Features Selection for EEG During Mental Arithmetic Task – Brain-Computer Interface

Abstract

Our objective is to find an efficient technique for obtaining features that can describe the continuous and underlying temporal dynamics of electroencephalography (EEG) data during the performance of mental tasks. The statistical technique known as the paired t-test is proposed to evaluate each feature. From the alpha, beta, and gamma bands of EEG, the features like mean, root mean square, skewness, mode, data range, interquartile range (IQR), and three Hjorth parameters are extracted to differentiate a signal before-during mental arithmetic task. The results suggest how essential the feature selection procedure is for the recognition task. Hjorth parameters seen between datasets during and before the arithmetic exercise have shown a significant statistical difference (p0.05). Studying an efficient way to distinguish mental task performance based on the extracted characteristics is of tremendous relevance and practical value because they accurately capture the continuity and internal dynamic changes of the EEG signals.