Data-determined window size and space-oriented segmentation of spontaneous EEG map series

Area of Science:

• Neuroscience
• Computational Neuroscience
• Brain Imaging Analysis

Background:

• Brain electric microstates represent quasi-stable topographical patterns of the scalp potential distribution.
• Accurate segmentation of these microstates is crucial for understanding brain dynamics.
• Current methods for microstate segmentation often rely on subjective parameter selection.

Purpose of the Study:

• To develop and validate a data-driven window-determining function for optimizing microstate segmentation.
• To establish normative data for brain electric microstate durations using the new method.
• To assess the reliability and objectivity of the proposed segmentation approach.

Main Methods:

• Extraction of landscape descriptors (extreme potentials, centroids) from momentary potential distribution maps.
• Development of a window-determining function balancing map similarity and dissimilarity.
• Segmentation of 211 two-second map epochs from 8 normal subjects using data-determined window sizes.
• Validation through random permutation analysis to rule out procedural artifacts.

Main Results:

• The data-driven window-determining function provided optimal window sizes for microstate segmentation.
• Mean microstate durations did not significantly differ between the two descriptors used (144 msec vs. 143 msec).
• Segmentation results were robust and not attributable to artifacts, as confirmed by permutation testing.

Conclusions:

• The proposed window-determining function offers an objective, data-driven approach to microstate segmentation.
• This method reliably identifies quasi-stable brain electric microstates and provides normative duration data.
• The findings support the validity of the segmentation technique for analyzing brain electric activity.