Classification of Motor Imagery Signals Using Neural Networks for Applications in Brain-Computer Interface

This study enhances Brain-Computer Interface (BCI) systems by analyzing Electroencephalogram signals related to motor imagery (MI). It focuses on accurate signal processing for reliable medical BCI applications, using specific statistical features from filtered Power Spectral Density (PSD) data. These features are crucial for classification and understanding brain activity during motor tasks. A sequential neural network model is employed, achieving 90.55% accuracy in distinguishing between left- and right-hand MI tasks using the BCI competition IV dataset 1. The research not only improves classification accuracy but also highlights the importance of suitable feature extraction. This precise interpretation of motor imagery signals has significant implications for applications such as gaming, neuro-prosthetics, and neuro-rehabilitation.