WO2022004841A1 - 推定システム、推定方法、プログラム、推定モデル、脳活動トレーニング装置、脳活動トレーニング方法、および、脳活動トレーニングプログラム - Google Patents
推定システム、推定方法、プログラム、推定モデル、脳活動トレーニング装置、脳活動トレーニング方法、および、脳活動トレーニングプログラム Download PDFInfo
- Publication number
- WO2022004841A1 WO2022004841A1 PCT/JP2021/024927 JP2021024927W WO2022004841A1 WO 2022004841 A1 WO2022004841 A1 WO 2022004841A1 JP 2021024927 W JP2021024927 W JP 2021024927W WO 2022004841 A1 WO2022004841 A1 WO 2022004841A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- disease
- likeness
- measurement data
- functional
- subject
- Prior art date
Links
Images
Classifications
-
- 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]
- A61B5/375—Electroencephalography [EEG] using biofeedback
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
- A61B5/055—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging
-
- 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]
-
- 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]
- A61B5/372—Analysis of electroencephalograms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2562/00—Details of sensors; Constructional details of sensor housings or probes; Accessories for sensors
- A61B2562/02—Details of sensors specially adapted for in-vivo measurements
Definitions
- the present invention relates to a technique for estimating disease-likeness based on measurement data related to brain activity.
- fMRI functional magnetic resonance imaging
- EEG electromagnetic field measurement method
- electroencephalogram hereinafter, also abbreviated as "EEG”
- EEG electromagnetic field measurement method
- fMRI electroencephalogram
- signal changes (time waveforms) measured by EEG are collectively referred to as “electroencephalograms”.
- an estimation model is created using measurement data obtained by simultaneously performing EEG and fMRI at rest (hereinafter, also abbreviated as “EEG / fMRI simultaneous measurement data”).
- EEG is advantageous over other measurement methods in terms of portability, portability, price, popularity, and the like. Therefore, by adopting the method disclosed in Patent Document 1 or the like, the cost can be reduced and the feasibility of the neurofeedback training can be enhanced.
- An estimation system includes an acquisition means for acquiring measurement data of brain waves simultaneously measured from a subject and measurement data of functional magnetic resonance imaging.
- the electroencephalogram measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the estimation system is included in the measurement data of the functional magnetic resonance imaging method and the first calculation means for calculating the first functional coupling for each channel combination based on the correlation between the channels included in the measurement data of the brain wave.
- a second calculation means for calculating a second functional connection for each brain network based on the correlation between the regions of interest, and a score indicating the disease-likeness of the estimation target using a plurality of the second functional connections.
- the estimation system may further include an estimation means for estimating the disease-likeness of the subject by inputting the measurement data of the brain waves measured from the subject into the estimation model.
- the estimation system may further include a presentation means for calculating a second score according to the estimated disease-likeness of the subject and presenting information according to the calculated second score to the subject.
- the estimation model may be prepared separately for each disease. At this time, the subject may be applied with an estimation model corresponding to the disease appearing in the subject.
- Changes in the subject's symptoms may be evaluated based on a second score according to the presumed subject's disease-likeness.
- the third calculation means calculates the score indicating the disease-likeness based on the sum of the weighting parameters corresponding to the plurality of second functional bonds associated with the disease-likeness to be estimated. May be good.
- the third calculation means may calculate the disease-likeness label by normalizing the score indicating the disease-likeness and then performing the threshold value processing.
- the estimation model includes information for selecting the first functional bond to be used for estimation among the first functional bonds for each channel combination, and the weighting parameter associated with the selected first functional bond. May be included.
- the first calculation means may calculate the first functional coupling from the correlation value between the time waveforms in the section included in the window commonly set for the time waveforms of the brain waves of the two target channels. good.
- the first calculation means may calculate the first functional coupling for each frequency band included in the electroencephalogram measurement data and / or for each window size of the window to be set.
- the estimation system may further include a condition setting means for predetermining the frequency band and / or the window size included in the electroencephalogram measurement data input to the estimation model according to the subject.
- the second calculation means is to calculate the second functional coupling from the correlation value between the time waveforms in the interval included in the window commonly set for the time waveform indicating the activity amount of the two target regions of interest. You may do it.
- the estimation method includes a step of acquiring measurement data of brain waves simultaneously measured from a subject and measurement data of functional magnetic resonance imaging.
- the electroencephalogram measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the estimation method is a step of calculating the first functional coupling for each channel combination based on the correlation between the channels included in the measurement data of the brain wave, and between the regions of interest included in the measurement data of the functional magnetic resonance imaging method.
- a program causes a computer to perform a step of acquiring measurement data of brain waves simultaneously measured from a subject and measurement data of functional magnetic resonance imaging.
- the electroencephalogram measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the program tells the computer the steps to calculate the first functional coupling for each channel combination based on the correlation between the channels contained in the EEG measurement data, and the interest contained in the functional magnetic resonance imaging measurement data.
- By calculating the second functional connection for each brain network based on the correlation between regions, and by calculating the score indicating the disease-likeness of the estimation target using multiple second functional connections.
- To estimate disease-likeness using a predetermined first functional binding by machine learning using the step of calculating the disease-like label and the first functional binding and the disease-likeness label for each channel combination. Perform the steps to determine the estimation model.
- a trained estimation model for estimating the disease-likeness of a subject is provided using the measurement data of the electroencephalogram measured from the subject.
- the process of constructing the estimation model includes the step of acquiring the measurement data of the electroencephalogram measured simultaneously from the subject and the measurement data of the functional magnetic resonance imaging method.
- the electroencephalogram measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the process of constructing the estimation model is included in the step of calculating the first functional coupling for each channel combination based on the correlation between the channels included in the EEG measurement data, and in the measurement data of the functional magnetic resonance imaging method.
- a brain activity training device for performing neurofeedback training.
- the brain activity training device is a storage device that stores an estimation model for estimating the subject's disease-likeness, which was generated before the execution of the neurofeedback training, and is for measuring the measurement data of the subject's brain waves in the neurofeedback training.
- the electroencephalogram measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the brain activity training device calculates the disease-likeness of the subject using an estimation model based on the measurement data from the electroencephalograph in the presentation device and the neurofeedback training, and outputs a signal for display corresponding to the disease-likeness.
- the estimation model is based on the process of acquiring the electroencephalogram measurement data measured simultaneously from the subject and the measurement data of the functional magnetic resonance imaging, and the correlation between the channels included in the electroencephalogram measurement data.
- the process of calculating the disease-likeness label by calculating the score indicating the disease-likeness of the estimation target using a plurality of the functional connections of 2 and the first functional connection and the disease-likeness label for each channel combination are used. It is generated by the process of determining the estimation model by estimating the disease-likeness using the predetermined first functional coupling by the machine learning.
- the electroencephalogram measurement data measured at the same time includes a time waveform for each channel corresponding to each channel of the electroencephalogram measurement data measured in the neurofeedback training.
- a brain activity training method for performing neurofeedback training includes a step of acquiring an estimation model for estimating the disease-likeness of the subject generated before the execution of the neurofeedback training, and a step of measuring the measurement data of the subject's brain wave in the neurofeedback training.
- the electroencephalogram measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the brain activity training method is a step in neurofeedback training in which a subject's disease-likeness is calculated using an estimation model based on brain wave measurement data, and a signal for display corresponding to the disease-likeness is output to a presentation device. including.
- the step of acquiring the estimation model is based on the process of acquiring the measurement data of the EEG measured simultaneously from the subject and the measurement data of the functional magnetic resonance imaging, and the correlation between the channels included in the measurement data of the brain wave.
- the second functional coupling is calculated for each brain network.
- a step of calculating the disease-likeness label by calculating a score indicating the disease-likeness of the estimation target using a step and a plurality of second functional connections, and a first functional connection and disease-likeness for each channel combination.
- the electroencephalogram measurement data measured at the same time includes a time waveform for each channel corresponding to each channel of the electroencephalogram measurement data measured in the neurofeedback training.
- a brain activity training program for performing neurofeedback training.
- the brain activity training program stores in a computer an estimation model for estimating the subject's disease-likeness, which was generated before the execution of the neurofeedback training, and acquires the measurement data of the subject's brain waves in the neurofeedback training.
- the electroencephalogram measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the brain activity training program calculates the disease-likeness of the subject using an estimation model based on the measured data of the brain waves in the neurofeedback training, and outputs a signal for display corresponding to the disease-likeness to the presentation device. To execute the steps to be performed.
- the estimation model is based on the process of acquiring the electroencephalogram measurement data measured simultaneously from the subject and the measurement data of the functional magnetic resonance imaging, and the correlation between the channels included in the electroencephalogram measurement data.
- the process of calculating the disease-likeness label by calculating the score indicating the disease-likeness of the estimation target using a plurality of the functional connections of 2 and the first functional connection and the disease-likeness label for each channel combination are used. It is generated by the process of determining the estimation model by estimating the disease-likeness using the predetermined first functional coupling by the machine learning.
- the electroencephalogram measurement data measured at the same time includes a time waveform for each channel corresponding to each channel of the electroencephalogram measurement data measured in the neurofeedback training.
- any disease associated with a plurality of brain networks can be estimated more easily.
- SCZ schizophrenia
- MDD depression
- FIG. 1 and 2 are schematic views showing an outline of an estimation method according to the present embodiment.
- FIG. 1 shows an outline of a process for determining an estimation model (learning phase)
- FIG. 2 shows an outline of a process for estimating disease-likeness (estimation phase) using the determined estimation model.
- disease is a term that includes not only pathological symptoms that occur in humans but also any mental or physical symptoms that are different from the conditions that appear in standard humans. The symptoms that appear in this case are also called “disease-like symptoms.” “Disease-likeness” means that the subject may have symptoms corresponding to the subject's “disease” (probability), and the subject may have symptoms corresponding to the subject's “disease”. It is a term that includes (probability).
- the “estimation model” is not limited to the estimation of these possibilities, and the healthy person has a brain activity having a gap of a predetermined degree or more from the standard "healthy brain activity state". In some cases, the possibility of the state (degree of gap) may be estimated. That is, an “estimation model” may be used to estimate the relative state of brain activity.
- the term “functional connection” is a term that includes an index indicating the degree of functional connection between regions in the brain.
- the “functional coupling” can be calculated by any method using the data measured by any measurement method. Unless the specific measurement data and the specific calculation method are specified in the present specification, the calculation method of "functional coupling” is not limited.
- EEG and fMRI are simultaneously performed on the same subject at rest, and EEG / fMRI simultaneous measurement data is acquired ((1)). EEG / fMRI simultaneous measurement).
- EEG measurement data the data obtained by EEG
- fMRI measurement data the data obtained by fMRI
- the EEG / fMRI simultaneous measurement data includes the electroencephalogram measurement data (EEG measurement data) simultaneously measured from the subject and the measurement data (fMRI measurement data) of the functional magnetic resonance imaging method.
- Each sensor is typically composed of a pair of electrodes.
- Each sensor is also referred to as a channel, and the EEG measurement data corresponds to a multi-channel brain wave. That is, the EEG measurement data includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors placed on the head of the subject.
- the functional coupling in each frequency band is calculated ((2) the functional coupling (FC) is calculated).
- FC Functional coupling
- FC Functional coupling
- EEG which is a measured value of the voltage generated by the electric activity of the brain
- MEG magnetoencephalography
- the fMRI measurement data is used to estimate the subject's disease-likeness from a specific brain network.
- a brain network is also called a resting state network (RSN), and a signal source belonging to a single brain region or a signal source belonging to multiple spatially separated brain regions cooperates. Is a general term for characteristic brain activity patterns. Brain networks are primarily defined using resting fMRI.
- the resting network includes (1) Control Network (CON), (2) Dorsal Attention Network (DAN), and (3) Default Mode Network (DMN). ), (4) Cerebral marginal system (LIM), (5) Somatomotor Network (SMN), (6) Ventral Attention Network (VAN), and (7) Visual network (7) Seven types of Visual Network (VIS) are known.
- CON Control Network
- DAN Dorsal Attention Network
- DNN Default Mode Network
- LIM Cerebral marginal system
- Somatomotor Network Somatomotor Network
- VAN Ventral Attention Network
- VIS Visual network (7) Seven types of Visual Network (VIS) are known.
- the (1) control network may be referred to as the frontal parietal network, and the (6) ventral attention network (VAN) may be referred to as the salency network. There is also.
- the above-mentioned resting network may be divided into several subnetworks. More specifically, (1) the control network (CON) is divided into three subnetworks, (3) the default mode network (DMN) is divided into four subnetworks, and the other networks are all divided. Divided into two subnetworks.
- the disease-likeness of a subject can be estimated based on one or more specific brain networks known in advance for each disease. Therefore, in the estimation method according to the present embodiment, the disease-likeness of the subject is estimated based on one or more specific brain networks ((3) the disease-likeness is estimated from a plurality of brain networks).
- the estimation result is also referred to as "disease-likeness label" (label).
- the disease-likeness label takes one of a plurality of values (levels).
- the input of EEG measurement data determines an estimation model for estimating the subject's disease-likeness ((((). 4) Determination of estimation model).
- the estimation model corresponds to a kind of trained model.
- the EEG measurement data measured by the EEG from the subject is input to the determined estimation model 10, and the estimation result of the subject's disease-likeness is output.
- the estimation model 10 also includes a function of selecting information suitable for estimating disease-likeness from the EEG measurement data measured by the EEG from the subject.
- the state of the subject's disease-likeness can be sequentially estimated, so that, for example, neurofeedback can be realized at low cost.
- the estimation model since the estimation model has target specificity, the estimation model is prepared for each disease. Then, an estimation model corresponding to the disease appearing in the subject will be applied.
- FIG. 3 is a schematic diagram showing a hardware configuration example of the disease-likeness estimation system 1 according to the present embodiment.
- the estimation system 1 includes a processing device 100, an EEG device 200, and an fMRI device 300.
- the processing device 100 acquires electroencephalogram measurement data (EEG measurement data) and functional magnetic resonance imaging measurement data (fMRI measurement data) simultaneously measured from the subject. More specifically, the processing device 100 receives the EEG measurement data measured by the EEG device 200 and the fMRI measurement data measured by the fMRI device 300, and determines an estimation model for estimating the disease-likeness.
- EEG measurement data electroencephalogram measurement data
- fMRI measurement data functional magnetic resonance imaging measurement data
- the EEG device 200 detects a signal (electrical signal) indicating an electroencephalogram generated by a plurality of sensors 220 arranged on the head of the subject S.
- the EEG device 200 includes a multiplexer 202, a noise filter 204, an A / D (Analog to Digital) converter 206, a storage unit 208, and an interface 210.
- the multiplexer 202 sequentially selects one set of cables from the cables 222 connected to the plurality of sensors 220, and electrically connects them to the noise filter 204.
- the noise filter 204 is a filter for removing noise such as a high frequency cut filter, and removes a noise component contained in a signal (electrical signal) indicating an electroencephalogram generated between a set of cables corresponding to the selected channel.
- the A / D converter 206 samples the electric signal (analog signal) output from the noise filter 204 at predetermined intervals and outputs it as a digital signal.
- the storage unit 208 sequentially stores the time-series data (digital signal) output from the A / D converter 206 in association with information indicating the selected channel and timing (for example, time or counter value).
- the interface 210 outputs time-series data indicating an electroencephalogram stored in the storage unit 208 to the processing device 100 in response to an access from the processing device 100 or the like.
- the fMRI apparatus 300 applies a high-frequency electromagnetic field having a resonance frequency toward a region (hereinafter, also referred to as a “region of interest”) for which information on the brain activity of the subject S is desired to be acquired, thereby applying a specific nuclear nucleus (for example, for example). Brain activity is measured by detecting electromagnetic waves generated by resonance from (hydrogen nuclei).
- the fMRI apparatus 300 includes a magnetic field application mechanism 310, a reception coil 302, a drive unit 320, and a data processing unit 350.
- the magnetic field application mechanism 310 applies a controlled magnetic field (static magnetic field and gradient magnetic field) to the region of interest of the subject S, and irradiates an RF (Radio Frequency) pulse. More specifically, the magnetic field application mechanism 310 includes a static magnetic field generation coil 312, a gradient magnetic field generation coil 314, an RF irradiation unit 316, and a sleeper 318 on which the subject S is placed in the bore.
- a controlled magnetic field static magnetic field generation coil 312
- a gradient magnetic field generation coil 314 an RF irradiation unit 316
- a sleeper 318 on which the subject S is placed in the bore.
- the drive unit 320 is connected to the magnetic field application mechanism 310 and controls the transmission and reception of the magnetic field applied to the subject S and the RF pulse wave. More specifically, the drive unit 320 includes a static magnetic field power supply 322, a gradient magnetic field power supply 324, a signal transmission unit 326, a signal reception unit 328, and a bed drive unit 330.
- the central axis of the cylindrical bore on which the subject S is placed is defined as the Z axis
- the horizontal and vertical directions orthogonal to the Z axis are defined as the X axis and the Y axis, respectively.
- the static magnetic field generation coil 312 generates a static magnetic field in the Z-axis direction in the bore from a spiral coil wound around the Z-axis.
- the gradient magnetic field generating coil 314 includes an X coil, a Y coil, and a Z coil (not shown) that generate gradient magnetic fields in the X-axis direction, the Y-axis direction, and the Z-axis direction in the bore, respectively.
- the RF irradiation unit 316 irradiates the region of interest of the subject S with an RF pulse based on the high frequency signal transmitted from the signal transmission unit 326 according to the control sequence.
- FIG 3 illustrates a configuration in which the RF irradiation unit 316 is built in the magnetic field application mechanism 310, but the RF irradiation unit 316 may be provided on the bed 318 side, and the RF irradiation unit 316 and the receiving coil 302 may be provided. It may be integrated.
- the receiving coil 302 receives the electromagnetic wave (NMR signal) emitted from the subject S and outputs an analog signal.
- the analog signal output from the receiving coil 302 is amplified and A / D converted in the signal receiving unit 328, and then output to the data processing unit 350.
- the receiving coil 302 is preferably arranged close to the subject S so that the NMR signal can be detected with high sensitivity.
- the data processing unit 350 sets a control sequence for the driving unit 320, and outputs a plurality of brain activity pattern images indicating the activity in the brain as information indicating the brain activity from the NMR signal received by the receiving coil 302. ..
- the data processing unit 350 includes a control unit 351, an input unit 352, a display unit 353, a storage unit 354, an image processing unit 356, a data collection unit 357, and an interface 358.
- the data processing unit 350 may be a dedicated computer or a general-purpose computer that realizes predetermined processing by executing a control program stored in a storage unit 354 or the like.
- the control unit 351 controls the operation of each functional unit such as generating a control sequence for driving the drive unit 320.
- the input unit 352 receives various operations and information input from an operator (not shown).
- the display unit 353 displays various images and various information regarding the region of interest of the subject S on the screen.
- the storage unit 354 stores control programs, parameters, image data (three-dimensional model image, etc.), other electronic data, and the like for executing processing related to fMRI.
- the image processing unit 356 generates a plurality of brain activity pattern images based on the detected NMR signal data.
- the interface 358 exchanges various signals with the drive unit 320.
- the data collection unit 357 collects data consisting of a group of NMR signals derived from the region of interest.
- FIG. 4 is a schematic diagram showing a hardware configuration example of the processing device 100 constituting the estimation system 1 that realizes the estimation method according to the present embodiment.
- the processing device 100 can typically employ a computer that follows a general-purpose architecture.
- the processing unit 100 has, as main components, a processor 102, a main storage unit 104, a control interface 106, a network interface 108, an input unit 110, a display unit 112, and a secondary storage unit. Includes 120 and.
- the processor 102 is composed of arithmetic processing circuits such as a CPU (Central Processing Unit) and a GPU (Graphical Processing Unit), and by executing the codes included in various programs stored in the secondary storage unit 120 in a specified order. , Realizes various functions described later.
- the main storage unit 104 is composed of a DRAM (Dynamic Random Access Memory) or the like, and holds the code of the program executed by the processor 102 and various work data necessary for executing the program.
- the processing device 100 has a communication function, and this communication function is mainly provided by the control interface 106 and the network interface 108.
- the control interface 106 exchanges data with the data processing unit 350 of the fMRI apparatus 300.
- the network interface 108 exchanges data with an external device (for example, a data server device on the cloud).
- the control interface 106 and the network interface 108 are composed of arbitrary communication components such as a wired LAN (Local Area Network), a wireless LAN, a USB (Universal Serial Bus), and Bluetooth (registered trademark).
- the input unit 110 is typically composed of a mouse, a keyboard, or the like, and accepts operations from the user.
- the display unit 112 is typically composed of a display or the like, and notifies the user of various information related to the execution state and operation of the process in the processing device 100.
- the secondary storage unit 120 is typically composed of a hard disk, SSD (Solid State Drive), or the like, and holds various programs executed by the processor 102, various data required for processing, setting values, and the like. More specifically, the secondary storage unit 120 stores the EEG measurement data 20, the fMRI measurement data 30, the estimation model determination program 121, the estimation program 122, and the estimation model parameter 124.
- SSD Solid State Drive
- FIG. 5 is a diagram for explaining a process of determining an estimation model in the estimation method according to the present embodiment.
- FIG. 6 is a diagram showing an example of data processing for determining an estimation model in the estimation method according to the present embodiment.
- the EEG measurement data 20 is a set of signal changes (time waveforms) indicating brain waves measured for each channel (sensor).
- the EEG measurement data 20 is converted into a time waveform 22 of power for each frequency band by preprocessing (corresponding to (1) preprocessing in FIG. 6).
- the time waveform 22 of the power means that the average value of the amplitude squared values of the corresponding frequency components included in the EEG measurement data 20 is sequentially calculated for each unit time.
- the EEG measurement data 20 (time waveform) is frequency-converted and the amplitude for each frequency is calculated. Then, the power can be calculated by selecting one or a plurality of frequencies included in the designated frequency band, squared the amplitude of the selected one or a plurality of frequencies, and calculating the average value.
- frequency analysis may be performed after downsampling to a predetermined sampling frequency.
- downsampling may be performed so that the sampling frequency is 1 / TR [Hz] so as to correspond to the irradiation cycle (TR: repetition time) of the RF pulse of fMRI.
- the time waveform 22 of the power of the number of channels N ⁇ the number of frequency bands M is generated by the preprocessing.
- the frequency band includes, for example, theta wave ( ⁇ wave: 4-8 Hz), alpha wave ( ⁇ wave: 8-12 Hz), low beta (low ⁇ wave: 12-20 Hz), and high beta (high ⁇ wave:). 20-30Hz) can be mentioned.
- the time correlation of the power time waveform 22 is calculated between different channels for the same frequency band (corresponding to (2) time correlation calculation in FIG. 6).
- time correlation means a correlation value between time waveforms and a time waveform of the correlation value in the section included in the window 26 commonly set for a plurality of time waveforms.
- a time correlation is calculated between the time waveforms 22 of the two powers.
- the time correlation means a time waveform having a correlation value focusing on the time width of the window 26 set for the time waveforms 22 of the two powers. That is, the functional coupling (FC) is calculated from the correlation value between the time waveforms in the section included in the window 26 commonly set for the time waveforms of the brain waves of the two target channels.
- the window 26 has a predetermined window size (time width), and the set position of the window 26 (time interval from the start time to the end time) is sequentially shifted by the step size, and each set position of the window 26 is sequentially shifted.
- the time waveform 24 of the EEG time correlation can be calculated.
- the calculated time waveform 24 of the EEG time correlation corresponds to FC.
- the functional coupling is calculated for each channel combination (each channel pair) based on the correlation between the channels included in the EEG measurement data 20.
- the time waveform 24 of the EEG time correlation is calculated for each frequency band. That is, a time waveform 24 having an EEG time correlation with a frequency band number M is generated.
- window size (time width) of the window 26 to be set may be different, and the time waveform 24 of the EEG time correlation may be calculated for each.
- the time waveform 24 of the EEG time correlation in which the channel pair, the frequency band, and the window size are different may be used.
- the time waveform 24 of the EEG time correlation is output as a vector of channel combination (channel pair) dimension ⁇ (number of time steps according to window 26) dimension for each frequency band and / or each window size. Will be done.
- the functional coupling may be calculated for each frequency band included in the EEG measurement data 20 and / or for each window size of the window 26 to be set.
- the fMRI measurement data 30 (that is, the brain activity pattern image) is a set of brain activity pattern images acquired for each RF pulse irradiation cycle.
- the fMRI measurement data 30 is a set of brain activity pattern images acquired for each RF pulse irradiation cycle.
- One or more regions corresponding to each of such brain networks correspond to "Region Of Interest” (hereinafter, also abbreviated as "ROI").
- the activity of each brain network is defined by the combination of two ROIs.
- the BOLD signal 32 for each ROI is calculated by preprocessing (corresponding to (1') preprocessing in FIG. 6).
- the BOLD signal means a temporal change in the amount of activity depending on the blood oxygen concentration for each ROI. More specifically, in the preprocessing for the fMRI measurement data 30, the BOLD signal is calculated based on the image feature amount corresponding to the ROI included in the brain activity pattern image.
- the time correlation of the BOLD signal 32 is calculated between the ROIs corresponding to each brain network (corresponding to the calculation of the (2') time correlation in FIG. 6).
- a time correlation is calculated between the two BOLD signals 32.
- the time correlation means a time waveform of a correlation value focusing on the time width of the window 26 set for the two BOLD signals 32. That is, the functional coupling (FC') is calculated from the correlation value between the time waveforms in the section included in the window 26 commonly set for the time waveform indicating the activity amount of the two ROIs of interest.
- the window 26 has a predetermined window size (time width), and the set position of the window 26 (time interval from the start time to the end time) is sequentially shifted by the step size, and each set position of the window 26 is sequentially shifted.
- time waveform 34 of the BOLD time correlation is a temporal change in the correlation value and corresponds to a functional coupling (FC').
- the time waveform 34 of the BOLD time correlation can be calculated for each ROI combination, that is, the target brain network.
- functional coupling FC' is calculated for each brain network based on the correlation between ROIs included in the fMRI measurement data 30.
- “Dynamic” in FIG. 5 means that a value is calculated for each window of interest, and “Static” means that a single value is calculated throughout the entire period. Therefore, “Static FC” in FIG. 5 means a correlation value (single functional combination) over the entire period.
- the disease to be estimated is used by using the time waveform 34 of the BOLD time correlation corresponding to a plurality of brain networks associated with the disease likeness of the estimation target by using such prior information.
- the WLS36 which is a score indicating the peculiarity, is calculated. More specifically, the WLS36 is calculated by multiplying and adding the corresponding weighting parameters to the time waveforms 34 (FC') of a plurality of target BOLD time correlations.
- Such a calculation method is known as a WLS (Weighted Linear Summation) method.
- WLS36 which is a score indicating the disease-likeness, is based on the sum of multiplying the plurality of functional couplings (time waveform 34 of the BOLD time correlation) associated with the disease-likeness to be estimated by the weighting parameters corresponding to each. Is calculated.
- the disease-likeness label 38 is calculated by calculating the score (WLS36) indicating the disease-likeness of the estimation target by using a plurality of time waveforms 34 (FC') of the BOLD time correlation.
- the score (WLS36) indicating the disease-likeness is normalized and then the threshold value is processed to calculate the disease-like label 38 (label). For example, when the binarization process is adopted as the threshold value process, the disease-likeness label 38 indicates “0” which means soundness or “1” which means disease.
- the disease-likeness label 38 is output as a vector of one dimension ⁇ (the number of settings in the window 26) dimension for the disease-likeness to be estimated.
- the disease-likeness label 38 is an explained variable.
- the estimation model defines the relationship between the time waveform 24 (FC) of the EEG time correlation, which is an explanatory variable, and the disease-likeness label 38 (label), which is an explained variable.
- FC time waveform 24
- label disease-likeness label 38
- the estimation method among the feature quantities included in the time waveform 24 of the EEG time correlation, which is a multidimensional vector, those suitable for estimating the disease-likeness label 38 are selected.
- the disease-likeness is estimated using the information of the selected features (time correlation calculated sequentially).
- the time waveform 24 (FC) of the EEG time correlation is used.
- An estimation model for estimating disease-likeness is determined.
- the dimension can be compressed and reduced, thereby reducing the calculation amount related to the estimation.
- the estimation process can be speeded up.
- SLR Sese Logistic Regression
- the time waveform 24 of the EEG time correlation is input to the machine learning algorithm SLR as an explanatory variable (corresponding to the input corresponding to (3) SLR in FIG. 6), and the disease-likeness label 38 is used as an explained variable.
- Is input to the machine learning algorithm SLR (corresponds to the input to (4') SLR in FIG. 6). Then, by machine learning, a feature amount suitable for estimating the disease-likeness label 38 is selected.
- FIG. 7 is a diagram for explaining an outline of an estimation model determined by the estimation method according to the present embodiment.
- the time waveform 24 of the EEG time correlation is a feature quantity group according to the number of EEG channels for each frequency band.
- the EEG time correlation time waveform 24 is also calculated for each window size.
- the weighting parameter W i is, the larger feature value F i suitable for estimation of disease likelihood label 38, a large value may be set.
- weighting parameter W i corresponding to the selected and selected feature amount F i of the feature quantity
- estimation model is determined, each channel combining the time waveform of EEG time correlation for the estimation of the time waveform of the EEG time correlation (each channel pair) 24 (FC) 24 (feature amount F i) It includes information for selecting, and weighting parameters W i associated with the time waveform 24 of the EEG time correlation that is selected.
- the subject's disease-likeness is sequentially estimated from only the EEG measurement data 20 using the feature amount and the corresponding weighting parameter determined by the procedure as described above.
- FIG. 8 is a flowchart showing a processing procedure of the estimation method according to the present embodiment. Some steps shown in FIG. 8 may be realized by executing a program in the processing apparatus 100.
- EEG and fMRI are simultaneously measured, and EEG measurement data 20 and fMRI measurement data 30 are acquired (step S100). That is, the processing device 100 acquires the electroencephalogram measurement data (EEG measurement data 20) and the fMRI measurement data (fMRI measurement data 30) simultaneously measured from the subject.
- the processing device 100 calculates the time waveform of the power for each frequency band by performing preprocessing on the acquired EEG measurement data 20 (step S102). Subsequently, the processing apparatus 100 calculates the time waveform of the EEG time correlation for each window size using the time waveform of the calculated power (step S104). That is, the processing apparatus 100 calculates the functional coupling (FC) for each channel combination based on the correlation between the channels included in the EEG measurement data 20.
- FC functional coupling
- the processing apparatus 100 performs preprocessing on the acquired fMRI measurement data 30 for each ROI constituting the brain network.
- the time waveform of the BOLD signal of (step S112) is calculated.
- the processing apparatus 100 uses the calculated time waveform of the BOLD signal to calculate the time waveform of the BOLD time correlation for each window size (step S114). That is, the processing apparatus 100 calculates the functional connection (FC') for each brain network based on the correlation between the ROIs included in the fMRI measurement data 30.
- the processing apparatus 100 selects the BOLD time correlation time waveform according to the disease-likeness of the estimation target from the calculated BOLD time correlation time waveforms, and multiplies and adds the corresponding weighting parameters. Then, WLS is calculated (step S116). Then, the processing apparatus 100 calculates the disease-likeness label indicating the disease-likeness by performing the calculated WLS in a normalized process and then binarizing it (step S118). That is, the processing apparatus 100 calculates the disease-likeness label by calculating the score (WLS) indicating the disease-likeness of the estimation target by using a plurality of functional couplings (FC').
- WLS the score
- FC' functional couplings
- the processing apparatus 100 determines the feature amount and the weighting parameter for estimating the disease-likeness label by machine learning using the time waveform of the EEG time correlation and the disease-likeness label (step S120). That is, the processing device 100 is an estimation model for estimating the disease-likeness using a predetermined functional connection (FC) by machine learning using the functional connection (FC) and the disease-likeness label for each channel combination. To decide.
- An estimation model can be determined by such a procedure.
- "(1) EEG / fMRI simultaneous measurement” shown in FIG. 1 and step S100 shown in FIG. 8 will be described.
- the subject S is placed on the bore of the fMRI apparatus 300 with the sensor attached to the head, and EEG and fMRI are executed in parallel.
- the processing device 100 stores the measurement data from the EEG device 200 and the fMRI device 300 in association with each other with respect to a common time. By associating the measurement data based on such a common time, the EEG measurement data 20 and the fMRI measurement data 30 having a common time axis can be acquired.
- the time waveform is frequency-converted as a preprocessing for the EEG measurement data 20 (time waveform).
- a fast Fourier transform or the like can be used.
- a Hilbert transform, a discrete Fourier transform, or the like may be used.
- the data in the frequency domain (relationship between frequency and amplitude) is calculated by frequency-converting the EEG measurement data 20.
- the power of the frequency band is calculated by calculating the average value of the amplitude squared values of the frequencies included in the frequency band for each target frequency band.
- step S104 of FIG. 8 any two channels are selected, the window is sequentially shifted along the time axis, and the correlation value is sequentially calculated between the time waveforms of the power in the window.
- FIG. 9 is a flowchart showing a more detailed processing procedure of steps S102 and S104 of FIG.
- the processing apparatus 100 selects one channel included in the acquired EEG measurement data 20 (step S1021), selects a time of interest for which power is calculated (step S1022), and selects the channel. Fast Fourier transform is performed on the time waveform included in the window with the time as the reference position (step S1023).
- the time waveform included in the window may be moved and averaged along the time axis, and then fast Fourier transform may be performed. By applying such a moving average, a high frequency noise component can be reduced.
- the processing apparatus 100 selects a frequency band for which power is to be calculated (step S1024), and calculates an average value of the amplitude squared values of the frequencies included in the selected frequency band (step S1025). Then, the processing device 100 stores the average value of the squared amplitude values in association with the selected time and the selected frequency band (step S1026).
- the processing device 100 determines whether or not the selection of all frequency bands has been completed (step S1027). If the selection of all frequency bands is not completed (NO in step S1027), the process of step S1024 or less is repeated.
- step S1027 the processing device 100 determines whether or not the selection of all time has been completed (step S1028). If all the time selections have not been completed (NO in step S1028), the processes of step S1022 and the like are repeated.
- step S1028 the processing device 100 determines whether or not the selection of all channels is completed (step S1029). If the selection of all channels is not completed (NO in step S1029), the process of step S1021 or less is repeated.
- step S1029 If the selection of all channels is completed (YES in step S1029), the calculation process of the power time waveform 22 for each frequency band is completed at this stage. Then, the calculation process of the time waveform 24 (FC) of the EEG time correlation continues.
- the processing apparatus 100 selects the window setting (window size and step size) for which the EEG time correlation is calculated (step S1041), and selects the frequency band for which the EEG time correlation is calculated (step S1042).
- window settings window size and step size
- a plurality of combinations may be prepared in advance, or only one type may be prepared.
- the processing apparatus 100 selects a target channel combination for calculating the EEG time correlation (step S1043).
- the processing apparatus 100 selects a target time for calculating the EEG time correlation (step S1044), and the power included in the window with the selected time as the reference position for the two channels corresponding to the selected channel combination.
- the time waveform of (step S1045) is extracted, and the correlation value of the time waveform of the extracted power is calculated (step S1046).
- the processing device 100 stores the calculated correlation value in association with the selected time, channel combination, frequency band, and window setting (step S1047).
- the processing device 100 determines whether or not all time selections have been completed (step S1048). If all the time selections have not been completed (NO in step S1048), the processes of step S1044 and the like are repeated.
- step S1048 the processing device 100 determines whether or not the selection of all the channel combinations is completed (step S1049). If the selection of all the channel combinations is not completed (NO in step S1049), the process of step S1043 or less is repeated.
- step S1049 the processing device 100 determines whether or not the selection of all frequency bands is completed (step S1050). If the selection of all frequency bands is not completed (NO in step S1050), the process of step S1042 or lower is repeated.
- step S1050 the processing device 100 determines whether or not the selection of all window settings is completed (step S1051). If the selection of all window settings is not completed (NO in step S1051), the process of step S1041 or less is repeated.
- step S1051 If the selection of all window settings is completed (YES in step S1051), the calculation process of the time waveform 24 (FC) of the EEG time correlation is completed at this stage.
- the BOLD signal 32 for each ROI is calculated from the brain activity pattern image.
- a process for compensating for the time delay that occurs in fMRI is executed.
- the BOLD signal 32 which is the neural state of the ROI of interest, is s (t) and the hemodynamic response function (HRF) is h (t).
- HRF hemodynamic response function
- HRF (t) depends on the irradiation cycle TR of the RF pulse of fMRI.
- the estimated value s ⁇ (t) of the brain state can be expressed by the following equation (2) using the Wiener filter d (t).
- H (x), Y (x), E (x), and D (x) are Fourier transforms of h (t), y (t), e (t), and d (t), the brain state.
- the estimated value s ⁇ (t) of can be expressed as the following equation (3).
- the estimated value s ⁇ (t) of the brain state shown in the above equation (3) corresponds to the BOLD signal. That is, the estimated value s (t) of the brain state is estimated by deconvolution of the observed y (t) with HRF. By deconvolution with HRF, the time delay (shift of measurement point) between the EEG measurement data 20 and the fMRI measurement data 30 is compensated.
- FIG. 10 is a diagram for explaining an outline of preprocessing for the EEG measurement data 20 and the fMRI measurement data 30.
- the window 26 is sequentially shifted and set by the step size with respect to the EEG measurement data 20, and the EEG time correlation is calculated for each window 26 to obtain the EEG time correlation time waveform 24. Can be calculated.
- the BOLD signal 32 is calculated after the time delay related to the irradiation of the RF pulse is compensated. That is, by deconvolution using HRF, the time axis of the EEG measurement data 20 and the time axis of the BOLD signal 32 can be substantially matched. Then, the disease-likeness label 38 is calculated using the BOLD signal 32.
- step S114 of FIG. 8 for each of the combinations of the two ROIs, the window is sequentially shifted along the time axis, and the correlation value is sequentially calculated between the time waveforms of the BOLD signals in the window.
- the combination of the two ROIs may be the same ROIs.
- the WLS is calculated by multiplying and adding the corresponding weighting parameters to the time waveforms of the plurality of BOLD time correlations according to the disease-likeness of the estimation target. Further, the calculated WLS is normalized and then binarized to calculate a disease-like label indicating the disease-likeness. More specifically, the WLS 36 uses the time waveform 34 (FC'(k)) of the k-th BOLD time correlation and the corresponding weighting parameter W FC (k) to be expressed in the following equation (4). Can be calculated as follows.
- WLS ⁇ FC'(k) ⁇ W FC (k) ⁇ ⁇ ⁇ (4)
- WLS is a score showing a larger numerical value as the degree of disease-likeness increases with 0 as a boundary.
- the WLS can be normalized to the probability p according to the following equation (5).
- Non-Patent Document 3 discloses a disease discriminator for schizophrenia (SCZ) using 16 functional bonds (FC').
- Non-Patent Document 4 discloses a disease discriminator for melancholic depression (MDD: melancholic) using 10 functional bonds (FC').
- FC' multiple functional bindings
- FC' weighted parameters associated with the selected functional bindings (FC'), depending on the disease likelihood of the presumed target.
- WLS can be determined by calculating the sum of the values obtained by multiplying each of W FC.
- FIG. 11 is a flowchart showing a more detailed processing procedure of steps S112 to S118 shown in FIG.
- the processing apparatus 100 selects the ROI for which the BOLD signal is to be calculated (step S1121), and determines the amount of activity from the image feature amount of the region corresponding to the ROI selected from each of the fMRI measurement data 30. Each is extracted (step S1122). By deconvolution of the extracted activity amount over time with HRF, the time waveform of the BOLD signal is calculated (step S1123) and stored in association with the selected ROI (step S1124).
- the processing device 100 determines whether or not all ROI selections have been completed (step S1125). If all ROI selections have not been completed (NO in step S1125), the process of step S1121 and below is repeated.
- step S1125 If all ROI selections have been completed (YES in step S1125), the BOLD signal calculation process for each ROI is completed at this stage. Then, the calculation process of the time waveform 34 of the BOLD time correlation continues.
- the processing apparatus 100 selects the ROI combination to be calculated for the BOLD time correlation (step S1141).
- the processing apparatus 100 selects a target time for calculating the BOLD time correlation (step S1142), and the BOLD included in the window with the selected time as the reference position for the two ROIs corresponding to the selected ROI combination.
- the time waveform of the signal 32 is extracted (step S1143), and the correlation value of the time waveform of the extracted BOLD signal is calculated (step S1144).
- the processing apparatus 100 stores the calculated correlation value in association with the selected time and ROI combination (step S1145).
- the processing device 100 determines whether or not all time selections have been completed (step S1146). If all the time selections have not been completed (NO in step S1146), the process of step S1142 or lower is repeated.
- step S1146 the processing apparatus 100 determines whether or not the selection of all ROI combinations has been completed. If the selection of all ROI combinations is not completed (NO in step S1147), the process of step S1141 or less is repeated.
- step S1147 If the selection of all ROI combinations is completed (YES in step S1147), the calculation process of the time waveform 34 of the BOLD time correlation is completed at this stage. Then, the calculation process of WLS continues.
- the processing device 100 selects the disease-likeness to be estimated (step S1161), and determines a plurality of brain networks (ROI combinations) associated with the selected disease-likeness (step S1162). Further, the processing device 100 determines a weighting parameter corresponding to each of the determined brain networks (step S1163). Then, the processing apparatus 100 calculates the total after multiplying the time waveform 34 of the BOLD time correlation of each of the determined brain networks by the corresponding weighting parameters (step S1164). The calculated sum is the WLS corresponding to the disease-likeness to be estimated.
- ROI combinations associated with the selected disease-likeness
- the processing device 100 determines whether or not all the disease-likeness selections have been completed (step S1165). If the selection of all disease-likeness is not completed (NO in step S1165), the process of step S1161 or less is repeated.
- step S1165 If the selection of all disease-likeness is completed (YES in step S1165), the WLS calculation process for each disease-likeness is completed at this stage. Then, the process of calculating the disease-likeness label continues.
- the processing apparatus normalizes the calculated WLS to calculate the probability p (step S1181), and performs the calculated probability p by the threshold value processing to output a sequence of 0 or 1 values (step S1182). ..
- the output column of 0 or 1 values serves as a disease-like label indicating the disease-likeness.
- the time waveform 24 of the EEG time correlation is machine-learned by machine learning the relationship between the EEG time correlation time waveform 24 (FC), which is an explanatory variable, and the disease-likeness label 38, which is the explained variable.
- FC EEG time correlation time waveform 24
- the disease-likeness label 38 which is the explained variable.
- a predetermined number for example, 30
- Any machine learning algorithm can be used as such a machine learning method, but as an example, a case where SLR is adopted will be described. Hereinafter, a specific processing procedure of SLR will be described.
- f (x; ⁇ ) hyperplane corresponding to 0 defines the boundary between the classes S 1 and class S 2.
- N pieces of input - data string containing the elements of the output ⁇ For (x 1 , y 1 ), (x 2 , y 2 ), ..., (X N , y N ) ⁇ , a probability function as shown in the following equation (8) can be defined.
- the purpose of machine learning is to introduce a probability function l ( ⁇ ) as shown in equation (10) and search for a weighted vector ⁇ that maximizes the value of the probability function l ( ⁇ ) defined by equation (10). do.
- a predetermined number for example, 30 is selected from the one with the highest weighting parameter size, and the feature amount corresponding to the selected weighting parameter is selected.
- the feature amount (specifying the channel pair, frequency band, window size to be used in the EEG time correlation) and the corresponding weighting parameter used for estimating the disease-likeness label 38 can be determined.
- FIG. 12 is a diagram for explaining the outline of the determined estimation model.
- the time waveform 24 of the EEG time correlation is input to the estimation model 10. More specifically, the amount of information of a predetermined window size (for example, 30 seconds) is input for each step size (for example, 30 seconds).
- the estimation model 10 includes a plurality of combinations of the feature amount information 11 and the weighting parameter 12. Of the input time waveforms 24 of the EEG time correlation, only the information corresponding to the feature amount information 11 included in the estimation model 10 (EEG time correlation selected as the feature amount) is used. Then, the weighting parameter 12 corresponding to the EEG time correlation used is multiplied, the sum of the respective results is calculated in the adder 13, and further binarized to 0 or 1 by the binarizer 14. The binarized result is output as disease-likeness.
- the channel pair, frequency band, and window size were all variable factors, but the frequency band and window size may be determined first as feature quantity conditions.
- EEG / fMRI simultaneous measurement data is acquired for a plurality of sessions, and discrimination performance (for example, AUC) is performed by a method such as cross-validation.
- discrimination performance for example, AUC
- the highest frequency band and window size index indicated by Area Under the Curve
- the amount of calculation required for the process of determining the estimation model can be reduced.
- the frequency band and / or the window size included in the EEG measurement data 20 to be input to the estimation model may be determined in advance according to the subject.
- the EEG measurement data measured from the subject is input to the estimation model to estimate the subject's disease-likeness.
- a typical application of such an estimation phase is neurofeedback training.
- FIG. 13 is a diagram for explaining an outline of neurofeedback training using the estimation method according to the present embodiment.
- the brain activity training device 2 for performing neurofeedback training includes an EEG device 200, a storage device 502, a display device 510, and a processing device 500.
- the storage device 502 stores the estimation model.
- the estimation model stored in the storage device 502 is generated before the execution of the neurofeedback training.
- the storage device 502 may be realized by using the storage included in the processing device 500, or may be realized by using the server device 400 shown in FIG.
- the display device 510 is an example of a presentation device, and provides visual or / or auditory information to the user.
- the EEG device 200 corresponds to an electroencephalograph and measures the measurement data of the electroencephalogram of the subject S in the neurofeedback training.
- the electroencephalogram measurement data measured by the EEG device 200 includes time waveforms for each of a plurality of channels corresponding to the plurality of sensors arranged on the head of the subject S, similarly to the EEG device 200 shown in FIG. .. That is, since substantially the same EEG device 200 is used at the time of generating the estimation model and at the time of executing the neurofeedback training, the EEG measurement data included in the EEG / fMRI simultaneous measurement data used when generating the estimation model. Will include the time waveform for each channel corresponding to each channel of the EEG measurement data measured in the neurofeedback training.
- the processing device 500 acquires EEG measurement data from the subject S by EEG, and estimates the disease-likeness using a predetermined estimation model. Disease likelihood is estimated on a cycle-by-cycle basis (typically, on a step-by-step size basis).
- the processing device 500 calculates a score according to the estimated disease-likeness, and provides a score display 520 according to the calculated score on the display device 510. In this way, the processing device 500 calculates the score according to the disease-likeness of the subject S using the estimation model based on the measurement data from the EEG device 200, and presents the information according to the calculated score to the subject. do. That is, the processing device 500 outputs a signal for display corresponding to the disease-likeness to the display device 510.
- the processing device 500 may be realized by executing a brain activity training program on a general-purpose computer.
- the score display 520 includes a reference circle 522 and a score circle 524 whose size changes according to the score.
- the size of the score circle 524 is sequentially updated according to the disease likelihood estimated based on the EEG measurement data measured from the subject S.
- Subject S is conscious of using his / her brain such as calculation, association, and meditation so that the size of the score circle 524 moves in the specified direction according to the instruction from the outside or himself / herself.
- the subject S is conscious of using the brain, it is possible to alleviate or treat the target disease.
- the disease-likeness can be estimated at any place. Taking advantage of these advantages, for example, it is possible to perform neurofeedback training at any place after simultaneous measurement of EEG and fMRI using a dedicated facility.
- FIG. 14 is a schematic diagram showing an implementation example of the estimation method according to the present embodiment.
- EEG and fMRI are simultaneously measured for each subject, and the processing apparatus 100 determines an estimation model 10 for each subject.
- the determined estimation model 10 is transmitted from the measurement station to the server device 400.
- the server device 400 holds subject data 402 including an estimation model for each subject.
- FIG. 15 is a schematic diagram showing an example of the functional configuration of the processing device 100 of the estimation system 1 according to the present embodiment. Each function shown in FIG. 15 is typically realized by the processor 102 of the processing apparatus 100 executing an estimation model determination program.
- the estimation model determination program 121 may be executed using one or a plurality of processors included in the processing device 100, or the plurality of processing devices may be linked to each other to execute the estimation model determination program 121. You may. In the latter case, a so-called cloud system, that is, a plurality of computers arranged on the network may be used. Furthermore, instead of the configuration (software implementation) realized by the processor executing the program, a hard-wired configuration such as FPGA (Field-Programmable Gate Array) or ASIC (Application Specific Integrated Circuit) is used for all or part of it. May be realized.
- FPGA Field-Programmable Gate Array
- ASIC Application Specific Integrated Circuit
- Each program according to the present embodiment may be implemented in a form that utilizes the function provided by the OS (Operating System), and even in such a case, it may be included in the technical scope of the present invention. ..
- the processing apparatus 100 includes preprocessing modules 150 and 160, time correlation calculation modules 152 and 162, WLS calculation module 164, binarization module 166, and model estimation module 168.
- the preprocessing module 150 converts the EEG measurement data 20 into a power time waveform 22.
- the time waveform 22 of the power may be calculated for each frequency band and / or for each window size.
- the time correlation calculation module 152 calculates the EEG time correlation time waveform 24 from the power time waveform 22 for each of the channel combinations (channel pairs).
- the preprocessing module 160 calculates the BOLD signal 32 for each ROI from the fMRI measurement data 30.
- the time correlation calculation module 162 calculates the time waveform 34 (FC') of the BOLD time correlation from the BOLD signal 32 for each ROI.
- the WLS calculation module 164 calculates WLS36, which is a score indicating the disease-likeness of the estimation target, using the time waveform 34 of the BOLD time correlation corresponding to the plurality of brain networks associated with the disease-likeness of the estimation target.
- the binarization module 166 normalizes the WLS36 and then calculates the disease-likeness label 38 (label), which is the result of binarizing the disease-likeness.
- the model estimation module 168 determines the feature amount and the weighting parameter for estimating the disease-likeness label 38 based on the time waveform 24 of the EEG time correlation and the disease-likeness label 38.
- the determined set of features and weighting parameters is output as the estimation model 10.
- FIG. 16 is a schematic diagram showing an example of the functional configuration of the processing device 500 of the estimation system 1 according to the present embodiment. Each function shown in FIG. 16 is realized by the processor of the processing device 500 executing an estimation program (similar to the estimation program 122 shown in FIG. 4).
- the estimation program may be executed using one or a plurality of processors included in the processing device 500, or a plurality of processing devices may be linked to each other to execute the estimation program. In the latter case, a so-called cloud system, that is, a plurality of computers arranged on the network may be used. Further, instead of the configuration (software implementation) realized by the processor executing the program, all or part of the configuration may be realized by using a hard-wired configuration such as FPGA or ASIC.
- Each program according to the present embodiment may be implemented in a form that utilizes the function provided by the OS, and even in such a case, it may be included in the technical scope of the present invention.
- the processing apparatus 500 includes a preprocessing module 550, a time correlation calculation module 552, a weighted sum calculation module 554, a binarization module 556, an estimation model acquisition module 558, and a display control module 560. And include.
- the preprocessing module 550 converts the EEG measurement data 20 into a power time waveform 22.
- the time waveform 22 of the power may be calculated for each frequency band and / or for each window size.
- the time correlation calculation module 552 calculates the EEG time correlation time waveform 24 from the power time waveform 22 for each of the channel combinations (channel pairs).
- the estimation model acquisition module 558 acquires the estimation model 10 corresponding to the subject from the server device 400 or the like.
- the estimation model 10 includes a set of features and weighting parameters for estimating the disease-likeness label 38.
- the weighted sum calculation module 554 selects one or a plurality of target feature quantities (EEG time correlation) from the time waveforms 24 of the EEG time correlation according to the estimation model 10 acquired by the estimation model acquisition module 558, and corresponds to them.
- the sum of the values obtained by multiplying the weighting parameters is calculated as WLS36.
- the binarization module 556 normalizes the WLS36 and then calculates the disease-likeness (0 or 1) which is the result of binarizing the disease-likeness.
- the display control module 560 calculates the score based on the disease-like value sequentially output from the binarization module 556, and calculates the score display 520 to be displayed on the display device 510. In this way, changes in the subject's symptoms will be evaluated based on the score according to the estimated subject's disease-likeness.
- Subclinical means a condition in which it is judged that there is a strong tendency to present at least some symptoms from the contents of the answers to the questions for evaluating the degree of symptoms for the disease of interest.
- EEG and fMRI were performed simultaneously at rest, and EEG / fMRI simultaneous measurement data was acquired.
- EEG / fMRI simultaneous measurement data was acquired for at least 8 sessions (5 minutes or less per session) for each subject.
- Two target diseases were assumed: schizophrenia (SCZ) (see Non-Patent Document 3) and depression (MDD) (see Non-Patent Document 4).
- FIG. 17 is a diagram showing an example of the evaluation result of the feature amount condition in the estimation method according to the present embodiment.
- FIG. 17 shows changes in the estimated performance when the window sizes for calculating the time correlation are different (8TR, 12TR, 16TR, 20TR, 24TR).
- TR means the irradiation cycle of the RF pulse.
- FIG. 17A shows an example of the evaluation result when schizophrenia (SCZ) is the target disease
- FIG. 17B shows the evaluation result when depression (MDD) is the target disease. An example is shown.
- the disease-likeness of schizophrenia is shown for the estimation model determined from the EEG measurement data for schizophrenia (SCZ) (hereinafter, also referred to as “schizophrenia estimation model”).
- the disease-likeness of depression is shown for the evaluation result of evaluating the estimation performance of the indicated score and the estimation model determined from the EEG measurement data for depression (MDD) (hereinafter, also referred to as "depression estimation model”).
- the evaluation result which evaluated the estimation performance of the score is shown. The following is an example of the results of cross-sectional evaluation of each model.
- FIG. 18 is a diagram showing an example of the evaluation result of the target specificity of the estimation model determined by the estimation method according to the present embodiment.
- FIG. 18A shows the estimation performance (mean AUC) when estimating the disease-likeness of schizophrenia (SCZ) and the disease-likeness of depression (MDD) using the schizophrenia estimation model.
- FIG. 18B shows the estimation performance when estimating the disease-likeness of depression (MDD) and the estimation performance when estimating the disease-likeness of schizophrenia (SCZ) using a depression estimation model. ..
- the schizophrenia estimation model shows specific estimation performance for estimating the disease-likeness of schizophrenia (SCZ).
- the depression estimation model shows specific estimation performance for estimating the disease-likeness of depression (MDD).
- the estimation model determined by the estimation method according to the present embodiment is subject-specific. It turns out that.
- J3 Neurofeedback training
- FIG. 19 is a diagram for explaining a method of neurofeedback training using an estimation model determined by an estimation method according to the present embodiment.
- a set 614 consisting of a plurality of blocks 612 is carried out over a plurality of days.
- Each of the blocks 612 contains a plurality of cycles 600.
- Each of the cycles 600 includes a series of processes consisting of an interval 602 (time T1), an induction period 604 (time T2), and a display period 606 (time T3).
- Interval 602 corresponds to a break period between the previous cycle 600 and the previous cycle 600.
- the induction period 604 corresponds to a period in which the subject is conscious of using the brain such as calculation, association, and meditation so that the subject can be evaluated as a larger score by himself / herself or according to an external instruction.
- the display period 606 corresponds to the period for displaying the score calculated from the subject in the induction period 604.
- the subject's disease-likeness is estimated using the EEG measurement data measured by the EEG from the subject.
- the subject's disease-likeness estimation may be repeated multiple times. Since the estimation result (0 or 1) of the disease-likeness is calculated a plurality of times, a score indicating the degree of the disease-likeness of the subject in the induction period 604 can be calculated by averaging these.
- a score display 520 according to the calculated score is provided to the subject.
- the score display 520 the smaller the score, the closer the score circle 524 is to the reference circle 522.
- the subject is given a reward such as money according to the score. Motivated by such rewards, subjects attempt to obtain higher scores.
- time T2 of the induction period 604 for example, about 50 to 70 seconds is set.
- Non-Patent Document 3 two target diseases are assumed: schizophrenia (SCZ) (see Non-Patent Document 3) and depression (MDD) (see Non-Patent Document 4).
- the EEG sampling frequency was set to 500 Hz, and the EEG measurement data was processed to remove artifacts (preliminarily extracted subject-specific independent components).
- the time T2 of the induction period 604 for schizophrenia (SCZ) was 70 seconds
- the time T2 of the induction period 604 for depression (MDD) was 85 seconds.
- FIG. 20 is a diagram showing an example of the results of neurofeedback training regarding schizophrenia (SCZ).
- FIG. 20A shows an experimental example of WLS, which is a score indicating the disease-likeness of the estimation target before and after training.
- FIG. 20B shows an experimental example of the ataxic personality disorder evaluation scale (SPQ) before and after training. SPQ is an example of a schizophrenia-like score.
- FIG. 20C shows an experimental example of the N-back task before and after training.
- FIGS. 20 (A) and 20 (B) indicate subjects.
- the WLS shown in FIG. 20 (A) and the SPQ shown in FIG. 20 (B) both mean that the smaller the value, the better the symptom. No significant result was shown for WLS shown in FIG. 20 (A), but improvement tendency was seen for SPQ shown in FIG. 20 (B) by training.
- the N-back task shown in FIG. 20 (C) is a test for evaluating the ability (cognitive function) to indicate whether or not the information presented N times before is memorized.
- the results of the N-back task are indicated by a score of "d prime".
- d prime means that the larger the value, the better the cognitive function.
- an improvement tendency can be seen by training.
- 4-back task (4-back test) a significant change was observed for the paired t-test.
- FIG. 21 is a diagram showing an example of the result of neurofeedback training regarding depression (MDD).
- FIG. 21 (A) shows an experimental example of WLS, which is a score indicating the disease-likeness of the estimation target before and after training.
- FIG. 21B shows experimental examples of the Beck Depression Inventory (BDI) and the self-assessed Depression Inventory (SDS) before and after training. BDI and SDS are examples of depression-like symptom scores.
- FIG. 21C shows an experimental example of the results of the N-back task before and after training.
- FIGS. 21 (A) and 21 (B) indicate subjects.
- the WLS shown in FIG. 21 (A) and the BDI and SDS shown in FIG. 21 (B) both mean that the smaller the value, the better the symptom.
- an improvement tendency is seen by training.
- the N-back task also shows an improvement tendency by training.
- FIG. 22 is a diagram showing an example of a procedure for evaluating the long-term effect of neurofeedback training using an estimation model determined by an estimation method according to the present embodiment.
- the training is carried out for 3 days, and the day before the training period (Pre training), the day after the training period (Post training), and the follow-up day 1 to 2 months after the training period (Pre training).
- Measurements were performed in each of the FUs.
- FIG. 23 is a diagram showing an example of the long-term effect of neurofeedback training on depression (MDD).
- FIG. 23 (A) shows an experimental example of WLS.
- FIG. 23B shows an experimental example of BDI.
- FIG. 23C shows an experimental example of RSS, which is a score indicating the frequency of ruminant thinking. It can be judged that the smaller the value of RSS is, the more preferable the state is.
- FIG. 23B shows the reduced state immediately after the training (Pre) and after 1 to 2 months (FU), and the effect of the training was long-term. It has been suggested to persist.
- FIG. 23B also shows the subscores used to calculate the BDI.
- the subscore shows the same tendency as BDI.
- FIG. 23C shows the reduced state was maintained immediately after the training (Pre) and after 1 to 2 months (FU), and the effect of the training was long-term. It has been suggested to persist.
- FIG. 23C also shows the subscore used to calculate the RRS. The subscore shows the same tendency as RRS.
- FIG. 24 is a diagram showing an example of the long-term effect of neurofeedback training on schizophrenia (SCZ).
- FIG. 24 shows the change in WLS of each subject in a line graph, and shows the change in the score obtained by averaging the WLS of all the subjects in a bar graph.
- CTRL indicates the result of the comparison target group.
- the comparison target group shows a set of subjects who have been trained by using the information of another person prepared in advance instead of the information from the subject as the information to be fed back. That is, the experimental example of the comparison target group shows the result of training on the assumption that the subject is based on his / her own brain activity even though his / her own brain activity is not referred to. The same applies to the following experimental examples.
- FIG. 25 is a diagram showing the effect of neurofeedback training on schizophrenia (SCZ) in comparison with the comparison target group.
- the vertical axis of the graph shown in FIG. 25 shows the change (Post-Pre) of the value before and after the neurofeedback.
- the comparison target group (CTRL) is distributed around the point where the values before and after training do not change (the value on the vertical axis is zero), whereas the group that has been properly trained is on the negative side. It can be seen that the distribution is centered on the points (that is, the SPQ value becomes smaller after training).
- FIG. 25 also shows the subscores used to calculate the SPQ. The subscore shows the same tendency as SPQ.
- the SPQ shown in FIG. 25 also had a significant difference from the comparison target group, indicating an improvement tendency due to training.
- FIG. 26 is another diagram showing the effect of neurofeedback training on schizophrenia (SCZ) in comparison with the comparison target group.
- the vertical axis of the graph shown in each of FIGS. 26 (A) to 26 (D) shows the change (Post-Pre) of the value before and after the neurofeedback.
- 26 (A) to 26 (D) each show an experimental example of the score of cognitive function.
- CTRL comparison target group
- 26 (C) and 26 (D) show an example of evaluation of cognitive function using CANTAB (Cambridge Neuropsychological Test Automated Battery) (see Non-Patent Document 5 and the like). More specifically, the sustainability attention task (RVP: rapid visual information processing) is evaluated. It has been reported that patients with schizophrenia have reduced function of persistent attention.
- CANTAB Cosmetic Neuropsychological Test Automated Battery
- the comparison target group (CTRL) is distributed around the point where the values before and after training do not change (the value on the vertical axis is zero), whereas the training is performed appropriately. It can be seen that the group is distributed around the points on the plus side (that is, the values of A'and p (Hit) are both large after training).
- J6 Specificity of the effect of neurofeedback training
- the neurofeedback training itself produces non-specific effects such as learning effects, and such non-specific effects can be achieved by neurofeedback training using an estimation model determined by the estimation method according to the present embodiment.
- An experimental example will be described to explain the occurrence of a specific effect beyond that.
- FIG. 27 is a diagram showing an experimental example for evaluating the specificity of the effect of neurofeedback training.
- FIG. 27 (A) shows an example of changes in RRS and subscore as a psychological index related to depression (MDD).
- FIG. 27 (B) shows an example of changes in SPQ and subscore as a psychological index for schizophrenia (SCZ).
- MDD means a group trained using an estimated model (depression estimated model) determined from EEG measurement data for depression (MDD).
- SCZ means a group trained using an estimation model (schizophrenia estimation model) determined from EEG measurement data for schizophrenia (SCZ).
- CRL means a group to be compared.
- both the SPQ (total score) and the subscore are specific for the group (SCZ) trained using the schizophrenia estimation model. It can be seen that changes are occurring.
- FIG. 28 is a diagram showing another experimental example for evaluating the specificity of the effect of neurofeedback training.
- FIG. 28 shows an example of changes in cognitive function.
- the tendency of improvement of cognitive function is shown regardless of the training using either the depression estimation model or the schizophrenia estimation model.
- FIG. 28 (B) a significant improvement tendency of cognitive function is shown in the training using the schizophrenia estimation model.
- the neurofeedback training using the estimation model determined by the estimation method according to the present embodiment shows an improvement tendency of cognitive function, and it is higher by using the schizophrenia estimation model. There is a tendency for improvement.
- the EEG measurement data can be used to more easily estimate the brain function represented by the plurality of brain networks and any disease associated with the plurality of brain networks.
- the estimation system according to the present embodiment, only the features of the EEG measurement data that are effective for estimating the disease-likeness are used in the estimation model, so that the dimension of the estimation model can be compressed and reduced, thereby causing the disease. It is possible to reduce the amount of calculation related to the estimation of the peculiarity and speed up the estimation of the peculiarity of the disease.
- the disease-likeness can be estimated for any disease associated with a plurality of brain networks, neurofeedback training can be applied to various diseases.
- the estimation model can be determined using the measurement data obtained by simultaneously performing EEG and fMRI at rest. Therefore, when performing simultaneous measurement of EEG and fMRI, the subject Since it is not necessary to give a task to the subject, the burden on the subject can be reduced in constructing the estimation model.
- the neurofeedback training provided by the estimation system according to the present embodiment gives an improvement tendency for some diseases, and the improvement tendency is maintained for a long period of time.
- estimation model used in the neurofeedback training provided by the estimation system according to the present embodiment shows the target specificity and is generated according to the disease.
- 1 estimation system 2 brain activity training device, 11 feature amount information, 12 weighting parameter, 13 adder, 14 binarizer, 20 EEG measurement data, 22 power time waveform, 24 EEG time correlation time waveform, 26 window , 30 fMRI measurement data, 32 BOLD signal, 34 BOLD time correlation time waveform, 38 disease-like label, 100,500 processing device, 102 processor, 104 main storage unit, 106 control interface, 108 network interface, 110,352 input unit , 112,353 Display unit, 120 secondary storage unit, 121 estimation model determination program, 122 estimation program, 124 estimation model parameters, 150,160,550 preprocessing module, 152,162,552 time correlation calculation module, 164WLS calculation Module, 166,556 binarization module, 168 model estimation module, 200 EEG device, 202 multiplexer, 204 noise filter, 206 A / D converter, 208,354 storage, 210,358 interface, 220 sensor, 222 cable, 300 fMRI device, 302 receiving coil, 310 magnetic field application mechanism, 3
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Veterinary Medicine (AREA)
- Molecular Biology (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Psychology (AREA)
- Psychiatry (AREA)
- High Energy & Nuclear Physics (AREA)
- Radiology & Medical Imaging (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
Description
まず、本実施の形態に従う推定方法の概要について説明する。図1および図2は、本実施の形態に従う推定方法の概要を示す模式図である。図1には、推定モデルを決定する処理(学習フェーズ)の概要を示し、図2には、決定された推定モデルを用いて疾患らしさを推定(推定フェーズ)する処理の概要を示す。
次に、本実施の形態に従う推定方法を実現するための推定システムのハードウェア構成例について説明する。
次に、本実施の形態に従う推定方法における推定モデルの決定処理について説明する。
EEG計測データ20は、チャネル(センサ)毎に計測される、脳波を示す信号変化(時間波形)の集合である。EEG計測データ20は、前処理(図6の(1)前処理に対応)によって、周波数帯域毎のパワーの時間波形22に変換される。パワーの時間波形22は、EEG計測データ20に含まれる対応する周波数成分の振幅二乗値の平均値を単位時間毎に順次算出したものを意味する。
fMRI計測データ30(すなわち、脳活動パターン画像)は、RFパルスの照射周期毎に取得される脳活動パターン画像の集合である。既知の脳内ネットワーク(安静時ネットワーク)の各々について、脳内のいずれの領域の脳活動に対応しているのかは既知である。このような脳内ネットワークの各々に対応する1または複数の領域が「関心領域」(Region Of Interest:以下、「ROI」とも略称する。)に相当する。
推定モデルは、説明変数であるEEG時間相関の時間波形24(FC)と被説明変数である疾患らしさラベル38(label)との関係を規定するものである。本実施の形態に従う推定方法においては、多次元ベクトルであるEEG時間相関の時間波形24に含まれる特徴量のうち、疾患らしさラベル38の推定に適したものが選択される。推定フェーズにおいては、選択された特徴量の情報(順次算出される時間相関)を利用して、疾患らしさが推定される。すなわち、チャネル組合せ毎(チャネルペア毎)のEEG時間相関の時間波形24(FC)と疾患らしさラベル38とを用いた機械学習により、所定のEEG時間相関の時間波形24(FC)を用いて、疾患らしさを推定するための推定モデルが決定される。
図8は、本実施の形態に従う推定方法の処理手順を示すフローチャートである。図8に示す一部のステップは、処理装置100においてプログラムが実行されることで実現されてもよい。
[D.EEG/fMRI同時計測]
次に、図1に示す「(1)EEG/fMRI同時計測」、および、図8に示すステップS100について説明する。図3に示す推定システム1を利用して、被験者Sは、頭部にセンサを装着した状態で、fMRI装置300のボアに載置されて、EEGおよびfMRIが並列的に実行される。
次に、図1に示す「(2)機能的結合(FC)を算出」、および、図8に示すステップS102~S104について詳述する。
図9は、図8のステップS102およびS104のより詳細な処理手順を示すフローチャートである。図9を参照して、処理装置100は、取得されたEEG計測データ20に含まれる1つのチャネルを選択し(ステップS1021)、パワーを算出する対象の時刻を選択し(ステップS1022)、選択した時刻を基準位置とするウィンドウに含まれる時間波形を高速フーリエ変換する(ステップS1023)。
次に、図1に示す「(3)複数の脳内ネットワークから疾患らしさを推定」、および、図8に示すステップS112~S118について詳述する。
脳状態の推定値s^(t)は、ウィーナーフィルタd(t)を用いて、以下の(2)式のように示すことができる。
WLSは、0を境界として、疾患らしさの度合いが大きくなるほど大きな数値を示すスコアである。WLSは、以下に示す(5)式に従って、確率pに正規化できる。
確率p(0≦p≦1)は、0.5を境界として、疾患らしさの度合いが大きくなるほど1に近付くことになる。
図11は、図8の示すステップS112~S118のより詳細な処理手順を示すフローチャートである。図11を参照して、処理装置100は、BOLD信号の算出対象となるROIを選択し(ステップS1121)、fMRI計測データ30のそれぞれから選択したROIに対応する領域の画像特徴量から活動量をそれぞれ抽出する(ステップS1122)。抽出した活動量の時間的変化をHRFで逆畳み込みすることで、BOLD信号の時間波形を算出し(ステップS1123)、選択中のROIに関連付けて格納する(ステップS1124)。
次に、図1に示す「(4)推定モデルの決定」、および、図8に示すステップS120について詳述する。
次に、上述したような学習フェーズの処理によって決定される推定モデルを用いた推定フェーズの処理例について説明する。
次に、本実施の形態に従う推定方法を実現する推定システム1に含まれる装置の機能構成の一例について説明する。
図15は、本実施の形態に従う推定システム1の処理装置100の機能構成の一例を示す模式図である。図15に示す各機能は、典型的には、処理装置100のプロセッサ102が推定モデル決定プログラムを実行することで実現される。
次に、図13および図14に示す処理装置500に実現される機能構成の一例について説明する。処理装置500のハードウェア構成は、上述の図4に示す処理装置100のハードウェア構成と同様であるので詳細な説明は繰り返さない。
次に、本実施の形態に従う推定方法を実際の被験者に適用して得られた結果のいくつかを説明する。
まず、推定モデルを決定するための特徴量条件の推定精度について評価した結果例について説明する。
次に、推定モデルの対象特異性について説明する。
次に、本実施の形態に従う推定方法により決定された推定モデルを用いたニューロフィードバックトレーニングの一例について説明する。
次に、ニューロフィードトレーニングの終了から1~2ヶ月後の追跡調査(フォローアップ:FU)における結果を含む長期効果の評価例について説明する。
次に、比較対象群をベンチマークとしたニューロフィードバックトレーニングの効果の一例について説明する。
ニューロフィードバックトレーニング自体は、学習効果などの非特異的な効果を生じるものであるが、本実施の形態に従う推定方法により決定された推定モデルを用いたニューロフィードバックトレーニングにより、そのような非特異的を超えた特異的な効果が生じることについて実験例を示して説明する。
本実施の形態に従う推定システムによれば、EEG計測データを用いて、複数の脳内ネットワークで表現される脳機能および複数の脳内ネットワークに関連付けられる任意の疾患をより簡便に推定できる。
Claims (18)
- 被験者から同時に計測された脳波の計測データおよび機能的磁気共鳴画像法の計測データを取得する取得手段を備え、前記脳波の計測データは前記被験者の頭部に配置される複数のセンサにそれぞれ対応する複数のチャネル毎の時間波形を含み、
前記脳波の計測データに含まれるチャネル間の相関に基づいて、チャネル組合せ毎に第1の機能的結合を算出する第1の算出手段と、
前記機能的磁気共鳴画像法の計測データに含まれる関心領域間の相関に基づいて、脳内ネットワーク毎に第2の機能的結合を算出する第2の算出手段と、
前記第2の機能的結合の複数を用いて推定対象の疾患らしさを示すスコアを算出することで、疾患らしさラベルを算出する第3の算出手段と、
前記チャネル組合せ毎の第1の機能的結合と前記疾患らしさラベルとを用いた機械学習により、所定の第1の機能的結合を用いて、前記疾患らしさを推定するための推定モデルを決定する機械学習手段とを備える、推定システム。 - 前記被験者から計測された脳波の計測データを前記推定モデルに入力して、前記被験者の疾患らしさを推定する推定手段をさらに備える、請求項1に記載の推定システム。
- 前記推定される被験者の疾患らしさに応じた第2のスコアを算出するとともに、前記算出した第2のスコアに応じた情報を前記被験者に提示する提示手段をさらに備える、請求項2に記載の推定システム。
- 前記推定モデルは、疾患の別に用意され、
前記被験者には、前記被験者に現れる疾患に対応する推定モデルが適用される、請求項3に記載の推定システム。 - 前記推定される被験者の疾患らしさに応じた第2のスコアに基づいて、前記被験者の症状の変化が評価される、請求項1~4のいずれか1項に記載の推定システム。
- 前記第3の算出手段は、前記推定対象の疾患らしさに対応付けられた複数の前記第2の機能的結合にそれぞれ対応する重み付けパラメータを乗じた総和に基づいて、前記疾患らしさを示すスコアを算出する、請求項1~5のいずれか1項に記載の推定システム。
- 前記第3の算出手段は、前記疾患らしさを示すスコアを正規化処理した上でしきい値処理することで、前記疾患らしさラベルを算出する、請求項6に記載の推定システム。
- 前記推定モデルは、チャネル組合せ毎の第1の機能的結合のうち推定に使用する第1の機能的結合を選択するための情報と、選択された第1の機能的結合に対応付けられる重み付けパラメータとを含む、請求項1~7のいずれか1項に記載の推定システム。
- 前記第1の算出手段は、対象の2つのチャネルの脳波の時間波形に対して共通に設定したウィンドウに含まれる区間における時間波形間の相関値から前記第1の機能的結合を算出する、請求項1~8のいずれか1項に記載の推定システム。
- 前記第1の算出手段は、前記脳波の計測データに含まれる周波数帯域毎、および/または、設定するウィンドウのウィンドウサイズ毎、に前記第1の機能的結合を算出する、請求項1~9のいずれか1項に記載の推定システム。
- 前記推定モデルに入力する前記脳波の計測データに含まれる周波数帯域および/またはウィンドウサイズ、を前記被験者に応じて事前に決定する条件設定手段をさらに備える、請求項10に記載の推定システム。
- 前記第2の算出手段は、対象の2つの関心領域の活動量を示す時間波形に対して共通に設定したウィンドウに含まれる区間における時間波形間の相関値から前記第2の機能的結合を算出する、請求項1~11のいずれか1項に記載の推定システム。
- 被験者から同時に計測された脳波の計測データおよび機能的磁気共鳴画像法の計測データを取得するステップを備え、前記脳波の計測データは前記被験者の頭部に配置される複数のセンサにそれぞれ対応する複数のチャネル毎の時間波形を含み、
前記脳波の計測データに含まれるチャネル間の相関に基づいて、チャネル組合せ毎に第1の機能的結合を算出するステップと、
前記機能的磁気共鳴画像法の計測データに含まれる関心領域間の相関に基づいて、脳内ネットワーク毎に第2の機能的結合を算出するステップと、
前記第2の機能的結合の複数を用いて推定対象の疾患らしさを示すスコアを算出することで、疾患らしさラベルを算出するステップと、
前記チャネル組合せ毎の第1の機能的結合と前記疾患らしさラベルとを用いた機械学習により、所定の第1の機能的結合を用いて、前記疾患らしさを推定するための推定モデルを決定するステップとを備える、推定方法。 - プログラムであって、コンピュータに、
被験者から同時に計測された脳波の計測データおよび機能的磁気共鳴画像法の計測データを取得するステップを実行させ、前記脳波の計測データは前記被験者の頭部に配置される複数のセンサにそれぞれ対応する複数のチャネル毎の時間波形を含み、
前記脳波の計測データに含まれるチャネル間の相関に基づいて、チャネル組合せ毎に第1の機能的結合を算出するステップと、
前記機能的磁気共鳴画像法の計測データに含まれる関心領域間の相関に基づいて、脳内ネットワーク毎に第2の機能的結合を算出するステップと、
前記第2の機能的結合の複数を用いて推定対象の疾患らしさを示すスコアを算出することで、疾患らしさラベルを算出するステップと、
前記チャネル組合せ毎の第1の機能的結合と前記疾患らしさラベルとを用いた機械学習により、所定の第1の機能的結合を用いて、前記疾患らしさを推定するための推定モデルを決定するステップとを実行させる、プログラム。 - 被験者から計測された脳波の計測データを用いて、前記被験者の疾患らしさを推定するための学習済の推定モデルであって、前記推定モデルを構築する処理は、
前記被験者から同時に計測された脳波の計測データおよび機能的磁気共鳴画像法の計測データを取得するステップを備え、前記脳波の計測データは前記被験者の頭部に配置される複数のセンサにそれぞれ対応する複数のチャネル毎の時間波形を含み、
前記脳波の計測データに含まれるチャネル間の相関に基づいて、チャネル組合せ毎に第1の機能的結合を算出するステップと、
前記機能的磁気共鳴画像法の計測データに含まれる関心領域間の相関に基づいて、脳内ネットワーク毎に第2の機能的結合を算出するステップと、
前記第2の機能的結合の複数を用いて推定対象の疾患らしさを示すスコアを算出することで、疾患らしさラベルを算出するステップと、
前記チャネル組合せ毎の第1の機能的結合と前記疾患らしさラベルとを用いた機械学習により、前記推定モデルを決定するステップとを備える、推定モデル。 - ニューロフィードバックトレーニングを実行するための脳活動トレーニング装置であって、
前記ニューロフィードバックトレーニングの実行前に生成された、被験者の疾患らしさを推定するための推定モデルを格納する記憶装置と、
前記ニューロフィードバックトレーニングにおいて、前記被験者の脳波の計測データを計測するため脳波計とを備え、前記脳波の計測データは、前記被験者の頭部に配置される複数のセンサにそれぞれ対応する複数のチャネル毎の時間波形を含み、
提示装置と、
前記ニューロフィードバックトレーニングにおいて、前記脳波計からの計測データに基づいて、前記推定モデルを用いて前記被験者の疾患らしさを算出し、当該疾患らしさに対応する表示のための信号を前記提示装置に出力する処理装置とを備え、
前記推定モデルは、
前記被験者から同時に計測された脳波の計測データおよび機能的磁気共鳴画像法の計測データを取得する処理であって、当該同時に計測される脳波の計測データは、前記ニューロフィードバックトレーニングにおいて計測される脳波の計測データのそれぞれのチャネルに対応するチャネル毎の時間波形を含んでいる処理と、
前記脳波の計測データに含まれるチャネル間の相関に基づいて、チャネル組合せ毎に第1の機能的結合を算出する処理と、
前記機能的磁気共鳴画像法の計測データに含まれる関心領域間の相関に基づいて、脳内ネットワーク毎に第2の機能的結合を算出する処理と、
前記第2の機能的結合の複数を用いて推定対象の疾患らしさを示すスコアを算出することで、疾患らしさラベルを算出する処理と、
前記チャネル組合せ毎の第1の機能的結合と前記疾患らしさラベルとを用いた機械学習により、所定の第1の機能的結合を用いて、前記疾患らしさを推定することにより、前記推定モデルを決定する処理とにより生成される、脳活動トレーニング装置。 - ニューロフィードバックトレーニングを実行するための脳活動トレーニング方法であって、
前記ニューロフィードバックトレーニングの実行前に生成された、被験者の疾患らしさを推定するための推定モデルを取得するステップと、
前記ニューロフィードバックトレーニングにおいて、前記被験者の脳波の計測データを計測するステップとを備え、前記脳波の計測データは、前記被験者の頭部に配置される複数のセンサにそれぞれ対応する複数のチャネル毎の時間波形を含み、
前記ニューロフィードバックトレーニングにおいて、前記脳波の計測データに基づいて、前記推定モデルを用いて前記被験者の疾患らしさを算出し、当該疾患らしさに対応する表示のための信号を提示装置に出力するステップを備え、
前記推定モデルを取得するステップは、
前記被験者から同時に計測された脳波の計測データおよび機能的磁気共鳴画像法の計測データを取得するステップであって、当該同時に計測される脳波の計測データは、前記ニューロフィードバックトレーニングにおいて計測される脳波の計測データのそれぞれのチャネルに対応するチャネル毎の時間波形を含んでいるステップと、
前記脳波の計測データに含まれるチャネル間の相関に基づいて、チャネル組合せ毎に第1の機能的結合を算出するステップと、
前記機能的磁気共鳴画像法の計測データに含まれる関心領域間の相関に基づいて、脳内ネットワーク毎に第2の機能的結合を算出するステップと、
前記第2の機能的結合の複数を用いて推定対象の疾患らしさを示すスコアを算出することで、疾患らしさラベルを算出するステップと、
前記チャネル組合せ毎の第1の機能的結合と前記疾患らしさラベルとを用いた機械学習により、所定の第1の機能的結合を用いて、前記疾患らしさを推定することにより、前記推定モデルを決定するステップとを含む、脳活動トレーニング方法。 - ニューロフィードバックトレーニングを実行するための脳活動トレーニングプログラムであって、前記脳活動トレーニングプログラムはコンピュータに、
前記ニューロフィードバックトレーニングの実行前に生成された、被験者の疾患らしさを推定するための推定モデルを格納するステップと、
前記ニューロフィードバックトレーニングにおいて、前記被験者の脳波の計測データを取得するステップとを実行させ、前記脳波の計測データは、前記被験者の頭部に配置される複数のセンサにそれぞれ対応する複数のチャネル毎の時間波形を含み、
前記ニューロフィードバックトレーニングにおいて、前記脳波の計測データに基づいて、前記推定モデルを用いて前記被験者の疾患らしさを算出し、当該疾患らしさに対応する表示のための信号を提示装置に出力するステップを実行させ、
前記推定モデルは、
前記被験者から同時に計測された脳波の計測データおよび機能的磁気共鳴画像法の計測データを取得する処理であって、当該同時に計測される脳波の計測データは、前記ニューロフィードバックトレーニングにおいて計測される脳波の計測データのそれぞれのチャネルに対応するチャネル毎の時間波形を含んでいる処理と、
前記脳波の計測データに含まれるチャネル間の相関に基づいて、チャネル組合せ毎に第1の機能的結合を算出する処理と、
前記機能的磁気共鳴画像法の計測データに含まれる関心領域間の相関に基づいて、脳内ネットワーク毎に第2の機能的結合を算出する処理と、
前記第2の機能的結合の複数を用いて推定対象の疾患らしさを示すスコアを算出することで、疾患らしさラベルを算出する処理と、
前記チャネル組合せ毎の第1の機能的結合と前記疾患らしさラベルとを用いた機械学習により、所定の第1の機能的結合を用いて、前記疾患らしさを推定することにより、前記推定モデルを決定する処理とにより生成される、脳活動トレーニングプログラム。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202180046953.XA CN115996667A (zh) | 2020-07-02 | 2021-07-01 | 估计系统、估计方法、程序、估计模型、脑活动训练装置、脑活动训练方法以及脑活动训练程序 |
US18/014,033 US20230293036A1 (en) | 2020-07-02 | 2021-07-01 | Estimation system, estimation method, program, estimation model, brain activity training apparatus, brain activity training method, and brain activity training program |
JP2022534109A JP7465970B2 (ja) | 2020-07-02 | 2021-07-01 | 推定システム、推定方法、プログラム、推定モデル、脳活動トレーニング装置、脳活動トレーニング方法、および、脳活動トレーニングプログラム |
EP21831592.7A EP4176805A1 (en) | 2020-07-02 | 2021-07-01 | Estimation system, estimation method, program, estimation model, brain activity training device, brain activity training method, and brain activity training program |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2020-115053 | 2020-07-02 | ||
JP2020115053 | 2020-07-02 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022004841A1 true WO2022004841A1 (ja) | 2022-01-06 |
Family
ID=79316602
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2021/024927 WO2022004841A1 (ja) | 2020-07-02 | 2021-07-01 | 推定システム、推定方法、プログラム、推定モデル、脳活動トレーニング装置、脳活動トレーニング方法、および、脳活動トレーニングプログラム |
Country Status (6)
Country | Link |
---|---|
US (1) | US20230293036A1 (ja) |
EP (1) | EP4176805A1 (ja) |
JP (1) | JP7465970B2 (ja) |
CN (1) | CN115996667A (ja) |
TW (1) | TW202218626A (ja) |
WO (1) | WO2022004841A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116491958A (zh) * | 2023-06-28 | 2023-07-28 | 南昌大学第一附属医院 | 靶点确定设备、电子设备及存储介质 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140148657A1 (en) * | 2011-02-03 | 2014-05-29 | Ramoot At Tel-Aviv University Ltd. | Method and system for use in monitoring neural activity in a subject's brain |
JP2017192425A (ja) * | 2016-04-18 | 2017-10-26 | 株式会社アラヤ | 認知状態推定システム及び認知状態推定方法 |
JP2019093008A (ja) | 2017-11-27 | 2019-06-20 | 株式会社国際電気通信基礎技術研究所 | 脳内ネットワークの活動推定システム、脳内ネットワークの活動推定方法、脳内ネットワークの活動推定プログラム、および、学習済み脳活動推定モデル |
JP2020062369A (ja) * | 2018-10-11 | 2020-04-23 | 株式会社国際電気通信基礎技術研究所 | 脳機能結合相関値の調整方法、脳機能結合相関値の調整システム、脳活動分類器のハーモナイズ方法、脳活動分類器のハーモナイズシステム、および脳活動バイオマーカシステム |
-
2021
- 2021-07-01 CN CN202180046953.XA patent/CN115996667A/zh active Pending
- 2021-07-01 EP EP21831592.7A patent/EP4176805A1/en active Pending
- 2021-07-01 TW TW110124198A patent/TW202218626A/zh unknown
- 2021-07-01 US US18/014,033 patent/US20230293036A1/en active Pending
- 2021-07-01 JP JP2022534109A patent/JP7465970B2/ja active Active
- 2021-07-01 WO PCT/JP2021/024927 patent/WO2022004841A1/ja active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140148657A1 (en) * | 2011-02-03 | 2014-05-29 | Ramoot At Tel-Aviv University Ltd. | Method and system for use in monitoring neural activity in a subject's brain |
JP2017192425A (ja) * | 2016-04-18 | 2017-10-26 | 株式会社アラヤ | 認知状態推定システム及び認知状態推定方法 |
JP2019093008A (ja) | 2017-11-27 | 2019-06-20 | 株式会社国際電気通信基礎技術研究所 | 脳内ネットワークの活動推定システム、脳内ネットワークの活動推定方法、脳内ネットワークの活動推定プログラム、および、学習済み脳活動推定モデル |
JP2020062369A (ja) * | 2018-10-11 | 2020-04-23 | 株式会社国際電気通信基礎技術研究所 | 脳機能結合相関値の調整方法、脳機能結合相関値の調整システム、脳活動分類器のハーモナイズ方法、脳活動分類器のハーモナイズシステム、および脳活動バイオマーカシステム |
Non-Patent Citations (7)
Title |
---|
ANDREW T DRYSDALEL ET AL.: "Resting-state connectivity biomarkers define neurophysiological subtypes of depression", NATURE MEDICINE, vol. 23, no. 1, 2017, pages 28 - 38, XP055453891, ISSN: ISSN: 1546-170X, DOI: 10.1038/nm.4246 |
ENIK6 BARTOK ET AL.: "Cognitive functions in prepsychotic patients", PROGRESS IN NEURO-PSYCHOPHARMACOLOGY & BIOLOGICAL PSYCHIATRY, vol. 29, 2005, pages 621 - 625, XP025311087, DOI: 10.1016/j.pnpbp.2005.01.008 |
HIRAYAMA J, HYVARINEN AJ, KAWANABE M: "SPLICE: Fully tractable hierarchical extension of ICA with pooling", PROCEEDINGS OF THE 34 TH INTERNATIONAL CONFERENCE ON MACHINE LEARNING, 1 August 2017 (2017-08-01), pages 1 - 10, XP055897233 * |
NAHO ICHIKAWA ET AL.: "Primary functional brain connections associated with melancholic major depressive disorder and modulation by antidepressants", SCIENTIFIC REPORTS, vol. 10, 2020, pages 3542, Retrieved from the Internet <URL:https://doi.org/10.1038/s41598-020-60527-z>> |
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, vol. 20, no. 10, 17 July 2017 (2017-07-17), pages 769 - 781, XP055589968, DOI: 10.1093/ijnp/pyx059 |
TERESA MURTA, MARCO LEITE, DAVID W. CARMICHAEL, PATRíCIA FIGUEIREDO, LOUIS LEMIEUX: "Electrophysiological correlates of the BOLD signal for EEG-informed fMRI : Electrophysiological Correlates of the BOLD Signal", HUMAN BRAIN MAPPING, vol. 36, no. 1, 1 January 2015 (2015-01-01), pages 391 - 414, XP055618887, ISSN: 1065-9471, DOI: 10.1002/hbm.22623 * |
YUJIRO YOSHIHARA ET AL.: "Overlapping but asymmetrical relationships between schizophrenia and autism revealed by brain connectivity", BIORXIV, 7 September 2018 (2018-09-07), Retrieved from the Internet <URL:URL:https://doi.org/10.1101/403212>> |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116491958A (zh) * | 2023-06-28 | 2023-07-28 | 南昌大学第一附属医院 | 靶点确定设备、电子设备及存储介质 |
CN116491958B (zh) * | 2023-06-28 | 2023-09-19 | 南昌大学第一附属医院 | 靶点确定设备、电子设备及存储介质 |
Also Published As
Publication number | Publication date |
---|---|
JP7465970B2 (ja) | 2024-04-11 |
US20230293036A1 (en) | 2023-09-21 |
CN115996667A (zh) | 2023-04-21 |
TW202218626A (zh) | 2022-05-16 |
JPWO2022004841A1 (ja) | 2022-01-06 |
EP4176805A1 (en) | 2023-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20210169395A1 (en) | Method and system for use in monitoring neural activity in a subject's brain | |
Deligianni et al. | Relating resting-state fMRI and EEG whole-brain connectomes across frequency bands | |
JP7221693B2 (ja) | ヒトの脳の皮質機能の目録を作るための方法および磁気画像化デバイス | |
Huster et al. | Methods for simultaneous EEG-fMRI: an introductory review | |
JP6884344B2 (ja) | 脳内ネットワークの活動推定システム、脳内ネットワークの活動推定方法、脳内ネットワークの活動推定プログラム、および、学習済み脳活動推定モデル | |
Gilbert et al. | Object repetition leads to local increases in the temporal coordination of neural responses | |
US20240057888A1 (en) | Method And System For Determining Brain-State Dependent Functional Areas Of Unitary Pooled Activity And Associated Dynamic Networks With Functional Magnetic Resonance Imaging | |
JP2020062369A (ja) | 脳機能結合相関値の調整方法、脳機能結合相関値の調整システム、脳活動分類器のハーモナイズ方法、脳活動分類器のハーモナイズシステム、および脳活動バイオマーカシステム | |
WO2012151453A2 (en) | Seizure detection and epileptogenic lesion localization | |
Novikov et al. | TMSmap–software for quantitative analysis of TMS mapping results | |
US11147501B2 (en) | Systems and methods for enhanced encoded source imaging | |
Wang et al. | Concordance of MEG and fMRI patterns in adolescents during verb generation | |
WO2022004841A1 (ja) | 推定システム、推定方法、プログラム、推定モデル、脳活動トレーニング装置、脳活動トレーニング方法、および、脳活動トレーニングプログラム | |
Khayretdinova et al. | Predicting age from resting-state scalp EEG signals with deep convolutional neural networks on TD-brain dataset | |
Abreu et al. | Optimizing EEG source reconstruction with concurrent FMRI-derived spatial priors | |
Ziani | Enhancing fetal electrocardiogram classification: A hybrid approach incorporating multimodal data fusion and advanced deep learning models | |
DeRamus et al. | Modular and state-relevant functional network connectivity in high-frequency eyes open vs eyes closed resting fMRI data | |
Aram et al. | Model-based estimation of intra-cortical connectivity using electrophysiological data | |
Candemir et al. | Automatic detection of emotional changes induced by social support loss using fmri | |
Olson et al. | Automated machine learning pipeline framework for classification of pediatric functional nausea using high-resolution electrogastrogram | |
Zafar et al. | A study of decoding human brain activities from simultaneous data of EEG and fMRI using MVPA | |
Ilias et al. | Overview of methods and available tools used in complex brain disorders | |
US20240156414A1 (en) | Method for Predicting Age from Resting-State Scalp EEG Signals Using Deep Convolutional Neural Networks | |
Soler et al. | Automated methodology for optimal selection of minimum electrode subsets for accurate eeg source estimation based on genetic algorithm optimization | |
Cordón Avila | Localization of sources in electroencephalographic registers during working memory tasks |
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: 21831592 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2022534109 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2021831592 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2021831592 Country of ref document: EP Effective date: 20230202 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |