Dipartimento di Ingegneria informatica, automatica e gestionale

Severe acquired brain injury often leads to a disorder of consciousness (DoC) which can be classified as vegetative state (VS) or minimally consciousness state (MCS) according to its severity. The golden standard for DoC diagnosis is currently represented by the standardized Coma Recovery Scale Revised which, due to fluctuations in DoCs’ vigilance, can lead up to 40% of misdiagnosis. EEG-based analyses have demonstrated promising results in DoC patients’ assessment and quantitative EEG measures are candidates as a reliable instrument to support clinical diagnosis with high accuracy. Methodologies for spectral analysis and connectivity estimation were applied to EEG resting state data from 58 DoC patients clinically assessed as VS or MCS. Indices describing the global properties of the resting networks and the characteristics of power spectra resulted significantly different between the two groups of patients. The significant indices were used as features to train a classifier able to discriminate VS from MCS with performance metrics equal to 67% for accuracy and 85% for precision. These findings boost the promising role of EEG-based indices as valuable and reliable tool to support DoC clinical diagnosis.