Journal of Advanced Clinical Neurology Research

Abstract

Spinal cord injury (SCI) causes severe motor impairments that significantly reduce patient independence, but cortical networks often remain intact. Non-invasive brain-computer interfaces (BCIs) hold promise for restoring motor control and facilitating rehabilitation after SCI. We conducted a systematic literature review of 100 recent studies on non-invasive BCIs for SCI motor recovery. Our analysis revealed that EEG-based motor-imagery (MI) BCIs paired with functional electrical stimulation (FES) were the predominant approach. These systems often incorporated rich multimodal feedback: many protocols combined visual cues, tactile sensations, and robotic assistance to reinforce the intended movement. We found that providing high density EEG recordings and personalized classifier calibration markedly improved decoding accuracy and clinical outcomes.Key implementation challenges included unstable FES electrode interfaces, user fatigue during extended training, and high system costs. Additionally, most studies tested only small patient cohorts, making it difficult to generalize results; patients with complete neural degeneration cannot benefit from conventional EEG-BCIs, indicating a need for alternative strategies. We highlight advanced adaptive techniques such as deep-learning decoders and transfer learning that have shown promise in recent studies. Overall, aligning neural intent detection with timely stimulation or feedback appears critical for driving neuroplasticity and enhancing motor recovery.