Steady-State Visual Evoked Potential based Brain-Computer Interface using Adaptive Correlation Technique

Extensive research in the area of Brain-Computer Interface (BCI) System has provided immense strength towards several improvements, which can widely benefit end-users. However, the Eigenvalue problem, illiteracy towards the BCI system, unsatisfactory EEG signal detection and unexciting user experience remain critical and challenging issues. The separation and extraction of SSVEP components from the non-SSVEP component is a challenging process. Thus, a robust and adaptable technique is required to enhance the efficiency of the BCI system. In this paper, Adaptive Correlation-based Component Analysis (ACCA) method is presented to enhance classification accuracy and to detect SSVEP components from EEG signals accurately. Besides, the practicality of the proposed ACCA method is analysed using an efficient encoding strategy. The correlation coefficients between single and multiple flickers are obtained to determine spatial probabilities. Further, adaptive filters are utilized to obtain SSVEP components for single and multiple flickers. The main objective of proposed Adaptive Correlation based Component Analysis (ACCA) method is to create maximum correlation among EEG signals for a specific time window invoked by single flicker as well as by multiple flickers for a group of specific frequencies with the help of maximum inter-assignment correlation technique, so that SSVEP components are effectively obtained by reducing eigen value problem. The performance of the proposed ACCA method is tested on a benchmark speller EEG dataset. The performance efficiency of the proposed ACCA method is measured against several traditional SSVEP extraction techniques considering performance matrices like accuracy, Information Transfer Rate (ITR) in bits per minute, and time segment in seconds.