Graph Ricci curvatures reveal atypical functional connectivity in autism spectrum disorder

Area of Science:

• Neuroscience
• Network Science
• Developmental Disorders

Background:

• Standard graph-theoretic measures are common for analyzing autism spectrum disorder (ASD) functional connectivity.
• Geometry-inspired network measures, like graph Ricci curvatures, have not been previously applied to ASD research.

Purpose of the Study:

• To apply Forman-Ricci and Ollivier-Ricci curvatures to investigate brain network differences in ASD.
• To identify brain regions with atypical connectivity in ASD using novel geometric measures.
• To explore the clinical relevance of these geometric network differences in ASD.

Main Methods:

• Utilized the Autism Brain Imaging Data Exchange I (ABIDE-I) dataset with 1112 participants.
• Applied Forman-Ricci and Ollivier-Ricci curvatures to resting-state functional connectivity networks.
• Conducted meta-analysis decoding to link curvature differences to cognitive domains and intervention targets.

Main Results:

• Identified significant brain-wide and region-specific differences in both Forman-Ricci and Ollivier-Ricci curvatures between ASD and typically developing individuals.
• Region-specific differences for Forman-Ricci curvature were concentrated in Default Mode, Somatomotor, and Ventral Attention networks.
• Brain regions exhibiting curvature differences are associated with cognitive impairments in ASD and respond to non-invasive stimulation.

Conclusions:

• Graph Ricci curvatures are a promising tool for characterizing atypical functional connectivity in ASD.
• These geometric measures can identify clinically relevant brain regions for understanding and potentially treating ASD and other neurodevelopmental disorders.