A Longitudinal MRI Study of Amygdala and Hippocampal Subfields for Infants with Risk of Autism

Area of Science:

• Neuroimaging
• Developmental Neuroscience
• Biomarkers

Background:

• Current autism spectrum disorder (ASD) diagnosis relies on behavioral observations at 3-4 years, missing critical early intervention windows.
• Early intervention is vital before 2 years old, but effective early biomarkers for infants at risk of ASD are lacking.
• Previous research shows altered amygdala and hippocampus development in older ASD individuals, but early postnatal trajectories are poorly understood.

Purpose of the Study:

• To investigate early developmental trajectories of the amygdala and hippocampal subfields in infants at risk of ASD.
• To identify potential neuroimaging biomarkers for early detection of autism spectrum disorder.
• To establish volume-based analysis of infant brain structures for ASD risk assessment.

Main Methods:

• A novel deep-learning approach, dilated-dense U-Net, was developed for accurate segmentation of infant amygdala and hippocampal subfields.
• Longitudinal brain imaging data from infants at risk of ASD (ages 6, 12, and 24 months) from the National Database for Autism Research (NDAR) were analyzed.
• Volume-based analysis was performed on segmented amygdala and hippocampal subfields to assess developmental changes.

Main Results:

• The study observed potential overgrowth in the amygdala and cornu ammonis (CA) sectors 1-3 starting as early as 6 months of age in infants at risk of ASD.
• Deep learning segmentation successfully addressed challenges of low tissue contrast and small structure size in infant brain imaging.
• These findings suggest early alterations in specific brain regions may precede typical behavioral diagnosis of ASD.

Conclusions:

• Overgrowth in the amygdala and specific hippocampal subfields (CA1-3) from 6 months of age may serve as an early neuroimaging biomarker for autism spectrum disorder risk.
• The developed deep learning segmentation method enables precise analysis of infant brain structures for early ASD detection.
• Identifying these early biomarkers facilitates timely intervention during a critical developmental window, potentially improving outcomes for infants at risk of ASD.