Abstract
INTRODUCTION: Disorders of consciousness (DoCs) are characterized by varying levels of arousal and awareness. Due to severe motor impairments often accompanying these conditions, differentiating between minimally conscious state (MCS) and unresponsive wakefulness syndrome (UWS) remains challenging. Accurate diagnosis, however, is critical for optimal treatment and prognosis. Functional near-infrared spectroscopy (fNIRS), due to its portability and noninvasiveness, holds promise for brain-based diagnostics, although current methods lack sufficient sensitivity and specificity. METHODS: We introduce a highly sensitive, and specific fNIRS-based diagnostic approach tailored to the individual cognitive state of DoC patients. Nine healthy participants received auditory cues instructing them to either perform individually tailored motor-speech imagery or remain at rest. In separate runs, participants were instructed to be responsive to these cues to mimic MCS, or unresponsive, to mimic UWS. fNIRS-channel covariance matrices were classified for responsive and unresponsive states as either imagery or rest using a Riemannian-geometry-based approach. RESULTS: Classification between responsive and unresponsive states achieved a sensitivity of 100% and a specificity of 89% across participants. Within the two states, imagery and rest were classified with 83,9% and 55,93% accuracy, respectively; the latter result, close to chance level, was expected in the unresponsive state. CONCLUSION: This individualized diagnostic approach may have the potential to significantly enhance diagnostic accuracy for DoCs. It provides a noninvasive, efficient, and objective assessment, potentially reducing the rate of misdiagnosis rates. The practicality and minimal technical requirements of fNIRS further support future clinical implementation.