Problem
Research questions and friction points this paper is trying to address.
EEG
gait decoding
motion artifacts
signal-to-noise ratio
cortical gait complexity
A 2-Block Architecture for Real-Time EEG Gait Decoding: A Pilot Study
🤖 AI Summary
Current EEG-based closed-loop control of lower-limb exoskeletons is hindered by motion artifacts, low signal-to-noise ratios, and reliance on binary gait states, which fail to capture the dynamic complexity of cortical gait regulation. To address these limitations, this work proposes a novel two-module real-time brain–computer interface architecture: the first module performs session-adaptive artifact suppression and multi-domain feature extraction, while the second introduces a trainable polynomial time-varying layer (PolyTVL) jointly with an LSTM to decode four distinct gait states—standing, initiation, execution, and termination. This system constitutes the first closed-loop framework enabling real-time four-state EEG decoding, achieving a Matthews correlation coefficient (MCC) of 0.435 on the validation set—significantly outperforming baseline methods. In closed-loop experiments, it attained success rates of 52.7% and 55.3% for self-initiated and assisted gait initiation, respectively, with an average prediction latency of only 70.5 ms, demonstrating high accuracy and real-time feasibility.
📝 Abstract
Closed-loop lower-limb exoskeleton control via Electroencephalography (EEG) remains limited by motion artifacts, low signal-to-noise ratio, and binary gait formulations that fail to capture full cortical gait complexity. We propose a 2-block Brain-Computer Interface (BCI) architecture: a trainable session-specific Feature Extraction Block with real-time artifact suppression and multi-domain feature extraction, coupled with a Decoder Block built on a novel Polynomial Time-Varying Layer (PolyTVL)+LSTM for four-state gait classification (Stand, Initiate, Execute, Terminate). Ablation confirmed v01 (PolyTVL+LSTM) outperformed all variants (validation MCC: 0.435, gap: 0.187), with consistent EEG feature discriminability across ROIs and sub-bands (p<0.05). Closed-loop deployment with v01 achieved 55.3% (Rex-assisted) and 52.7% (volitional) gait initiation success, with a mean prediction time of 70.5~ms (+/-41.5), validating real-time feasibility in this pilot study.
Innovation
Methods, ideas, or system contributions that make the work stand out.
2-block BCI architecture
Polynomial Time-Varying Layer
real-time EEG decoding
gait state classification
artifact suppression
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
