Enhancing Detection of SSVEP-based BCIs Using Adjacent Frequencies Fusion Method

Brain-computer interfaces (BCIs) have emerged as transformative technologies, enabling direct communication between the human brain and external devices. Steady-state visual evoked potentials (SSVEP) have gained particular attention due to their potential in BCIs. Current decoding algorithms do not take advantage of the correlation coefficients of adjacent frequencies. We propose adjacent frequencies fusion filter bank canonical correlation analysis (AFF-FBCCA), which enhances accuracy and robustness by utilizing information from adjacent frequencies. This weighted fusion aims to capitalize on the inherent similarity between electroencephalogram (EEG) signals at closely spaced frequencies. The determination of weight coefficients, incorporating dynamic adjustments based on the time window, further contributes to the adaptability of AFF-FBCCA. The proposed method is validated using public benchmark datasets. The results show that AFF-FBCCA is always superior to standard FBCCA in terms of classification accuracy and information transfer rate (ITR) in all test time windows. This method maintains the advantage of training-free and provides a more accurate and user-friendly solution for SSVEP-based BCI.