Abstract
Visual Evoked Potential (VEP)-based Brain–Computer Interfaces (BCIs) are common due to their advantages. Particularly Steady-State VEP (SSVEP) and code-modulated VEP (c-VEP) BCIs are dominating the field. While SSVEP BCIs are more practical and easy to be implemented with common displays, c-VEP BCIs require synchronization but provide more targets with sufficiently long encoding codes. However, a novel paradigm as known as ‘Quasi-Steady State VEP (QSS-VEP)’ incorporates the advantages of both c-VEP and SSVEP paradigms. With specially designed low jittered stimulation sequences and the well-established Continuous Loop Averaging Deconvolution (CLAD) method, it enables simultaneous acquisition of both QSS-VEP and transient VEPs at higher stimulation rates. In this study, pattern reversal stimulation at 10, 32, 50, and 70 reversal per second (rps) stimuli are presented to a dual display stimulator. The optimal stimulation rate and the feasibility of QSS-VEP paradigm in a simple BCI system are investigated. Moreover, a comparison of the QSS-VEP-based template matching and SSVEP-based template matching is given for one low (10rps) and one medium stimulation rate (32rps). It has been shown that the QSS-VEP method increases the SNR and therefore single sweep QSS-VEPs provide significantly higher performance in template matching-based classification in a BCI system. Recently, transient VEP retrieval from linear regression and random code estimation from the EEG using these transient VEPs as unitary signals has gained attention due to high performance in BCI applications. The CLAD method-based transient VEP extraction is an easier, yet efficient way compared to linear regression-based methods. Thus, EEG codes can be recognized with sufficiently high accuracy using the QSS-VEP and CLAD-based BCI paradigm in addition to extraction of clinically relevant transient VEP at high rates.