FBLSTM: A Filter-Bank LSTM-based deep learning method for MI-EEG classification

Yawei Gui; Ziwei Tian; Xi Liu; Bingliang Hu; Quan Wang

doi:10.1117/12.2673936

6 April 2023 FBLSTM: A Filter-Bank LSTM-based deep learning method for MI-EEG classification

Yawei Gui, Ziwei Tian, Xi Liu, Bingliang Hu, Quan Wang

Proceedings Volume 12615, International Conference on Signal Processing and Communication Technology (SPCT 2022); 126151W (2023) https://doi.org/10.1117/12.2673936
Event: International Conference on Signal Processing and Communication Technology (SPCT 2022), 2022, Harbin, China

Abstract

Motor imagery electroencephalogram (MI-EEG) classification is a vital task for brain computer interface (BCI) system. But the classification accuracy is not satisfactory, which hinders its generalizability. This study proposes a Filter-Bank Long Short-Term Memory (LSTM) Network (FBLSTM), which adopts a series of band-pass filters to obtain the different frequency information in EEG signals, and a convolution neural network to obtain spatial information. Moreover, a LSTM with attention mechanism is employed to process time series information. The open BCI Competition IV dataset 2a is applied to validate the performance of the proposed FBLSTM. Compared with recent methods, our method shows advantages on the within-subject and cross-subject 4-class classification performance and outperformed existing models, achieving an average accuracy of 72.4% and 53.6%, respectively.

Citation Download Citation

Yawei Gui, Ziwei Tian, Xi Liu, Bingliang Hu, and Quan Wang "FBLSTM: A Filter-Bank LSTM-based deep learning method for MI-EEG classification", Proc. SPIE 12615, International Conference on Signal Processing and Communication Technology (SPCT 2022), 126151W (6 April 2023); https://doi.org/10.1117/12.2673936

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available