Brain-Computer Interfacing and Classification of Cognitive Activities

Authors

  • Hussein Al-Huraibi Department of Electrical and Computer Engineering, Faculty of Engineering, King Abdulaziz University, Jeddah, Saudi Arabia
  • K. Prahlad Rao Department of Electrical and Computer Engineering, Faculty of Engineering, King Abdulaziz University, Jeddah, Saudi Arabia

Keywords:

EEG, fNIRS, Brain Computer Interface, SVM and Classification

Abstract

Human intellect can be straightforwardly associated with computers through a modern innovation known as Brain-Computer Interface (BCI). Electroencephalography (EEG) and functional Near Infrared Spectroscopy (fNIRS) based BCI empowers to associate the individuals with the encompassing world through brain signals noninvasively. This strategy of perusing the intellect through physiological signals by EEG and fNIRS sensors has made critical advances in neurological science and engine control inquire about. The BCI framework can record, analyze, and decipher the framework input, procured from the brain in terms of commands. These commands can assist be utilized to activate outside gadgets of choice concurring to the user’s intellect. The BCI is rising as one of the capable instruments in reasonable biomedical applications such as recovery, cognitive forms, prosthetics, and numerous neuro-feedback utilitarian exercises. Be that as it may, the usefulness of BCI depends upon the acknowledgment and classification of brain signals for segregating errand and resting exercises of the brain. We have developed two algorithms for assessment and classification of EEG and fNIRS alone and combined as hybrid (EEG+fNIRS) signals to recognize brain activities under the given tasks. We have tested our classifiers from the open-source EEG-fNIRS dataset. The dataset is consisting of EEG and fNIRS simultaneously recordings acquired from twenty-six healthy participants during word generation (WG) tasks. In this work, we have achieved an average classification accuracy peak of 85 %, 84 %, and 78 % Hybrid, EEG, and fNIRS respectively for SVM with the dataset.

Published

2020-09-03

Issue

Section

Articles