Multivariate classification of autism spectrum disorder using frequency-specific resting-state functional...

Area of Science:

• Neuroscience
• Developmental Neuroscience
• Clinical Neuroscience

Background:

• Autism spectrum disorder (ASD) is characterized by atypical whole-brain functional connectivity patterns in adolescents.
• Previous research focused on low-frequency fluctuations (0.01-0.08 Hz) for ASD discrimination.
• The role of specific frequency bands in ASD classification and symptom severity remains unclear.

Purpose of the Study:

• To investigate if functional connectivity in specific frequency bands can discriminate individuals with ASD from controls.
• To determine if relationships between connectivity and symptom severity are frequency-dependent.

Main Methods:

• Utilized resting-state fMRI data from 240 adolescents (112 with ASD, 128 controls) from the Autism Brain Imaging Data Exchange database.
• Constructed whole-brain functional connectivity networks in the Slow-5 (0.01-0.027 Hz) and Slow-4 (0.027-0.073 Hz) frequency bands.
• Employed support vector machine (SVM) for classification analysis.

Main Results:

• Achieved 79.17% classification accuracy (p<0.001) using SVM.
• Discriminative features were predominantly identified in the Slow-4 band.
• Atypical connectivity was observed between the default mode, fronto-parietal, and cingulo-opercular networks in individuals with ASD.
• Thalamic connections showed the highest classification weight in the Slow-4 band.
• A significant correlation was found between social/communication deficits (ADOS) and connectivity within the default mode and cingulo-opercular networks.

Conclusions:

• Preliminary evidence suggests frequency-specific functional connectivity indices can aid ASD detection.
• The Slow-4 frequency band shows particular promise for identifying ASD-related connectivity alterations.
• These findings may contribute to developing novel biomarkers for ASD diagnosis.