Detection of Eyeblink using Hybrid Brain-Computer Interface

Authors

  • S. Sridhar , U. Ramachandraiah , G. Muthukumaran , E. Sathish

Abstract

Quadriplegia brings impairment to the physical functions of a person. Electroencephalography
(EEG) signals using Brain-Computer Interface (BCI) can be implemented to make them partially
independent. EEG is a technique used to monitor, analyse and interpret the electrophysiological activities of
the brain, thereby activating the assistive devices to help them with their brain intents. In this paper, the
study is conducted on a BCI using eye blink signals. The presence and absence of eyeblink in EEG signal
are detected by employing the Linear Discriminant Analysis (LDA). Simultaneously, the eyeblink event is
also recorded using a web camera and the eyeblinks are detected using Convolutional Neural Network
(CNN) algorithm. The final decision of the eyeblink is determined by correlating both the outputs from LDA
and CNN. This methodology analysed on healthy persons, it can be fine-tuned for the patients of a
quadriplegic, amyotrophic lateral sclerosis and spinal cord injury. The classification accuracy of LDA and
CNN is 98.88% and 95% respectively.

Published

2020-01-31

Issue

Section

Articles