System and method for calculation of an index of brain activity
Inventors
HEDDI, Mohcine • Le Van Quyen, Michel • LE DOUGET, Jean-Eudes
Assignees
Bioserentiy • Centre National de la Recherche Scienttifique • Centre National de la Recherche Scientifique CNRS • Assistance Publique Hopitaux de Paris APHP • Institut National de la Sante et de la Recherche Medicale INSERM • Sorbonne Universite • Bioserenity SAS • Institut du Cerveau et de La Moelle Epiniere ICM
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11642073-B2
Publication Date
2023-05-09
Expiration Date
Abstract
A system for calculating an indicator associated to a brain activity of a subject, the system including an acquisition module configured to acquire at least an epoch of electroencephalographic signal of a subject from a plurality of electrodes and a data processing module configured to carry out the steps of: calculating an average vector (VA) using as input of an autoencoder neural network (aNN) an electroencephalographic signals (ES) of a subject acquired from a plurality of electrodes; detecting (DET) the presence of at least a predefined pattern in the consecutive average values of the average vector (VA); and generating an indicator of brain activity (Idx) of the subject when detecting the predefined pattern.
Core Innovation
The invention relates to a system for calculating an indicator associated to a determined brain activity of a subject using electroencephalographic signals acquired from a plurality of electrodes. The system receives at least an epoch of electroencephalographic signals, generates an input matrix from the electroencephalographic signals, and provides the input matrix as input to an auto-encoder neural network. The auto-encoder generates a reconstructed output matrix from the input matrix.
The system generates a loss values vector by linear combination of the input matrix and the reconstructed output matrix, where each element of the loss values vector is associated to a channel. For each epoch, an average value of the elements of the loss values vector is calculated, and the calculating step is repeated multiple times for consecutive electroencephalographic epochs. This repetition generates an average vector comprising average values obtained for each consecutive epoch.
The system detects the presence of at least a predefined pattern in the consecutive average values of the average vector. When the predefined pattern is detected, the system generates an indicator of brain activity of the subject, where the indicator is representative of the determined brain activity, and an output generator reports the indicator of the brain activity. In the described embodiments, the predefined pattern can be detected based on consecutive averaged values and/or additional constraints.
Claims Coverage
The independent claims are configured to calculate an indicator associated with a determined brain activity by repeatedly processing consecutive electroencephalographic epochs into an average vector and detecting a predefined pattern. Across the independent claims, the main inventive workflow includes input-matrix reconstruction via an auto-encoder, channel-associated loss values, consecutive-epoch average-vector formation, pattern detection, and reporting via an output generator.
Indicator generation from consecutive epoch average-vector pattern detection
Detecting the presence of at least a predefined pattern in the consecutive average values of the average vector, and generating an indicator of brain activity when the predefined pattern is detected, with an output generator for reporting the indicator representative of the determined brain activity.
Input-matrix reconstruction using an auto-encoder neural network
Receiving at least an epoch of electroencephalographic signals acquired from a plurality of electrodes; generating an input matrix having the n dimension equal to N, number of channels of the electroencephalographic recording; providing the input matrix as input to an auto-encoder neural network; and generating a reconstructed output matrix using the auto-encoder.
Channel-associated loss values vector from input and reconstructed output
Generating a loss values vector by linear combination of the input matrix and output matrix, wherein each element of the loss value vector is associated to a channel.
Average-vector formation across consecutive electroencephalographic epochs
Calculating an average value of the elements of the loss values vector, repeating the calculating step multiple times to generate an average vector comprising the average values obtained for each consecutive electroencephalographic epoch.
Both independent claims share the same core calculation pipeline: per-epoch input-matrix encoding and reconstruction using an auto-encoder, computing channel-associated loss values, forming an average vector across consecutive epochs, detecting a predefined pattern in the consecutive average values, and generating and reporting an indicator representative of the determined brain activity.
Stated Advantages
Provides an indicator of brain activity representative of the determined brain activity by detecting a predefined pattern in consecutive average values derived from EEG epochs.
Documented Applications
Seizure detection is reported using seizure-detection performance on 23 pediatric subjects from the CHB-MIT scalp EEG database.
Computing an indicator associated to a determined brain activity is described for pathological brain activity such as epileptic seizures.
Interested in licensing this patent?