Spatio-functional parcellation of resting state fMRI

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain Network Analysis

Background:

• Resting-state functional Magnetic Resonance Imaging (rs-fMRI) reveals spontaneous brain activity crucial for understanding neural function.
• Accurate analysis of rs-fMRI data necessitates brain parcellation that balances spatial and functional homogeneity.
• Existing parcellation methods often present a trade-off between functional similarity within clusters and anatomical alignment.

Purpose of the Study:

• To present a novel multistage clustering scheme for rs-fMRI data.
• To achieve spatially and functionally homogenous brain parcellations.
• To improve the identification and characterization of brain networks from rs-fMRI data.

Main Methods:

• Development of a multistage clustering approach for rs-fMRI data.
• Evaluation of cluster spatial and functional homogeneity compared to existing methods.
• Application of the scheme to identify distinct brain networks.

Main Results:

• The proposed multistage approach successfully identifies various brain networks.
• Functional homogeneity of the clusters surpasses that of functional atlases and k-means clustering.
• Spatial homogeneity of the clusters is superior to Independent Component Analysis (ICA) and k-means clustering.

Conclusions:

• The novel multistage clustering scheme offers improved spatial and functional homogeneity for rs-fMRI brain parcellation.
• This method enhances the identification and analysis of brain networks derived from resting-state fMRI data.
• The findings suggest a more robust approach for understanding brain organization and function.