Complex tensor based blind source separation of EEG for tracking P300 subcomponents.

Area of Science:

• Neuroscience
• Signal Processing
• Biomedical Engineering

Background:

• Event-related potentials (ERPs), such as P300, are crucial for understanding cognitive information processing.
• P300 comprises correlated subcomponents (P3a and P3b) that challenge traditional source separation techniques.
• Existing methods struggle to effectively disentangle these correlated brain signals.

Purpose of the Study:

• To develop a novel method for complex tensor factorization of correlated brain sources.
• To accurately separate the P3a and P3b subcomponents of the P300 event-related potential.
• To enable tracking of P3a and P3b variations for improved brain state assessment.

Main Methods:

• Introduction of a complex-valued tensor factorization technique applied to electroencephalography (EEG) data.
• Utilization of complex-valued statistics to leverage correlations within the EEG signals.
• Comparison of the proposed method with spatial principal component analysis (SPCA).

Main Results:

• The proposed complex-valued tensor factorization successfully separates correlated P300 subcomponents.
• The method effectively exploits data correlations using complex-valued statistics.
• Tracking of P3a and P3b variations becomes feasible for brain state analysis.

Conclusions:

• Complex tensor factorization offers a powerful approach for separating correlated brain sources like P300 subcomponents.
• The developed method provides a significant advancement in analyzing cognitive processes via EEG.
• This technique holds promise for more accurate brain state assessment and related research.