Feature study of hysterical blindness EEG based on FastICA with combined-channel information.

BACKGROUND An appropriate feature study of hysteria electroencephalograms (EEG) would provide new insights into neural mechanisms of the disease, and also make improvements in patient diagnosis and management. OBJECTIVE The objective of this paper is to provide an explanation for what causes a particular visual loss, by associating the features of hysterical blindness EEG with brain function. METHODS An idea for the novel feature extraction for hysterical blindness EEG, utilizing combined-channel information, was applied in this paper. After channels had been combined, the sliding-window-FastICA was applied to process the combined normal EEG and hysteria EEG, respectively. Kurtosis features were calculated from the processed signals. As the comparison feature, the power spectral density of normal and hysteria EEG were computed. RESULTS According to the feature analysis results, a region of brain dysfunction was located at the occipital lobe, O1 and O2. Furthermore, new abnormality was found at the parietal lobe, C3, C4, P3, and P4, that provided us with a new perspective for understanding hysterical blindness. CONCLUSIONS Indicated by the kurtosis results which were consistent with brain function and the clinical diagnosis, our method was found to be a useful tool to capture features in hysterical blindness EEG.