Leveraging Grating-Based Flickers: A Leap Toward Practical, Visually Comfortable, and High-Performance Dry EEG Code-VEP BCI

Abstract

AbstractO_ST_ABSPurposeC_ST_ABSReactive Brain-Computer Interfaces (rBCIs) typically rely on repetitive visual stimuli, which can strain the eyes and cause attentional distraction. To address these challenges, we propose a novel approach rooted in visual neuroscience to design visual Stimuli for Augmented Response (StAR). The StAR stimuli consist of small randomly-oriented Gabor or Ricker patches that optimize foveal neural response while reducing peripheral distraction. MethodsIn a factorial design study, 24 participants equipped with an 8-dry electrodes EEG system focused on series of target flickers presented under three formats: traditional Plain flickers, Gabor-based, or Ricker-based flickers. These flickers were part of a five-classes Code Visually Evoked Potentials (c-VEP) paradigm featuring low frequency, short, and aperiodic visual flashes. ResultsSubjective ratings revealed that Gabor and Ricker gratings were visually comfortable and nearly invisible in peripheral vision compared to plain flickers. Moreover, Gabor and Ricker-based textures achieved higher accuracy (93.6% and 96.3%, respectively) with only 88 seconds of calibration data, compared to plain flickers (65.6%). A follow-up online implementation of this experiment was conducted to validate our findings within the frame of naturalistic operations. During this trial, remarkable accuracies of 97.5% in a cued task and 94.3% in an asynchronous digicode task were achieved, with a mean decoding time as low as 1.68 seconds. ConclusionThis work demonstrates the potential to expand BCI applications beyond the lab by integrating visually unobtrusive systems with gel-free, low density EEG technology, thereby making BCIs more accessible and efficient. The datasets, algorithms, and BCI implementations are shared through open-access repositories.