White Matter Connectome Correlates of Auditory Over-Responsivity: Edge Density Imaging and Machine-Learning...

Area of Science:

• Neuroimaging
• Developmental Neuroscience
• Clinical Psychology

Background:

• Sensory over-responsivity (SOR), particularly auditory over-responsivity (AOR), affects children with and without neurodevelopmental challenges.
• Understanding the neural underpinnings of AOR is crucial for developing effective interventions.

Purpose of the Study:

• To investigate white matter microstructural and connectome correlates of AOR in school-aged boys.
• To identify potential imaging biomarkers for AOR using diffusion tensor imaging (DTI) and connectome analysis.
• To evaluate the efficacy of machine learning algorithms in classifying children with AOR.

Main Methods:

• Analysis of prospectively collected DTI and high-resolution T1 data from 39 boys (aged 8-12 years).
• Calculation of DTI metrics (FA, MD, RD, AD) and connectome Edge Density (ED) maps.
• Voxel-wise analysis using tract-based spatial statistics and stepwise logistic regression for biomarker identification.
• Classification of AOR using various machine learning models (naïve Bayes, random forest, SVM).

Main Results:

• Children with AOR exhibited widespread white matter microstructural impairments, including lower FA and higher MD/RD.
• Reduced connectome ED was observed in the anterior-superior corona radiata and corpus callosum.
• Average FA of the left superior longitudinal fasciculus (SLF) emerged as a significant predictor of AOR (p=0.007), achieving an AUC of 0.756.
• Random forest models utilizing ED demonstrated higher classification accuracy for AOR.

Conclusions:

• Extensive white matter microstructural impairments and altered connectomic organization are present in children with AOR.
• Average FA of the left SLF serves as a potential region-of-interest-based imaging biomarker for predicting SOR.
• Machine learning models, particularly those using connectome ED, offer accurate and objective image-based classification for AOR.