Subject-Specific Pre-Processing Pipeline for EEG Artifact Removal Aimed at Enhancing SSVEP Detection

Steady-State Visually Evoked Potentials (SSVEP) based Brain-Computer Interfaces (BCIs) offer significant potential in assistive and rehabilitation technologies. However, the performance of SSVEP detection is often compromised by artifacts in Electroencephalography (EEG) signals and inter-subject variability, among other factors. This study introduces a subject-specific p reprocessing p ipeline a imed a t enhancing SSVEP signal detection accuracy. The proposed method involves a dual-stage approach, utilizing EEG signals recorded both during a baseline rest state and during tasks known to induce artifacts. By establishing a personalized reference for each subject, the pipeline enables more accurate artifact detection and removal. The pipeline's efficacy w as e valuated u sing Canonical Correlation Analysis (CCA) on an open EEG dataset, with performance assessed across varying time segments (1, 2, 3 y 4 s). Detection accuracy was calculated to quantify performance in SSVEP detection. The performance of the proposed method for the 4-second time segment was superior (96.44%) compared to the standard method (93.78%). The results demonstrate the pipeline's potential to improve the reliability of SSVEP-based BCIs, particularly in highly contaminated signals.