Temporal and spatial variability of dynamic microstate brain network in early Parkinson's disease

Changes of brain network dynamics reveal variations in macroscopic neural activity patterns in behavioral and cognitive aspects. Quantification and application of changed dynamics in brain functional connectivity networks may contribute to a better understanding of brain diseases, and ultimately provide better prognostic indicators or auxiliary diagnostic tools. At present, most studies are focused on the properties of brain functional connectivity network constructed by sliding window method. However, few studies have explored evidence-based brain network construction algorithms that reflect disease specificity. In this work, we first proposed a novel approach to characterize the spatiotemporal variability of dynamic functional connectivity networks based on electroencephalography (EEG) microstate, and then developed a classification framework for integrating spatiotemporal variability of brain networks to improve early Parkinson's disease (PD) diagnostic performance. The experimental results indicated that compared with the brain network construction method based on conventional sliding window, the proposed method significantly improved the performance of early PD recognition, demonstrating that the dynamic spatiotemporal variability of microstate-based brain networks can reflect the pathological changes in the early PD brain. Furthermore, we observed that the spatiotemporal variability of early PD brain network has a specific distribution pattern in brain regions, which can be quantified as the degree of motor and cognitive impairment, respectively. Our work offers innovative methodological support for future research on brain network, and provides deeper insights into the spatiotemporal interaction patterns of brain activity and their variabilities in early PD.

Automated summary

The study proposes a novel method to describe the spatiotemporal variability of dynamic functional connectivity networks and develops a classification framework to improve the diagnostic performance of early Parkinson's disease. The results show that the proposed method significantly improves the recognition performance of early Parkinson's disease compared to the traditional sliding window method, indicating that the dynamic spatiotemporal variability of microstate-based brain networks can reflect the pathological changes in the early Parkinson's disease brain. Furthermore, the spatiotemporal variability of the early Parkinson's disease brain network has a specific distribution pattern in different brain regions, which can be quantified as the degree of motor and cognitive impairment.