Development of an EEG-based Motor Imagery Brain-Computer Interface System for Lower Limb Assistive Technologies

Abstract

Recent studies proved that advancements in Brain-Computer Interface (BCI) have played a vital role for the rehabilitation of people suffering from motor neuron diseases, which affect either the upper limb, lower limb, or both. Despite the fact that significant progress has been done for the field of BCI systems relating to upper-limb impairment, studies supporting BCI for the lower limb are still insufficient. This study aims to develop an Electroencephalogram (EEG)-based Motor Imagery BCI system for lower limb movements usually used for everyday actions (i.e. walking up and down the stairs, and their sub-phases such as stance and swing) which will support the motor-impaired to be able to conduct daily life activities. In line with this, the brain activities were measured via EEG-based BCI system, using a 14-channel EMOTIV EPOC + device which provided raw signals. The acquired signals were then preprocessed using a built-in denoising filter and the stance and swing gait phases were segmented with the aid of a gyroscope. Each segment underwent feature extraction using Discrete Wavelet Transform (DWT). The interclass correlations of the extracted features were determined. Using the Artificial Neural Network (ANN) as the classification technique with a 10fold cross validation, an average accuracy of 94.59% was achieved.