Volumetric Analysis of Amygdala and Hippocampal Subfields for Infants with Autism

Area of Science:

• Neuroimaging
• Developmental Neuroscience
• Biomedical Engineering

Background:

• Autism spectrum disorder (ASD) is associated with abnormal brain overgrowth in children.
• Developmental trajectories of specific brain regions like the amygdala and hippocampal subfields in infants with ASD are not well-documented.

Purpose of the Study:

• To investigate the development of amygdala and hippocampal subfields in infants at high risk for ASD.
• To establish early neuroanatomical markers for ASD using advanced imaging techniques.

Main Methods:

• Utilized a longitudinal dataset of infant brain MRI scans from 6 to 24 months of age.
• Developed and applied a novel deep learning model (Dilated-Dense U-Net) for accurate segmentation of amygdala and hippocampal subfields.
• Conducted volume-based analysis on the segmented brain regions.

Main Results:

• Infants later diagnosed with ASD exhibited significantly larger left and right amygdala volumes compared to typically developing controls.
• Infants later diagnosed with ASD also showed larger volumes in hippocampal subfields than typically developing controls.
• These findings suggest early abnormal brain overgrowth in key limbic structures in infants who develop ASD.

Conclusions:

• Early overgrowth in amygdala and hippocampal subfields may be a potential neuroimaging biomarker for autism spectrum disorder.
• The findings highlight the importance of studying infant brain development to understand the early neuropathology of ASD.
• Advanced deep learning techniques can improve the analysis of small and low-contrast brain structures in pediatric neuroimaging.