Classification of EEG signals related to real and imagery knee movements using deep learning for brain computer interfaces.

This article focuses on comparing the classification accuracy of motor imagery (MI) and movement execution (ME) of knee movements using electroencephalogram (EEG)-based non-invasive Brain-Computer Interface (BCI) technology. Data from ten healthy subjects performing and imagining four types of knee exercises were collected and classified using a modified Lenet-5 convolutional neural network (CNN) model combined with support vector machine (SVM). Results showed that ME data achieved significantly higher classification accuracy (98.91%) than MI data (98.37%), suggesting ME provides more reliable signals for BCI applications involving lower limb movements. The study highlights the potential for improving BCI accuracy in recognizing movement intentions, which could benefit patient communication, rehabilitation, and assistive robotic control, while recommending further research on other body movements to enhance BCI development.