Motor imagery EEG signal classification using minimally random convolutional kernel transform and hybrid deep learning

The brain-computer interface (BCI) establishes a non-muscle channel that enables direct communication between the human body and an external device. Electroencephalography (EEG) is a popular non-invasive technique for recording brain signals. It is critical to process and comprehend the hidden patterns linked to a specific cognitive or motor task, for instance, measured through the motor imagery brain-computer interface (MI-BCI). A significant challenge is presented by classifying motor imagery-based electroencephalogram (MI-EEG) tasks, given that EEG signals exhibit nonstationarity, time-variance, and individual diversity. Achieving good classification accuracy is also challenging due to the increasing number of classes and the inherent variability among individuals. To overcome these issues, this paper proposes a novel method for classifying EEG motor imagery signals that efficiently extracts features using the Minimally Random Convolutional Kernel Transform (MiniRocket). A linear classifier then utilises the extracted features for activity recognition. Furthermore, a novel deep learning model based on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) architecture was proposed and demonstrated to serve as a baseline. The classification via MiniRocket's features achieved higher performance than the best deep learning models at a lower computational cost. PhysioNet and BCI Comp IV 2a datasets were used to evaluate the performance of the proposed approaches. Using PhysioNet, the proposed models achieved mean accuracy values of 98.63% and 98.06%, respectively, for the MiniRocket and CNN-LSTM. With the BCI-CompIV-2a dataset, proposed models achieved mean accuracy values of 92.57% and 92.32%, respectively. The findings demonstrate that the proposed approach can significantly enhance motor imagery EEG accuracy and provide new insights into the feature extraction and classification of MI-EEG. An additional future direction is non-additive electrode-source fusion (Choquet-integral/coalition formulations) to improve robustness under low-SNR EEG and inter-subject variability.