EEG Feature Selection via Frequency Domain Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing electroencephalogram (EEG) signals face challenges in selecting pertinent features for discriminating between brain conditions due to the complexity of the signals, low signal-to-noise ratio, and insufficient spatial resolution, making it difficult to reliably identify brain functions and conditions such as Alzheimer’s disease, Mild Cognitive Impairment, and Subjective Cognitive Impairment.
Innovation Solution
A method that computes values quantifying brain activity for each electrode, performs thresholding on these values, and uses a feature selector to rank and select relevant feature vectors based on electrode, frequency, and threshold percentage, allowing for the discrimination of brain conditions like Alzheimer’s disease, Mild Cognitive Impairment, and Subjective Cognitive Impairment, even with limited EEG electrodes and low signal quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional EEG analysis methods are used to extract information from EEG signals, then brain activity can be quantified, but the identification of brain regions of interest becomes difficult due to the need to specify electrode groups based on subject position and study-dependent criteria
Solution Approach 1:
The patent applies parameter changes by transforming the EEG analysis from spatial domain (electrode selection based on position) to frequency domain (spectral features). The method computes power spectral density and extracts features based on frequency bands (delta, theta, alpha, beta, gamma) rather than relying on specific electrode locations, thereby resolving the complexity of electrode selection while maintaining measurement precision
Solution Approach 2:
The patent transitions from analyzing EEG signals in the time/space domain to the frequency domain through spectral analysis. By computing power spectral density and analyzing frequency components, the method adds a dimensional transformation that eliminates the need for subject-specific electrode positioning while preserving brain activity information
2Device complexity
If EEG signals are smoothed by averaging over multiple electrodes to reduce complexity, then analysis becomes simpler, but spatial resolution and ability to identify specific brain functions deteriorate
Solution Approach 1:
The patent resolves this contradiction by transforming the analysis to the frequency domain where signals are decomposed into spectral components. This dimensional change allows maintaining information from individual electrodes without requiring spatial averaging, thus preserving spatial resolution while simplifying analysis through frequency-based feature extraction
3Ease of manufacture
If visual analysis of EEG signals is performed to identify brain conditions, then non-invasive detection is achieved, but reliability is insufficient due to signal complexity, low signal-to-noise ratio, and small variations
Solution Approach 1:
The patent replaces manual visual analysis with automated computational methods. By implementing algorithms that compute power spectral density, extract spectral features, and apply classification rules, the system substitutes human visual inspection with machine-based automated analysis, thereby improving reliability while maintaining the non-invasive nature of EEG detection
Solution Approach 2:
The patent introduces spectral features as an intermediary between raw EEG signals and brain condition diagnosis. By computing power spectral density and extracting meaningful frequency-domain features, the method creates an intermediate representation that enhances signal-to-noise ratio and makes automated classification more reliable compared to direct visual analysis
4Measurement precision
If graph theory methods are used to analyze brain functional connectivity, then network topology can be studied, but the choice of connectivity threshold becomes study-dependent and varies across different brain regions
Solution Approach 1:
The patent applies parameter changes by shifting from spatial connectivity analysis (graph theory requiring threshold selection) to frequency-domain spectral analysis. By computing power spectral density and analyzing frequency components at each electrode, the method eliminates the need for connectivity threshold selection while maintaining precise measurement of brain activity patterns
Data Source
AI summary
A method for selecting features for the detection of a given brain condition, the method using a feature selector able to discriminate between at least two brain conditions, and using a plurality of electroencephalogram signals, relative to several brain electrodes and filtered on at least one frequency, the method comprising the following steps:e) for each signal, computing at least one value quantifying brain activity for each brain electrode;f) for each signal, performing a thresholding of said computed quantifying values according to at least one threshold percentage, thus forming at least one group of thresholded values corresponding to each signal;g) using said feature selector to rank said thresholded values; andh) based on this ranking, selecting at least one feature vector comprising at least one electrode, and one frequency and/or one threshold percentage, and corresponding to the best ranked thresholded values for the detection of said given brain condition.