Subject-Specific Classification of the Lower Limb Motor Imagery Using Emotive EEG Device

Brain–Computer Interfaces (BCIs) establish direct communication between the brain and external devices, offering potential benefits for individuals with severe motor impairments by interpreting neural signals and bypassing neuromuscular pathways. Motor Imagery (MI)-based BCIs rely on mentally simulating movements, enabling users to control external devices without actual limb motion. Although high-density EEG systems can improve MI-BCI accuracy, their complexity and preparation time limit widespread use in rehabilitation and everyday life. This study addresses these challenges by employing the Muse EEG headband, a low-density, user-friendly device with four dry electrodes, to classify lower limb motor imagery with minimal technical complexity. The targeted movements are flexion-extension in the knee joints of the right and left legs. Following a standardized experimental protocol, EEG signals were recorded during motor imagery tasks, then preprocessed and segmented. Feature extraction was performed using various analyses, followed by feature selection and dimensionality reduction to optimize classification performance. Multiple classifiers were employed to distinguish between MI and rest states, as well as between right and left knee MIs. The results demonstrated an intra-subject classification accuracy of ۸۵.۳۱% for MI versus rest and ۸۸.۵۷% for right versus left knee MIs.

کلیدواژه ها:

Electroencephalography (EEG) ، Brain-Computer Interface (BCI) ، Motor Imagery (MI) ، Movement Intention ، Lower Limb ، Classification