Detecting spatially highly resolved network modules: a multi subject approach.

Area of Science:

• Neuroscience
• Network Science
• Data Science

Background:

• High-dimensional (HD) network analysis, common in neuroscience, generates vast datasets that are difficult to interpret holistically.
• Current methods for segmenting these networks into functional modules lack cross-subject consistency, hindering group-level investigations.

Purpose of the Study:

• To develop a novel computational framework for analyzing and segmenting HD brain networks at a group level.
• To overcome the challenge of non-aligned network modules across different subjects in connectivity analyses.

Main Methods:

• The proposed method rearranges subject-specific connectivity data into an integrative tensor.
• This tensor undergoes decomposition into additive factors, yielding subject-independent network structures and subject-specific loadings.
• This approach facilitates large-scale, group-wide network segmentation.

Main Results:

• The integrative tensor decomposition successfully generates subject-independent network representations.
• Subject-specific loadings enable consistent identification and comparison of network modules across individuals.
• The method allows for robust group-level segmentation of high-dimensional brain networks.

Conclusions:

• The proposed tensor decomposition method offers a robust solution for group-level analysis of HD brain networks.
• This approach enhances the ability to identify consistent functional network modules across a population.
• The findings pave the way for more comprehensive and scalable network neuroscience research.