Alzheimer's Disease Detection Using Combined EEG Source Connectivity and Microstate Features.

Background/Objectives: Electroencephalography (EEG) connectivity and microstate analysis have shown great potential for Alzheimer's disease (AD) diagnosis; however, their clinical application remains limited by the low spatial resolution of EEG and the lack of standardized microstate analysis. To address these challenges, this study proposes a multi-domain feature fusion framework, namely Source-localized Microstate and Multi-frequency Synchronization (SMMS), which integrates EEG source localization (ESL)-based weighted phase lag index (wPLI) functional connectivity with EEG microstate features. Methods: Specifically, ESL was employed to improve the spatial resolution of EEG signals for constructing functional connectivity matrices, while multi-frequency-band wPLI features were extracted to characterize functional synchronization among cortical regions. Meanwhile, EEG microstate features were utilized to capture the temporal dynamics of brain functional states. The proposed framework was evaluated on a public OpenNeuro dataset comprising 36 AD patients, 23 frontotemporal dementia (FTD) patients, and 29 healthy controls (HCs), as well as an additional clinical dataset collected from 48 AD patients at Sir Run Run Shaw Hospital, Hangzhou, China. Results: Experimental results showed that the proposed SMMS framework achieved high classification performance on both datasets. Conclusions: These findings demonstrate its effectiveness for EEG-based Alzheimer's disease diagnosis.