Related Experiment Videos
Synergistic EEG signal processing for brain-computer interfaces using hybrid MothCray optimization and deep learning
Annu Kumari1, Damodar Reddy Edla1, Dharavath Ramesh2
1Department of Computer Science and Engineering, National Institute of Technology Goa, Cuncolim, Goa, 403703, India.
Neuroscience
|August 4, 2026
Summary
This study introduces an optimized framework for electroencephalography (EEG) brain-computer interfaces (BCIs). Our novel moth-crayfish optimization (MCO) improves channel selection and classification accuracy for practical BCI applications.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Computer Science
Background:
- Electroencephalography (EEG) based brain-computer interface (BCI) systems necessitate efficient channel selection for optimal performance.
- High signal quality, reduced complexity, and robust usability are critical for practical EEG-BCI deployment.
Purpose of the Study:
- To develop a comprehensive signal-processing framework to enhance the efficiency and accuracy of EEG-driven BCIs.
- To introduce a novel moth-crayfish optimization (MCO) algorithm for optimal EEG channel selection.
- To improve the practical applicability of BCIs through reduced channel usage and enhanced portability.
Main Methods:
- Signal pre-processing included notch filtering, independent component analysis (ICA), and temporal segmentation for artifact removal.
- The moth-crayfish optimization (MCO) algorithm, a hybrid of moth flame and crayfish optimization, was employed for informative channel selection.
- Feature extraction involved time-domain statistics, frequency-domain attributes, and connectivity measures, followed by classification using a tailored deep neural network (EEGNet).
Main Results:
- The proposed framework achieved a classification accuracy of 93.92% on the BCI competition IV dataset IIa.
- The MCO-based channel selection and enhanced EEGNet classifier outperformed existing methods.
- The system demonstrated effectiveness for applications requiring fewer EEG channels, enabling portable and comfortable headsets.
Conclusions:
- The MCO-based channel selection and multi-domain feature fusion significantly advance EEG-BCI performance.
- The enhanced EEGNet classifier provides robust and accurate classification for spatial-temporal EEG data.
- This framework supports practical BCI deployment in neurorehabilitation and assistive communication, offering efficient resource utilization and stable performance.