Evaluation of Brain Function using Microstate Analysis in Electroencephalography

Abstract

Mechanism of cognitive function and pathophysiology of neuropsychiatric disorders based on neural activity. Electroencephalography (EEG) microstate analysis is a method wherein spontaneous neural activity is segmented at sub-second levels to analyze quasi-stable states. In particular, four archetype microstates and their features are known to reflect disruptions in large-scale brain networks. This dissertation discusses EEG microstate features and their application using machine-learning algorithms for evaluating brain function. The first part of this dissertation validated the use of microstate features to diagnose schizophrenia. EEG recordings of patients and controls showed different microstate features. EEG microstate features outperformed conventional EEG ones in classification. The performance of the microstate features outperformed that of conventional EEGs, even after optimization using recursive feature elimination. The results suggested that EEG microstate features are useful for diagnosing schizophrenia. In the second part, the mental arithmetic task performance was evaluated and discriminated using the EEG microstates. Type D and type C features associated with the dorsal attention system and default mode network were different in resting and task state. The model that features selected by recursive feature elimination showed excellent classification performance for differentiating between groups, suggesting that EEG microstate features can reflect task performance. Those features also differed in EEG recordings of depressive patients and healthy controls. The findings suggest that EEG microstate features and machine-learning algorithms using them can be used as state and trait markers to evaluate brain function.