Classification of left and right-hand motor imagery in acute stroke patients using EEG microstate

Abstract Background Stroke is one of the leading causes of adult disability, often resulting in motor dysfunction and brain network reorganization. Brain-computer interface (BCI) systems offer a novel approach to post-stroke motor rehabilitation, with motor imagery (MI) serving as a key paradigm tha...

Full description

Saved in:
Bibliographic Details
Main Authors: Shiyang Lv, Xiangying Ran, Mengsheng Xia, Yehong Zhang, Ting Pang, Xuezhi Zhou, Zongya Zhao, Yi Yu, Zhixian Gao
Format: Article
Language:English
Published: BMC 2025-06-01
Series:Journal of NeuroEngineering and Rehabilitation
Subjects:
Online Access:https://doi.org/10.1186/s12984-025-01668-y
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Background Stroke is one of the leading causes of adult disability, often resulting in motor dysfunction and brain network reorganization. Brain-computer interface (BCI) systems offer a novel approach to post-stroke motor rehabilitation, with motor imagery (MI) serving as a key paradigm that requires decoding left and right-hand MI differences to optimize system performance. However, the neural dynamics underlying these differences, especially from the perspective of Electroencephalography(EEG) microstate, remain poorly understood in acute stroke patients. Methods This study enrolled 14 acute stroke patients and recorded their EEG data during left and right-hand MI tasks. Four EEG microstate (A, B, C, and D) were analyzed to extract temporal feature parameters, including Duration, Occurrence Coverage, and transition probabilities(TP). Significant features were used to construct classification models using Linear Discriminant Analysis(LDA), Support Vector Machines(SVM), and K-Nearest Neighbors(KNN) algorithms. Results Microstate analysis revealed significant differences in temporal features of microstate A and C during left and right-hand MI tasks. During left-hand MI, microstate A exhibited longer Duration(P fdr=0.032), higher Occurrence(P fdr=0.018), and greater Coverage(P fdr=0.004) compared to the right-hand, whereas microstate C showed the opposite pattern(P fdr=0.044, P fdr=0.004, P fdr=0.004). Additionally, the TP from microstate B→A, D→A and D→C also demonstrated significant differences(P fdr=0.04, P fdr<0.001, P fdr=0.006). Among classification models, the KNN algorithm achieved the highest accuracy of 75.00%, outperforming LDA and SVM. Fisher analysis indicated that the Occurrence of microstate C was the most discriminative feature for distinguishing between left and right-hand MI tasks in acute stroke patients. Conclusion Differences in EEG microstate features during left and right-hand MI tasks in acute stroke patients may reflect lateralized mechanisms of brain network reorganization. Microstate features hold significant potential for both post-stroke brain function assessment and the optimization of BCI systems. These features could enhance adaptive BCI strategies in acute stroke rehabilitation.
ISSN:1743-0003