Emerging Neuro-Technologies for Detecting Consciousness in Locked-In Syndrome: A Review of fMRI, EEG, and Machine Learning Applications

Abstract

Introduction: The clinical differentiation between disorders of consciousness (DOC), such as unresponsive wakefulness syndrome (UWS), and locked-in syndrome (LIS/CLIS) remains a formidable challenge, with misdiagnosis rates reaching up to 40%. Because traditional bedside assessments are motor-dependent, they fail to detect "covert consciousness" or "cognitive motor dissociation" (CMD) in patients who are cognitively aware but physically paralyzed. This review analyzes evidence on emerging neurotechnologies, specifically functional magnetic resonance imaging (fMRI), electroencephalogram (EEG), and machine learning (ML)—aimed to establish objective, motor-independent markers of awareness.

Methods: PubMed, Google Scholar and Scopus were searched for articles that focus on locked-in syndrome and the neurotechnologies that can detect consciousness, using keywords such as “Locked-in syndrome”, “electroencephalography”, “fMRI”. Analysis was done from 22 studies, focusing on different technologies and different subgroups of unconsciousness.

Results: Out of 22 studies, transcranial magnetic stimulation-electro-EEG-derived perturbational complexity index reliably discriminates conscious from unconscious states with high sensitivity, providing a "gold standard" for bedside complexity mapping. Task-based fMRI and EEG detected command-following in 15% of behaviorally unresponsive acute patients, which serves as a powerful predictor of 12-month functional recovery. In chronic cases, resting-state fMRI demonstrated that LIS patients maintain higher default mode network (DMN) connectivity than those in UWS. Machine learning models successfully automated the detection of awareness by training on pharmacological datasets and applying them to pathological states. For the transition to completely locked-in state (CLIS), metabolic signals from functional near-infrared spectroscopy (fNIRS) and invasive Brain-Computer Interfaces (BCI) using ECoG or LFP signals provided stable, long-term communication channels where traditional EEG paradigms often failed. Furthermore, hybrid systems integrating EEG with eye-tracking significantly improved communication accuracy in fluctuating states.

Conclusion: Emerging neurotechnologies have transformed the diagnostic landscape for locked-in syndrome, shifting reliance from overt behavior to internal neural dynamics. A multimodal diagnostic framework, combining the spatial resolution of fMRI, the temporal flexibility of EEG, and the predictive power of machine learning, is essential for accurate stratification. These advancements not only reduce misdiagnosis but also provide the technical foundation for permanent, high-speed communication interfaces for the most severely disabled patients.