Eye Movement Detection Based on SM-SSA and Quantum CNN from EMG of EOM Signals

Analysis and classification of electromyogram (EMG) signals are pivotal for developing assistive technologies. These signals are nonstationary in nature and require nonstationary signal processing methods for their analysis. In this study, the sliding mode singular spectrum analysis (SM-SSA) method is considered for analysis of EMG of extraocular muscles (EOM) signals. Furthermore, we propose a new framework combining SM-SSA and quantum convolutional neural network (QCNN) for the task of eye movement detection. The SM-SSA decomposes EMG of EOM signals into its constituent components from which features are extracted. The neighborhood component analysis is used to obtain the optimal set of features which are classified into different eye movement classes using QCNN classifier. The proposed framework achieved a classification accuracy of 98.70%, outperforming compared methods from literature.