An Integrated Stimulus-to-Analysis Pipeline for Consumer-Grade SSVEP BCIs: A Pilot Comparison of Checkerboard, Dartboard, and Flicker Stimuli

Running steady-state visual evoked potential (SSVEP) experiments on consumer-grade EEG hardware typically requires combining separate tools for stimulus presentation, data acquisition, and offline analysis. We present a single Python tool that integrates all three stages: it designs the stimuli, presents them on screen, and processes the resulting recordings. The stimulus module supports static images, phase-reversing checkerboard and dartboard patterns, and full-field flicker. The analysis pipeline reads the EDF recording, segments it by marker, computes Welch power spectral density (PSD), estimates the signal-to-noise ratio (SNR) relative to the nearest non-harmonic bins, generates spectrograms, and applies canonical correlation analysis (CCA) against sine/cosine references at the target frequency and its second harmonic. To validate the tool, we conducted a pilot study using a 14-channel Emotiv EPOC X and a laptop driving a 60 Hz external display, recording one participant at 12 Hz across three stimulus types. Contrary to expectation, the dartboard condition yielded the highest SNR at O2 (peak 2.82, mean 2.17), followed by the checkerboard (maximum 2.35, mean 1.97 over five full-length trials) and the full-field flicker (maximum 1.79, mean 1.33). This ordering is the reverse of that predicted by total luminance modulation. Part of the difference may reflect a stimulus confound rather than a physiological effect: the on–off flicker may have responded at a 6 Hz fundamental, with only its second harmonic falling at the 12 Hz analysis frequency, which the present data could not confirm. Single-channel CCA did not exceed r = 0.25 in any condition, indicating that multi-channel spatial filtering is the primary next step. The tool operated end-to-end, and the per-condition ranking, although unexpected, was consistent between the PSD- and CCA-based metrics within this single participant. These findings are preliminary observations from a one-participant pilot rather than an established effect and require confirmation through more comprehensive, multi-subject measurements.