Impact of the preprocessing block on the performance of the ΒCI system

Laboratory of Metrology and Information Processing, Faculty of Sciences, Ibn Zohr University, Agadir, Morocco

^* Corresponding author: soukaina.hamou@edu.uiz.ac.ma

Abstract

Electroencephalography (ΕΕG) is considered as one of the famous and efficient used methods in the Brain Computer Interface (ΒCI). This is due to its simplicity for implementation, low cost and being portable. The ΕΕG is a technique that examines the electrical activity of the brain using a non-invasive electrodes placed on the scalp. ΕΕG-based BCI system is constituted of five blocks: signal acquisition, preprocessing, feature extraction, classification and command block. In this paper, we will study the impact of the filter type and its order on the performance of the considered BCI system. This system is composed of: bandpass (ΒΡ) filter for the preprocessing step, Common Spatial Pattern (CSP) in the feature extraction block, and for the classification block, we used Support Vector Machine (SVΜ). The obtained results show a good improvement of the proposed BCI system. Indeed, the accuracy of this system can achieve 88.17% and the kappa coefficient is almost 0.76.