Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network Design
Recently, robot-assisted rehabilitation has emerged as a promising solution to increase the training intensity of stroke patients while reducing workload on therapists, whilst surface electromyography (sEMG) is expected to serve as a viable control source. In this paper, we delve into the potential...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Transactions on Neural Systems and Rehabilitation Engineering |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10812756/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841557088632832000 |
|---|---|
| author | Tianzhe Bao Zhiyuan Lu Ping Zhou |
| author_facet | Tianzhe Bao Zhiyuan Lu Ping Zhou |
| author_sort | Tianzhe Bao |
| collection | DOAJ |
| description | Recently, robot-assisted rehabilitation has emerged as a promising solution to increase the training intensity of stroke patients while reducing workload on therapists, whilst surface electromyography (sEMG) is expected to serve as a viable control source. In this paper, we delve into the potential of deep learning (DL) for post-stroke hand gesture recognition by collecting the sEMG signals of eight chronic stroke subjects, focusing on three primary aspects: feature domains of sEMG (time, frequency, and wavelet), data structures (one or two-dimensional images), and neural network architectures (CNN, CNN-LSTM, and CNN-LSTM-Attention). A total of 18 DL models were comprehensively evaluated in both intra-subject testing and inter-subject transfer learning tasks, with two post-processing algorithms (Model Voting and Bayesian Fusion) analysed subsequently. Experiment results infer that for intra-subject testing, the average accuracy of CNN-LSTM using two-dimensional frequency features is the highest, reaching 72.95%. For inter-subject transfer learning, the average accuracy of CNN-LSTM-Attention using one-dimensional frequency features is the highest, reaching 68.38%. Through these two experiments, it was found that frequency features had significant advantages over other features in gesture recognition after stroke. Moreover, the post-processing algorithm can further improve the recognition accuracy, and the recognition effect can be increased by 2.03% through the model voting algorithm. |
| format | Article |
| id | doaj-art-76e5e6cecba44016b5705ed7c8cdfd8f |
| institution | Kabale University |
| issn | 1534-4320 1558-0210 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Transactions on Neural Systems and Rehabilitation Engineering |
| spelling | doaj-art-76e5e6cecba44016b5705ed7c8cdfd8f2025-01-07T00:00:13ZengIEEEIEEE Transactions on Neural Systems and Rehabilitation Engineering1534-43201558-02102025-01-013319120010.1109/TNSRE.2024.352158310812756Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network DesignTianzhe Bao0https://orcid.org/0000-0002-1103-2660Zhiyuan Lu1https://orcid.org/0009-0004-7651-6625Ping Zhou2https://orcid.org/0000-0002-4394-2677School of Rehabilitation Science and Engineering, University of Health and Rehabilitation Sciences, Qingdao, ChinaSchool of Rehabilitation Science and Engineering, University of Health and Rehabilitation Sciences, Qingdao, ChinaSchool of Rehabilitation Science and Engineering, University of Health and Rehabilitation Sciences, Qingdao, ChinaRecently, robot-assisted rehabilitation has emerged as a promising solution to increase the training intensity of stroke patients while reducing workload on therapists, whilst surface electromyography (sEMG) is expected to serve as a viable control source. In this paper, we delve into the potential of deep learning (DL) for post-stroke hand gesture recognition by collecting the sEMG signals of eight chronic stroke subjects, focusing on three primary aspects: feature domains of sEMG (time, frequency, and wavelet), data structures (one or two-dimensional images), and neural network architectures (CNN, CNN-LSTM, and CNN-LSTM-Attention). A total of 18 DL models were comprehensively evaluated in both intra-subject testing and inter-subject transfer learning tasks, with two post-processing algorithms (Model Voting and Bayesian Fusion) analysed subsequently. Experiment results infer that for intra-subject testing, the average accuracy of CNN-LSTM using two-dimensional frequency features is the highest, reaching 72.95%. For inter-subject transfer learning, the average accuracy of CNN-LSTM-Attention using one-dimensional frequency features is the highest, reaching 68.38%. Through these two experiments, it was found that frequency features had significant advantages over other features in gesture recognition after stroke. Moreover, the post-processing algorithm can further improve the recognition accuracy, and the recognition effect can be increased by 2.03% through the model voting algorithm.https://ieeexplore.ieee.org/document/10812756/Deep learninghand gesture recognitionpost-processingsurface electromyographystroke patients |
| spellingShingle | Tianzhe Bao Zhiyuan Lu Ping Zhou Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network Design IEEE Transactions on Neural Systems and Rehabilitation Engineering Deep learning hand gesture recognition post-processing surface electromyography stroke patients |
| title | Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network Design |
| title_full | Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network Design |
| title_fullStr | Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network Design |
| title_full_unstemmed | Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network Design |
| title_short | Deep Learning-Based Post-Stroke Myoelectric Gesture Recognition: From Feature Construction to Network Design |
| title_sort | deep learning based post stroke myoelectric gesture recognition from feature construction to network design |
| topic | Deep learning hand gesture recognition post-processing surface electromyography stroke patients |
| url | https://ieeexplore.ieee.org/document/10812756/ |
| work_keys_str_mv | AT tianzhebao deeplearningbasedpoststrokemyoelectricgesturerecognitionfromfeatureconstructiontonetworkdesign AT zhiyuanlu deeplearningbasedpoststrokemyoelectricgesturerecognitionfromfeatureconstructiontonetworkdesign AT pingzhou deeplearningbasedpoststrokemyoelectricgesturerecognitionfromfeatureconstructiontonetworkdesign |