Mohammad Naser
(Advisor: Dr. Sylvia Bhattacharya)

will defend a doctoral thesis entitled,

MAPPING RESTING-STATE ELECTROENCEPHALOGRAPHY (EEG) ONTO MOTOR IMAGERY FOR CROSS-SUBJECT BRAIN–COMPUTER INTERFACE APPLICATIONS

On

Tuesday, November 18th at 8:30 AM
Atrium Building, Room J1223

Abstract

This dissertation proposes a framework for reducing subject-specific calibration requirements in motor imagery (MI) brain–computer interface (BCI) systems by using resting-state electroencephalography (EEG) as a predictive marker of MI characteristics. Traditional MI-BCI models suffer from severe inter-subject variability and non-stationarity, requiring each user to provide extensive MI data for model training, an impractical barrier for real-world and clinical use. To address this, the study investigates whether intrinsic resting-state dynamics can infer an individual’s MI signal patterns and inform cluster-based model generalization. Using data from multiple benchmark EEG datasets (BCI Competition IV-2a and PhysioNet), spectral and connectivity features of resting EEG were analyzed to identify hemispheric asymmetries and cluster subjects into physiologically homogeneous groups. Deep learning classifiers trained within these rest-informed clusters were evaluated for MI decoding performance under minimal or zero training conditions. Results demonstrate that subjects exhibiting resting-state hemispheric dominance show correspondingly stronger lateralized MI responses, validating the predictive relationship between rest and task states. Furthermore, models trained within these rest-based clusters achieved significantly higher cross-subject accuracy compared to mixed-subject baselines, reducing the dependence on individualized calibration. This work establishes resting-state EEG as both a meaningful neurophysiological descriptor and a practical calibration surrogate, offering a pathway toward plug-and-play MI-BCI systems. The findings have direct implications for assistive technologies such as EEG-controlled wheelchairs and broader applications in adaptive neuroengineering, where resting-state activity can inform user-specific adaptation without additional task data.

Committee

• Dr. Sylvia Bhattacharya – Department of Engineering Technology (advisor)
• Dr. Awatef Ergai – Department of Industrial and Systems Engineering
• Dr. Hoseon Lee – Department of Electrical and Computer Engineering
• Dr. Rodney Roberts – College of Engineering, Florida A&M-Florida State University
• Dr. Sumit Chakravarty – Department of Electrical and Computer Engineering