Improving Motor Imagery Classification in BCI Systems Using EMD and Multi-Layer CNNs

Brain-Computer Interface (BCI) systems are relatively new technologies that could play a significant role in aiding the recovery of impaired activities resulting from neuromuscular disabilities in affected individuals. Accurate recognition and classification of motor imagery in BCI systems present a challenge, leading to extensive research in recent years aimed at improving the accuracy of these systems. In this study, a combination of the Empirical Mode Decomposition (EMD) method and a multi-layer Convolutional Neural Network (CNN) is employed. Initially, the signal is decomposed into Intrinsic Mode Functions (IMFs) using EMD, and all IMFs across all trials are analyzed to select the best ones, which are then fed into the CNN as inputs. Additionally, the study incorporates the fusion of three CNN networks, each corresponding to a different IMF. The features extracted from these networks are combined and used to train an SVM classifier. The proposed method achieved an accuracy of 86.1% on the BCI-2a 2008 dataset, outperforming other state-of-the-art approaches.