EEG microstate analysis of PD+ and PD-: a multiband fusion approach

This study investigates the neural activity differences between Parkinson's disease patients with freezing of gait (PD+) and those without (PD-) using EEG microstate analysis, while evaluating the impact of different frequency bands on classification performance. Resting-state EEG signals from PD+ and PD- patients were analyzed to extract microstate features across five frequency bands: delta, theta, alpha, beta, and gamma. A frequency band fusion strategy was employed to optimize classification performance. Experimental results revealed that the theta band exhibited the most prominent performance in distinguishing PD+ from PD- patients, achieving a classification accuracy of 96.6% ± 2.8%. By integrating multi-band features, the classification performance was further enhanced to 98.5% ± 1.2%, significantly outperforming results from single-band analysis. Additionally, statistical analysis of theta band microstates demonstrated a significant difference in the occurrence frequency (Occurrence Per Second, OPS) of microstate B in PD+ patients (p ⪅0.05). These findings highlight the critical role of the theta band in classifying PD+ and PD- patients and demonstrate the effectiveness of the frequency band fusion strategy in improving classification accuracy. This study provides novel insights into the neural mechanisms underlying freezing of gait (FOG) in Parkinson's disease and offers theoretical support for EEG-based disease diagnosis and classification.