Classification of finger pairs from one hand based on spectral features in human EEG

Individual finger movements are well-articulated movements of fine body parts, the successful decoding of which can provide extra degrees of freedom to drive brain computer interface (BCI) applications. Past studies present some unique features revealed from spectral principal component analysis (PCA) on electrophysiological data recorded in both the surface of the brain (electrocorticography, ECoG) and the scalp (electroencephalography, EEG). These features contain discriminable information about fine individual finger movements from one hand. However, the efficacy of these spectral features has not been well investigated under the application of various classifiers. In the present study, we set out to investigate the topic using noninvasive human EEG. Several classifiers were chosen to explore their capability in capturing the spectral PC features to decode individual finger movements pairwisely from one hand using noninvasive EEG, aiming to investigate the efficacy of these spectral features in a decoding task.