Loading...

Investigation and Evaluation of Rehabilitation by Lowerlimb Exoskeleton Based on Patient Intention Detection

Arjmand Mazidi, Alireza | 2025

45 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58576 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Behzadipour, Saeed
  7. Abstract:
  8. Rehabilitation of individuals with gait impairments—especially those caused by spinal cord injury or accidents—requires timely and effective interventions. Lower-limb exoskeletons, as advanced rehabilitation tools, enable precise, safe, and repeatable movement execution, thereby enhancing training efficiency and accelerating motor recovery. Mental engagement during such training is known to promote neuroplasticity and improve functional outcomes. This study investigates the integration of a brain–computer interface (BCI) based on electroencephalographic (EEG) signals with a lower-limb exoskeleton to control movement based on the user's motor intention. The main objective is to evaluate the feasibility of inducing neuroplasticity through rehabilitation exercises driven by this hybrid system. To evaluate the proposed system, 10 healthy male participants aged 24–30 years, with no prior experience using BCI systems, took part in the study. EEG signals were recorded using a four-channel Muse headband with a sampling rate of 256 Hz across fifteen sessions, each consisting of a 400-second motor imagery task. After preprocessing, 196 time- and frequency-domain features were extracted and reduced to the top 20 features using recursive feature elimination. A three-class support vector machine (SVM) model (rest, left leg, right leg) was trained, achieving a maximum classification latency of 200~ms (typically below 100~ms), making it suitable for real-time use in exoskeleton control. Participants were divided into two groups: “with-exoskeleton” and “without-exoskeleton”. Both groups underwent 15 one-hour motor imagery training sessions with real-time feedback. The first group received visual, auditory, and physical feedback, while the second group received only visual and auditory feedback. A dedicated training protocol and graphical user interface (GUI) were designed for each group, and all participants provided informed consent under ethical approval. Results showed that the difference between training and test accuracies for all participants was below 5%, indicating no overfitting. Participants S3, S8, and S2 achieved the highest test accuracies, around 50%, and the overall classification accuracy across all subjects ranged between 44% and 50%. After completing the training protocol, all participants in the exoskeleton group exhibited consistent improvement in classification accuracy and task performance, while those in the non-exoskeleton group showed mixed trends—some required more sessions to achieve progress. The observed rehabilitation improvement ranged from 5.53% to 62.83% across participants. Furthermore, time–frequency analysis revealed post-training changes in brain activity patterns in both groups, suggesting that neuroplasticity may have occurred as a result of the BCI-guided rehabilitation training
  9. Keywords:
  10. Rehabilitation Robots ; Brain-Computer Interface (BCI) ; Lower Limb Exoskeleton ; Electroencphalogram Signal ; Neuroplasticity ; Neural Recovery

 Digital Object List

  • محتواي کتاب
  •   view
  •   download

 Bookmark

No TOC