Parametric Bispectral Estimation of EEG Signals in Disorders of Consciousness

Since rhythmic oscillations are a general phenomenon of neuronal activity, the use of EEG-based spectral analysis is one of the most important methods for studying how the human brain processes information. Usually, during the functional integration and isolation of information in the brain, it is often accompanied by simultaneous rhythmic oscillations of different frequencies. Therefore, studying the phase-synchronous coupling between different frequencies of neurons may have important physiological significance. In this paper, we apply high order bispectral (BIS) analysis to investigate non-linear phase coupling of different oscillators of patients with chronic disorders of consciousness (DOC). We found that non-linear phase coupling of different oscillators of patients in minimally conscious state (MCS) showed significantly different from those with vegetative state (VS). In order to quantify the group differences, three relevant parameters of bispectrum are introduced. Moreover, Bispectral entropy (BE) and the weighted center frequencies fl of the bispectrum significantly correlated with the level of behavioral responsiveness measured using the JFK Coma Recovery Scale-Revised (CRS-R). And classifier based on support vector machine (SVM) is trained to predict states of consciousness of patients with DOC. What's more, sharpening the current knowledge of BIS by using EEG-based approaches may shed light on the processes of consciousness and their pathophysiology, and may contribute to clinical treatment and assessment of patents with DOC.