EEG Motor Imagery Classification Using Multi-Domain Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motor imagery classification technologies using EEG signals face challenges in accuracy due to noise contamination and the inability to effectively extract features from multiple domains, limiting the performance of brain-computer interface applications.
Innovation Solution
A motor imagery classification apparatus and method that extracts features from EEG signals by mapping them into a matrix structure, analyzing spatial and temporal features using CNN and RNN models, and classifying intentions through a deep learning neural network, while applying the Common Spatial Pattern algorithm and filtering to enhance feature extraction and noise reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning models are used to extract features from EEG signals, then the accuracy of motor imagery intention prediction is improved, but the models can only extract single domain features which limits overall performance
Solution Approach 1:
The patent segments the feature extraction process into three independent domain-specific modules: spatial domain feature extraction, temporal domain feature extraction, and frequency domain feature extraction. Each module uses appropriate deep learning models tailored to its domain characteristics, allowing simultaneous extraction of multiple domain features without interference, thereby resolving the contradiction between accuracy and adaptability.
Solution Approach 2:
The patent creates a universal multi-domain feature extraction framework that can handle spatial, temporal, and frequency domains within a single system. This multi-functional architecture enables the system to extract features from all three domains simultaneously, making the model adaptable to different feature types while maintaining high prediction accuracy through domain-specific processing.
2Measurement precision
If features from multiple domains are combined to improve performance, then the classification accuracy increases, but the complexity of feature selection and model design increases significantly
Solution Approach 1:
The patent divides the complex multi-domain feature extraction task into three separate, manageable modules corresponding to spatial, temporal, and frequency domains. Each module has its own dedicated deep learning model and feature selection process, which reduces the overall complexity compared to a monolithic approach while still achieving high classification accuracy through the combination of domain-specific features.
Solution Approach 2:
The patent implements a feedback mechanism where the performance of each domain-specific feature extraction module is evaluated independently, and the features are selectively combined based on their contribution to overall classification accuracy. This feedback-driven approach simplifies the feature selection process by providing clear guidance on which domain features are most valuable, reducing the complexity of model design.
3Productivity
If EEG signals are analyzed in real-time for motor imagery classification, then the practical application value increases, but noise contamination and low accuracy of real-time feature classification hinder effective implementation
Solution Approach 1:
The patent segments the real-time EEG signal processing into three parallel domain-specific processing streams (spatial, temporal, and frequency domains). This segmentation allows each stream to process features independently with optimized algorithms for its specific domain, improving real-time classification accuracy while maintaining high processing speed through parallel computation.
Solution Approach 2:
The patent converts the challenge of noise contamination into an advantage by implementing domain-specific noise filtering and feature selection mechanisms. Each domain module independently identifies and extracts robust features that are less susceptible to noise, and the combination of multiple domain features provides redundancy that further mitigates noise impact, thereby improving reliability in real-time applications.
Data Source
AI summary
The present disclosure relates to an apparatus and method for motor imagery classification using electroencephalography (EEG), and more particularly, to an apparatus and method for motor imagery classification that extracts features in different domains included in EEG signals generated during motor imagery in real time and classifies a user's intentions using the features.