Related Experiment Video
Updated: Oct 18, 2025
09:42
Author Spotlight: Using Motor Imagery Brain-Computer Interface to Improve Motor and Cognitive Function in Stroke Patients
Published on: September 1, 2023
1.6K
A transfer learning framework based on motor imagery rehabilitation for stroke
Fangzhou Xu1, Yunjing Miao2, Yanan Sun2
1School of Electronic and Information Engineering (Department of Physics), Qilu University of Technology (Shandong Academy of Sciences), Jinan, 250353, China. xfz@qlu.edu.cn.
Scientific Reports
|October 6, 2021
Summary
This study enhances stroke rehabilitation using deep learning for brain-computer interfaces (BCI). The EEGNet model with transfer learning achieved 66.36% accuracy in motor imagery recognition, reducing training time and complexity.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Deep learning models can adapt to new domains.
- Electroencephalogram (EEG) signals are used in brain-computer interface (BCI) systems.
- Stroke rehabilitation can benefit from BCI technology.
Purpose of the Study:
- To design an effective motor imagery (MI) BCI system for stroke patients using deep learning.
- To adapt deep learning models from source to target domains for EEG signal processing.
- To improve the efficiency and reduce the training time of BCI systems through transfer learning.
Main Methods:
- Utilized multiple convolutional neural networks (CNNs) to decode EEG signals.
- Implemented a 'fine-tune' approach for transferring model parameters.
- Evaluated model performance on two-class MI recognition tasks.
Main Results:
- The combination of EEGNet and 'fine-tune' transfer learning yielded the best performance.
- Achieved an average classification accuracy of 66.36% across 11 subjects.
- Demonstrated significantly lower algorithm complexity compared to other models.
Conclusions:
- The EEGNet model shows significant potential for MI-based stroke rehabilitation BCI systems.
- Transfer learning effectively improves the performance of EEG-based stroke rehabilitation BCI.
- The proposed framework offers an efficient solution for stroke rehabilitation.
