Novel methods for the detection of consciousness in severely brain-injured subjects

People who suffer severe brain injuries leading to disorders of consciousness often have significant motor impairment that confounds the measurement of residual cognitive function at the bedside. In order to avoid this confound, recent work has focused on developing mental imagery paradigms for command following and communication. Subjects have been asked to imagine they are playing tennis or walking around their house in response to a verbal command during functional imaging, or to use those same mental imagery responses to respond to questions. While such studies have revealed dissociations between behavior and cognitive abilities, it remains possible that many subjects who follow commands or communicate at the bedside will not successfully perform such tasks.;To address this question, we studied seven subjects using a hierarchical functional magnetic resonance (fMRI) imaging assessment utilizing mental imagery responses. The subjects showed a wide variation in the correlation of blood-oxygen-level-dependent (BOLD) responses and overt behavioral responses. Specifically, our analysis demonstrated a common dissociation between bedside and fMRI-based command following and communication capabilities. These observations reveal significant challenges in developing validated fMRI-based methods for clinical use.;Given these limitations, we sought to utilize a passive approach to study the brain activity of brain-injured subjects in response to salient stimuli. We studied subjects with EEG while they listened to naturalistic narratives lasting 2-3 minutes, as well as time-reversed versions of those narratives. We identified changes in brain activity that were selective for speech stimuli by comparing the forward and time-reversed speech presentations. All control and MCS subjects showed differences between conditions. Across subjects in both groups, these differences were in similar frequencies bands, and were accompanied by increases in coherence between frontal and temporal areas, suggesting a common pattern of neuronal response. VS subjects showed no change to presentation of the narrative. These data indicate that naturalistic language content leads to reliable changes in the EEG of control and MCS subjects. With further validation, our methods may provide a basis for clinical assessments of diagnosis and prognosis in severe brain injury.