EEG-Based Classification of Motor Imagery and Movement Across Brain Regions

The brain plays a crucial role in all actions, including motor imagery and movement. To better understand its function in these contexts, we utilize brain-computer interfaces. This study aims to enhance comprehension of brain activity during both imagined and actual motions, using this knowledge to inform neurorehabilitation strategies for individuals with movement impairments. An EEG dataset from 52 healthy participants performing motor imagery and hand movement tasks was analyzed. EEG signals were recorded using a 64-channel setup, with preprocessing involving noise filtering, artifact removal, and Principal Component Analysis (PCA) for dimensionality reduction. Features were extracted from temporal and spectral domains, and machine learning models, including Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Random Forest, were evaluated. The polynomial SVM yielded the highest accuracy (58.06%) and was used for further cortical region analysis. The parietal cortex demonstrated the best performance (58%), highlighting its role in sensorimotor integration. Although promising, further optimization of feature selection and model refinement is necessary. Future work should investigate region-specific improvements and applications in motor rehabilitation.