Research Article: Dynamic time window mechanism for time synchronous VEP-based BCIs—Performance evaluation with a dictionary-supported BCI speller employing SSVEP and c-VEP

Date Published: June 13, 2019

Publisher: Public Library of Science

Author(s): Felix Gembler, Piotr Stawicki, Abdul Saboor, Ivan Volosyak, Zhishun Wang.


Brain-Computer Interfaces (BCIs) based on visual evoked potentials (VEPs) allow high communication speeds and accuracies. The fastest speeds can be achieved if targets are identified in a synchronous way (i.e., after a pre-set time period the system will produce a command output). The duration a target needs to be fixated on until the system classifies an output command affects the overall system performance. Hence, extracting a data window dedicated for the classification is of critical importance for VEP-based BCIs. Secondly, unintentional fixation on a target could easily lead to its selection. For the practical usability of BCI applications it is desirable to distinguish between intentional and unintentional fixations. This can be achieved by using threshold-based target identification methods. The study explores personalized dynamic classification time windows for threshold-based time synchronous VEP BCIs. The proposed techniques were tested employing the SSVEP and the c-VEP paradigm. Spelling performance was evaluated using an 8-target dictionary-supported BCI utilizing an n-gram word prediction model. The performance of twelve healthy participants was assessed with the information transfer rate (ITR) and accuracy. All participants completed sentence spelling tasks, reaching average accuracies of 94% and 96.3% for the c-VEP and the SSVEP paradigm, respectively. Average ITRs around 57 bpm were achieved for both paradigms.

Partial Text

Brain-Computer Interfaces (BCIs) detect, analyze, and decode brain activities to provide communication with the external environment, without involving any muscle activities [1]. The brain activities are usually recorded non-invasively by an electroencephalogram (EEG). BCIs may be used as a communication tool for severely impaired people [2, 3].

The main purpose of the presented study was to investigate methods of dynamic gaze classification time windows for time synchronous VEP BCIs. The proposed methods lead to a more natural user-BCI interaction. To demonstrate the robustness of the approach, a dictionary-driven spelling application was tested with the SSVEP and the c-VEP paradigm. In this sense, the study also provides a direct comparison between c-VEP and SSVEP stimulation, both in terms of performance and user-friendliness.

A dynamic classification time window approach for time synchronous VEP BCIs was proposed. The optimal time window was determined individually deduced from a training session which was also used for the generation of templates and spatial filters. An 8-target spelling application utilizing n-gram-based word suggestions was used to evaluate the usability of the developed methods. Twelve participants tested the system in on-line spelling tasks with the SSVEP and the c-VEP paradigm. The presented study demonstrates the robustness of the proposed approach. All participants completed sentence spelling and word spelling tasks with accuracies well above 90% for the two paradigms. Ensemble-based classification strategies were employed in both cases. The proposed methods were equally effective for c-VEP and SSVEP based systems in terms of ITR; mean ITRs of approximately 57 bpm were achieved in both cases. Nevertheless, in word spelling tasks, the c-VEP system (mean word spelling ITR 92.7 bpm) outperformed the SSVEP system (ITR 75.1 bpm). In terms of user-friendliness however, the SSVEP paradigm was preferred by most participants. The results suggest that the stimulation pattern (SSVEP vs. c-VEP) could be selected based on the user preference. In terms of speed, the optimal paradigm could be determined individually for each user in a short training session. However, the perceived level of user-friendliness should also be taken into account, as it might be more relevant for end users than pure system speed.