Enhancing motor imagery EEG signal decoding through machine learning: A systematic review of recent progress

Area of Science:

• Neuroscience
• Deep Learning
• Biomedical Engineering

Background:

• Electroencephalogram (EEG) is a key non-invasive tool for measuring brain activity due to its high temporal resolution, usability, and safety.
• Brain-Computer Interfaces (BCIs) utilizing EEG signals offer a novel communication pathway for individuals with motor disabilities and neurological disorders.
• Real-world application of EEG-based BCIs for motor imagery recognition faces challenges like inter-individual variability and low signal-to-noise ratio (SNR).

Purpose of the Study:

• To systematically review the intersection of neuroscience and deep learning for decoding motor imagery EEG signals.
• To provide a comprehensive summary of studies published since 2017, focusing on datasets, preprocessing, feature extraction, and deep learning models.
• To serve as a resource for researchers and practitioners aiming to improve EEG-based BCI systems for motor disability assistance.

Main Methods:

• Systematic literature review of studies from 2017 onwards.
• Focus on analysis of datasets used for motor imagery EEG signal decoding.
• Examination of preprocessing techniques, feature extraction methods, and deep learning architectures applied in BCI research.

Main Results:

• Identified key trends and methodologies in deep learning applications for EEG-based motor imagery.
• Highlighted common datasets and preprocessing pipelines used in the field.
• Summarized various deep learning models and their performance in BCI applications.

Conclusions:

• Deep learning shows significant promise in decoding motor imagery EEG signals for BCI applications.
• Addressing challenges in signal variability and SNR is crucial for practical BCI implementation.
• This review provides insights to advance the development of brain-computer interfaces, bridging human cognition and machine interaction.