Linear Discriminant Analysis Classifies the EEG Spectral Features Obtained from Three Class Motor Imagination

Abstract

This paper explores the changes in power associated with the imagination of motor imagery (MI) signals corresponding to hand movement and foot movement. The study presented in this paper establishes a correlation between MI signals and Event-RelatedSynchronization (ERS)and Event-Related Desynchronization (ERD) phenomenon with the help of Band power features. This fact has been further validated with the help of Linear Discriminant Analysis (LDA) Classifier. The analysis is conducted on dataset IIIa, obtained from BCI Competition III conducted by Graz University, Austria. The dataset is a multiclass data recorded for three subjects. The accuracy of individual classes and inter-class is calculated. The average accuracy for subject K3b, K6b and L1b obtained is 91.92%, 59.97%, 57.89% respectively when training data is used as test data and when a set of test data is used the resultant accuracy obtained is 85.66%, 65.33%, 58% for the three subjects respectively. Foot data showed very less accuracy in comparison with the data of hands. It is analyzed through the studies conducted in this paper that simple phenomenon of ERD/ERS can be utilized in a multiclass Brain-Computer Interface (BCI) system operated for MI.