Detection of brain stimuli using Ramanujan periodicity transforms

Area of Science:

• Neuroscience
• Signal Processing
• Biomedical Engineering

Background:

• Reliable brain-computer interfacing (BCI) relies on accurately matching brain responses to visual stimuli.
• Steady-state visually evoked potentials (SSVEPs) are key signals for SSVEP-based BCI.

Purpose of the Study:

• To develop and analyze a novel SSVEP detection method using Ramanujan periodicity transform (RPT).
• To improve the efficiency and accuracy of SSVEP detection for real-time BCI applications.

Main Methods:

• SSVEPs are modeled with sparse representations in an RPT dictionary.
• A generalized likelihood ratio test (GLRT) is developed for binary SSVEP detection.
• The approach is extended to multi-hypothesis, multi-electrode settings, incorporating spatial correlation using pre-stimulus data.

Main Results:

• The RPT detector significantly outperforms spectral-based methods and state-of-the-art CCA/FBCCA.
• Achieves superior accuracy and lower sample complexity in short data length regimes.
• Demonstrates asymptotic optimality, closing the gap to perfect measurement bounds.

Conclusions:

• The RPT detector advances real-time BCI by offering a robust and data-efficient SSVEP detection framework.
• Provides a new standard for evaluating SSVEP detection schemes across various operating conditions.