Leveraging Exoskeleton-Induced Finger Proprioception from EEG for Enhanced Motor Intention Decoding

Qiang Sun, E. Calvo, Marc M. Van Hulle2025 · DOI

Abstract

Repetitive training with hand exoskeletons can improve rehabilitation outcomes in individuals with hand motor impairments. Proprioceptive sensations induced by exoskeleton-assisted movements can be noninvasively captured via electroencephalography (EEG), often reflected as pronounced sensorimotor power desynchronization. However, it remains unclear whether fingerspecific proprioception can be differentiated from EEG signals and whether this information can enhance motor intention decoding. To investigate this, we recorded EEG from healthy participants (n = 25) performing finger tasks under three movement modalities: imagined movement (IM), passive movement (PM, performed by the R-TOUCH ® hand exoskeleton), and congruent imagined and passive movement (IPM). The tasks involved flexion of the thumb, index finger, and a group of three fingers (middle-ring-little, MRL). Across all modalities, finger movements elicited similar time, frequency, and spatial domain EEG patterns, with stronger responses observed during PM and IPM. Finger movement decoding was significantly more accurate for PM and IPM compared to IM, with the IPM obtaining the highest accuracies of 0.7364 ± 0.0755 for binary (Thumb vs. Index) and 0.5197 ± 0.0698 for ternary (Thumb vs. Index vs. MRL) tasks. While models trained or fine-tuned on passive movement data of a single subject successfully generalized to the same subject's imagined movement data, they faced challenges due to distributional differences-particularly affecting the performance of neural network-based models. These findings demonstrate that proprioceptive feedback from finger movements can be decoded from EEG signals, offering valuable insights for the development of precise brain-controlled hand exoskeletons-a promising technology in neurorehabilitation.
BibTeX
@misc{sun2025leveraging,
  author = {Qiang Sun and E. Calvo and Marc M. Van Hulle},
  title = {{Leveraging Exoskeleton-Induced Finger Proprioception from EEG for Enhanced Motor Intention Decoding}},
  year = {2025},
  doi = {10.36227/techrxiv.174702151.11095405/v1},
}

Loading related content...

Figure