|
KMID : 1137820220430060375
|
|
의공학회지
2022 Volume.43 No. 6 p.375 ~ p.381
|
|
|
A Comparative Analysis of Motor Imagery, Execution, and Observation for Motor Imagery-based Brain-Computer Interface
|
|
Gwon Da-Eun
Hwang Min-Joo
Kwon Ji-Hyun
Shin Ye-Eun
Ahn Min-Kyu
|
|
|
Abstract
|
|
|
|
Brain-computer interface (BCI) is a technology that allows users with motor disturbance to control machines by brainwaves without a physical controller. Motor imagery (MI)-BCI is one of the popular BCI techniques, but it needs a long calibration time for users to perform a mental task that causes high fatigue to the users. MI is reported as showing a similar neural mechanism as motor execution (ME) and motor observation (MO). However, integrative investigations of these three tasks are rarely conducted. In this study, we propose a new paradigm that incorporates three tasks (MI, ME, and MO) and conducted a comparative analysis. For this study, we collected Electroencephalograms (EEG) of motor imagery/execution/observation from 28 healthy subjects and investigated alpha event-related (de)syn- chronization (ERD/ERS) and classification accuracy (left vs. right motor tasks). As result, we observed ERD and ERS in MI, MO and ME although the timing is different across tasks. In addition, the MI showed strong ERD on the contralateral hemisphere, while the MO showed strong ERD on the ipsilateral side. In the classification analysis using a Riemannian geom- etry-based classifier, we obtained classification accuracies as MO (66.34%), MI (60.06%) and ME (58.57%). We con- clude that there are similarities and differences in fundamental neural mechanisms across the three motor tasks and that these results could be used to advance the current MI-BCI further by incorporating data from ME and MO.
|
|
|
KEYWORD
|
|
|
Motor imagery, Motor execution, Motor observation, Brain-computer interface
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
 
|
|