MAICA: an ICA-based method for source separation in a low-channel EEG recording

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Independent Component Analysis (ICA) is a powerful technique for signal decomposition.
• Applying ICA to low-channel electroencephalography (EEG) recordings presents challenges due to limited data.
• Existing methods may struggle to effectively extract meaningful components from sparse EEG data.

Purpose of the Study:

• To introduce a novel method, Moving Average ICA (MAICA), for applying ICA to low-channel EEG recordings.
• To demonstrate the efficacy of MAICA in signal decomposition and artifact identification.
• To evaluate MAICA's performance in real-time applications.

Main Methods:

• MAICA extends low-sensor EEG signal matrices using zero-phase moving average filters.
• The algorithm's theoretical background is discussed.
• MAICA's performance is validated using a mathematical system, 64-channel EEG data, and motor imagery BCI data.

Main Results:

• MAICA successfully decomposed mixed sinusoidal signals into source components with high correlation (99%-100%).
• On 5-channel EEG, MAICA identified more artifact components than classic ICA on 64-channel data.
• MAICA operated in real-time with minimal delay (<6ms per trial).

Conclusions:

• MAICA enables effective ICA application on low-channel EEG recordings.
• The method significantly improves artifact detection and component separation.
• MAICA is suitable for real-time medical signal processing applications requiring pattern retrieval from limited sensor data.