RVEB—An algorithm for recognizing voluntary eye blinks based on the signal recorded from prefrontal EEG channels. (May 2020)

• Record Type: Journal Article
• Authors: Rejer, Izabela; Cieszyński, Łukasz
• Is Part Of: Biomedical signal processing and control. Volume 59(2020)
• Publication Date: 2020-05
• Subjects: EOG activity -- EEG signal -- Human-machine interface -- Eye blinks -- Blinking patterns recognition; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2020.101876
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Eye blinks can establish a successful communication channel for Human-Computer interactions. RVEB algorithm can recognize 4 blinking schemes with 95 % accuracy and high speed (1 blinking scheme in 2 s). RVEB algorithm can recognize 9 blinking schemes with 94 % accuracy and only two EEG channels (Fp1 and Fp2). Abstract: Objective: The paper presents an algorithm for recognizing voluntary eye blinks (RVEB). The algorithm is based on the analysis of time waveforms recorded from two prefrontal EEG (electroencephalographic) channels. Two versions of the algorithm are provided: 1) for recognizing different numbers of winks with the left or right eye (left-right RVEB algorithm); 2) for recognizing different numbers of winks with the left or right eye, or different numbers of blinks with both eyes (left-right-both RVEB algorithm). Methods: The paper provides a description of the algorithm and its verification via a two-session experiment conducted with 12 healthy subjects. Results: (i) the average detection accuracy reached 96.9% and 93.8% for the left-right and left-right-both RVEB algorithm, respectively; (ii) the recognition rate was stable across different eye conditions and different numbers of winks/blinks; (iii) the average time needed to perform one command in a free user session with four blinking schemes was equal to 2.04 s and the average accuracy was 95%. Conclusion: The comparison with other papers shows that the proposed algorithm is able to recognize: a higher number of control states (9 states) than reference algorithms (1–4 states) with a high detection rate and short calibration period (15–21 s). Significance: Comparing to nowadays Brain-Computer Interfaces (BCIs), the RVEB algorithm provides more comfortable communication (for people able to control their eyelid muscles) than BCIs based on steady state visually evoked potentials or P300 component and more control states than BCIs based on motor imagery.