A radiomics-based brain network in T1 images: construction, attributes, and applications

Area of Science:

• Neuroimaging
• Brain Network Analysis
• Radiomics

Background:

• T1-weighted magnetic resonance imaging (MRI) is common but underutilized for individual brain network construction.
• Existing methods often fail to capture the nuances of individual structural brain connectivity.

Purpose of the Study:

• To develop a novel individualized radiomics-based structural similarity network (iRSSN) from T1 images.
• To evaluate the network characteristics and clinical applicability of the iRSSN.

Main Methods:

• Utilized voxel-based morphometry on T1 images to obtain gray matter density.
• Extracted radiomic features from regions of interest defined by the Brainnetome atlas.
• Constructed the iRSSN based on correlations of radiomic features between brain regions.
• Assessed network properties including graph theory metrics, reliability, and individual identification (fingerprinting).

Main Results:

• The iRSSN demonstrated robust network characteristics and high test-retest reliability.
• The iRSSN showed significant individual identification ability.
• Outperformed other network types in mild cognitive impairment subtype discrimination and fluid intelligence prediction on large datasets.

Conclusions:

• The iRSSN offers a unique, reliable, and informative individualized structural brain network.
• This method provides a valuable tool for clinical and phenotypic applications, potentially integrating with other neuroimaging modalities like resting-state functional connectivity.