Hybrid CNN-LSTM Model for EEG-Based Motor Imagery Classification

BCIs offer a direct communication channel between the external device and the brain. The study targets EEG signal-based MI classification with a hybrid model that utilizes CNN-LSTM in spatial and temporal correlation recording. The hybrid model was trained on the BCI Competition IV 2a dataset, including four-class MI tasks of nine subjects. Preprocessing involves normalization, segmentation, and artifact removal.Execute network is efficient with high accuracy and resilience and effectively exploits inter-subject variability and class imbalance with high efficiency. Its performance is confirmed using metrics like precision, recall, and F1-score.The present work depicts the potential of hybrid deep learning in improving BCI accuracy for paving the way for improving assistive technology and neurorehabilitation