Eigenvector Centrality Characterization on fMRI Data: Gender and Node Differences in Normal and ASD Subjects : Author...

Area of Science:

• Neuroscience
• Brain Imaging
• Machine Learning

Background:

• Abnormal brain network characteristics are increasingly studied for diagnosing neurological conditions.
• Functional magnetic resonance imaging (fMRI) provides valuable data for analyzing brain networks.
• Simpler evaluation methods for brain network analysis are needed.

Purpose of the Study:

• To evaluate the effectiveness of eigenvector centrality measures from fMRI in distinguishing individuals with autism spectrum disorder (ASD) from typically developing controls.
• To explore the utility of network node centrality values for ASD diagnosis.

Main Methods:

• Functional magnetic resonance imaging (fMRI) was used to obtain brain data.
• Eigenvector centrality was calculated to represent network properties.
• Boxplot formalism and classification and regression tree models were employed for analysis.

Main Results:

• Significant region-wise differences in network centrality were observed between ASD subjects and controls.
• These differences were prominent in the frontoparietal, limbic, ventral attention, default mode, and visual networks.
• A reduced number of regions-of-interest (ROI) were identified as discriminative.

Conclusions:

• Eigenvector centrality measures derived from fMRI are suitable for discriminating ASD subjects.
• Automated machine learning algorithms offer advantages over manual classification for ASD diagnosis based on brain networks.