the subject, the brain wave measurement data including time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject; calculating first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data; calculating second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data; calculating a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities; and determining the estimation model by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label; and outputting a signal including a disorder likelihood to a presentation apparatus.
16 . A brain activity training apparatus for conducting neurofeedback training, the brain activity training apparatus comprising: a storage device where an estimation model for estimating disorder-likelihood of a subject generated before the neurofeedback training is conducted is stored; an electroencephalograph configured to measure brain wave measurement data of the subject in the neurofeedback training, the brain wave measurement data including first time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject; a presentation apparatus; and a processing apparatus configured to calculate, in the neurofeedback training, disorder-likelihood of the subject with the estimation model based on measurement data from the electroencephalograph and outputs a signal for representation corresponding to the disorder-likelihood to the presentation apparatus, wherein the estimation model is generated by processing for obtaining, from the electroencephalograph, the brain wave measurement data and, from a functional magnetic resonance imaging device, functional magnetic resonance imaging measurement data simultaneously measured from the subject, the simultaneously measured brain wave measurement data including a second time waveform for each channel corresponding to each channel of the brain wave measurement data measured in the neurofeedback training, processing for calculating first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data, processing for calculating second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data, processing for calculating a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities, processing for determining the estimation model by estimating the disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label; and processing for outputting the disorder-likelihood by outputting a signal including the disorder-likelihood to the presentation apparatus.
17 . A brain activity training method for conducting neurofeedback training, the brain activity training method comprising: obtaining an estimation model for estimating disorder-likelihood of a subject generated before the neurofeedback training is conducted; measuring brain wave measurement data of the subject in the neurofeedback training, the brain wave measurement data including first time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject; and calculating, in the neurofeedback training, disorder-likelihood of the subject with the estimation model based on the brain wave measurement data and outputting a signal for representation corresponding to the disorder-likelihood to a presentation apparatus, wherein the obtaining an estimation model includes obtaining, from an electroencephalograph, the brain wave measurement data and, from a functional magnetic resonance imaging device, functional magnetic resonance imaging measurement data simultaneously measured from the subject, the simultaneously measured brain wave measurement data including a second time waveform for each channel corresponding to each channel of the brain wave measurement data measured in the neurofeedback training, calculating first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data, calculating second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data, calculating a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities, determining the estimation model by estimating the disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label; and outputting a signal including a disorder likelihood to a presentation apparatus.
18 . A non-transitory storage medium storing thereon a brain activity training program for conducting neurofeedback training, when executed by one or more processors, the brain activity training program causing the one or more processors to perform: storing an estimation model for estimating disorder-likelihood of a subject generated before the neurofeedback training is conducted; obtaining brain wave measurement data of the subject in the neurofeedback training, the brain wave measurement data including first time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject; and calculating, in the neurofeedback training, disorder-likelihood of the subject with the estimation model based on the brain wave measurement data and outputting a signal for representation corresponding to the disorder-likelihood to a presentation apparatus, wherein the estimation model is generated by processing for obtaining, from an electroencephalograph, the brain wave measurement data and, from a functional magnetic resonance imaging device, functional magnetic resonance imaging measurement data simultaneously measured from the subject, the simultaneously measured brain wave measurement data including a second time waveform for each channel corresponding to each channel of the brain wave measurement data measured in the neurofeedback training, processing for calculating first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data, processing for calculating second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data, processing for calculating a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities, processing for determining the estimation model by estimating the disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label; and outputting a signal including a disorder likelihood to a presentation apparatus.

Description

TECHNICAL FIELD The present invention relates to a technology for estimating disorder-likelihood based on measurement data on a brain activity. BACKGROUND ART Neurofeedback training or the like aiming at estimation of a brain function with the use of functional magnetic resonance imaging (which will also be abbreviated as “fMRI” below) representing one of techniques for non-invasive measurement of a brain activity and modulation of the brain function has been known. The neurofeedback training only with fMRI faces a challenge in viability such as cost. Then, a method of combining an electromagnetic field measurement method such as electroencephalogram (which will also be abbreviated as “EEG” below) and fMRI has been proposed (see, for example, PTL 1 or the like). Variation in signal (a time waveform) measured in EEG is herein collectively referred to as “brain waves.” With the technique disclosed in PTL 1 or the like, an estimation model is created from measurement data obtained by simultaneous EEG and fMRI in a resting state (which will also be abbreviated as “EEG/fMRI simultaneous measurement data” below), and based on the created estimation model, neurofeedback is given using only EEG measurement data. EEG is more advantageous than other measurement techniques in terms of portability, mobility, a price, and possibility of prevalence. Therefore, by adopting the technique disclosed in PTL 1 or the like, cost can be reduced to thereby enhance viability of neurofeedback training. It has been proposed to estimate an activity of each brain network based on fMRI measurement data in a resting state and to estimate “disorder-likelihood” based on a brain function expressed in a plurality of brain networks (see NPL 1 or the like). Estimation of “disorder-likelihood” is expected to be applied to diagnosis of psychiatric disorders, identification of a subtype of an identical disorder, and selection of a therapy. CITATION LIST Patent Literature PTL 1: Japanese Patent Laying-Open No. 2019-093008 Non Patent Literature NPL 1: Andrew T Drysdale et al., “Resting-state connectivity biomarkers define neurophysiological subtypes of depression,” Nature Medicine, Volume 23, Number 1, pp. 28-38 (ISSN: 1546-170X), 2017. 1NPL 2: Takashi Yamada et al., “Resting-State Functional Connectivity-Based Biomarkers and Functional MRI-Based Neurofeedback for Psychiatric Disorders: A Challenge for Developing Theranostic Biomarkers,” International Journal of Neuropsychopharmacology (2017) 20 (10), pp. 769-781, 2017. 7. 17NPL 3: Yujiro Yoshihara et al., “Overlapping but asymmetrical relationships between schizophrenia and autism revealed by brain connectivity,” bioRxiv, <URL:https://doi.org/10.1101/403212>, 2018. 9. 7NPL 4: Naho Ichikawa et al., “Primary functional brain connections associated with melancholic major depressive disorder and modulation by antidepressants,” Scientific Reports (2020) 10:3542<URL:https://doi.org/10.1038/s41598-020-60527-z>, 2020NPL 5: Enikö Bartók et al., “Cognitive functions in prepsychotic patients,” Progress in Neuro-Psychopharmacology & Biological Psychiatry 29 (2005) 621-625 SUMMARY OF INVENTION Technical Problem Conventional neurofeedback training is directed to change in activity in a specific brain region or a specific brain network (change in time correlation of activities between a plurality of brain regions) (see, NPL 2). A technique that allows easier estimation of any disorder associated with a plurality of brain networks has been demanded. Solution to Problem An estimation system according to one embodiment of the present invention includes obtaining means configured to obtain brain wave measurement data and functional magnetic resonance imaging measurement data simultaneously measured from a subject. The brain wave measurement data includes time waveforms for a plurality of channels corresponding to respective ones of a plurality of sensors arranged in a head of the subject. The estimation system includes first calculation means configured to calculate first functional connectivity for each channel combination based on correlation between channels included in the brain wave measurement data, second calculation means configured to calculate second functional connectivity for each brain network based on correlation between regions of interest included in the functional magnetic resonance imaging measurement data, third calculation means configured to calculate a disorder-likelihood label by calculating a score representing disorder-likelihood to be estimated based on a plurality of second functional connectivities, and machine learning means configured to determine an estimation model for estimating the disorder-likelihood based on prescribed first functional connectivity by machine learning using the first functional connectivity for each channel combination and the disorder-likelihood label. The estimation system may further include estimation means configured to estimate disorder-likelihood of the su