Understanding Human Limb Movements Through EEG Signal Analysis

Abstract

This paper investigates the efficacy of EEG signal analysis in relation to human limb movements, utilizing support vector machines (SVM) and k-nearest neighbors (KNN) classifiers. A pivotal aspect of this analysis is the comparative assessment of frequency band performance and amplitude performance across three distinct features: the mean, standard deviation, and sum of discrete wavelet transforms. EEG signals are intrinsically linked to motor activities, exhibiting varied neural rhythms when individuals execute motor tasks. These neural patterns hold significance within brain-computer interface (BCI) frameworks. The paper meticulously examines and interprets various EEG signal configurations corresponding to human limb activities, aiming to delineate the spatial relationships inherent between EEG signals and specific limb movements. The classifiers are applied to EEG data collected during experimental trials that encompassed both active motor tasks and resting conditions. The analysis reveals that SVM, evaluated on both unsegmented and segmented data sets, significantly outperforms KNN, achieving a peak accuracy of 98.62 % across both data types. Notably, one key advantage of SVM in this context is its applicability for online systems, as it circumvents the need for computational processes post-training. Conversely, KNN's reliance on extensive computational resources for new data inputs renders it less optimal for realtime applications. Subsequent research efforts may concentrate on the development of a refined model aimed at classifying diverse limb motor activities utilizing EEG data, thereby propelling advancements in the brain-computer interface domain.

Topics & Concepts