Filter bank sinc-convolutional network with channel self-attention for high performance motor imagery decoding.
Uloženo v:
| Název: | Filter bank sinc-convolutional network with channel self-attention for high performance motor imagery decoding. |
|---|---|
| Autoři: | Chen J; Faculty of Information Technology, Beijing University of Technology, Beijing, People's Republic of China., Wang D; Faculty of Information Technology, Beijing University of Technology, Beijing, People's Republic of China., Yi W; Beijing Machine and Equipment Institute, Beijing, People's Republic of China., Xu M; Faculty of Information Technology, Beijing University of Technology, Beijing, People's Republic of China., Tan X; Faculty of Information Technology, Beijing University of Technology, Beijing, People's Republic of China. |
| Zdroj: | Journal of neural engineering [J Neural Eng] 2023 Mar 03; Vol. 20 (2). Date of Electronic Publication: 2023 Mar 03. |
| Způsob vydávání: | Journal Article; Research Support, Non-U.S. Gov't |
| Jazyk: | English |
| Informace o časopise: | Publisher: Institute of Physics Pub Country of Publication: England NLM ID: 101217933 Publication Model: Electronic Cited Medium: Internet ISSN: 1741-2552 (Electronic) Linking ISSN: 17412552 NLM ISO Abbreviation: J Neural Eng Subsets: MEDLINE |
| Imprint Name(s): | Original Publication: Bristol, U.K. : Institute of Physics Pub., 2004- |
| Výrazy ze slovníku MeSH: | Imagination* , Brain-Computer Interfaces*, Imagery, Psychotherapy ; Electroencephalography/methods ; Intention ; Algorithms |
| Abstrakt: | Objective. Motor Imagery Brain-Computer Interface (MI-BCI) is an active Brain-Computer Interface (BCI) paradigm focusing on the identification of motor intention, which is one of the most important non-invasive BCI paradigms. In MI-BCI studies, deep learning-based methods (especially lightweight networks) have attracted more attention in recent years, but the decoding performance still needs further improving. Approach. To solve this problem, we designed a filter bank structure with sinc-convolutional layers for spatio-temporal feature extraction of MI-electroencephalography in four motor rhythms. The Channel Self-Attention method was introduced for feature selection based on both global and local information, so as to build a model called Filter Bank Sinc-convolutional Network with Channel Self-Attention for high performance MI-decoding. Also, we proposed a data augmentation method based on multivariate empirical mode decomposition to improve the generalization capability of the model. Main results. We performed an intra-subject evaluation experiment on unseen data of three open MI datasets. The proposed method achieved mean accuracy of 78.20% (4-class scenario) on BCI Competition IV IIa, 87.34% (2-class scenario) on BCI Competition IV IIb, and 72.03% (2-class scenario) on Open Brain Machine Interface (OpenBMI) dataset, which are significantly higher than those of compared deep learning-based methods by at least 3.05% (p= 0.0469), 3.18% (p= 0.0371), and 2.27% (p= 0.0024) respectively. Significance. This work provides a new option for deep learning-based MI decoding, which can be employed for building BCI systems for motor rehabilitation. (© 2023 IOP Publishing Ltd.) |
| Contributed Indexing: | Keywords: brain–computer interface; data augmentation; deep learning; motor imagery; self-attention |
| Substance Nomenclature: | 92396-88-8 (bis(tri-n-hexylsiloxy)(2,3-naphthalocyaninato)silicon) |
| Entry Date(s): | Date Created: 20230210 Date Completed: 20230307 Latest Revision: 20230310 |
| Update Code: | 20250114 |
| DOI: | 10.1088/1741-2552/acbb2c |
| PMID: | 36763992 |
| Databáze: | MEDLINE |
Nájsť tento článok vo Web of Science | Abstrakt: | Objective. Motor Imagery Brain-Computer Interface (MI-BCI) is an active Brain-Computer Interface (BCI) paradigm focusing on the identification of motor intention, which is one of the most important non-invasive BCI paradigms. In MI-BCI studies, deep learning-based methods (especially lightweight networks) have attracted more attention in recent years, but the decoding performance still needs further improving. Approach. To solve this problem, we designed a filter bank structure with sinc-convolutional layers for spatio-temporal feature extraction of MI-electroencephalography in four motor rhythms. The Channel Self-Attention method was introduced for feature selection based on both global and local information, so as to build a model called Filter Bank Sinc-convolutional Network with Channel Self-Attention for high performance MI-decoding. Also, we proposed a data augmentation method based on multivariate empirical mode decomposition to improve the generalization capability of the model. Main results. We performed an intra-subject evaluation experiment on unseen data of three open MI datasets. The proposed method achieved mean accuracy of 78.20% (4-class scenario) on BCI Competition IV IIa, 87.34% (2-class scenario) on BCI Competition IV IIb, and 72.03% (2-class scenario) on Open Brain Machine Interface (OpenBMI) dataset, which are significantly higher than those of compared deep learning-based methods by at least 3.05% (p= 0.0469), 3.18% (p= 0.0371), and 2.27% (p= 0.0024) respectively. Significance. This work provides a new option for deep learning-based MI decoding, which can be employed for building BCI systems for motor rehabilitation.<br /> (© 2023 IOP Publishing Ltd.) |
|---|---|
| ISSN: | 1741-2552 |
| DOI: | 10.1088/1741-2552/acbb2c |