Classification of Preschoolers with Low-Functioning Autism Spectrum Disorder Using Multimodal MRI Data
Johanna Inhyang Kim1, Sungkyu Bang2, Jin-Ju Yang2
1Department of Psychiatry, Hanyang University Medical Center, 222-1 Wangsimni-ro, Sungdong-gu, Seoul, 04763, Republic of Korea.
Insights
Machine learning accurately identified autism spectrum disorder (ASD) in preschoolers using MRI scans. Combining T1 and DTI data improved diagnostic performance, highlighting key brain features.
Area of Science:
- Neuroimaging
- Machine Learning
- Developmental Neuroscience
Background:
- Autism spectrum disorder (ASD) diagnosis in preschoolers is challenging.
- Multimodal neuroimaging studies in this population are limited.
- Low-functioning autism spectrum disorder (ASD) requires specific diagnostic approaches.
Purpose of the Study:
- To apply machine learning classifiers to T1-weighted MRI and DTI data for distinguishing preschoolers with ASD from typically developing controls (TDC).
- To identify key neuroimaging features differentiating low-functioning ASD preschoolers.
- To evaluate the added value of combining T1 and DTI data for classification accuracy.
Main Methods:
- Machine learning classifiers were applied to T1-weighted MRI and diffusion tensor imaging (DTI) data.
- The study included 58 children with ASD (age 3-6 years) and 48 TDC.
- Feature importance analysis identified significant brain regions and metrics.
Main Results:
- Classification accuracy reached 88.8%, with 93.0% sensitivity and 83.8% specificity.
- Key distinguishing features included cortical thickness (right inferior occipital gyrus), mean diffusivity (middle cerebellar peduncle), and nodal efficiency (left posterior cingulate gyrus).
- Combining T1 and DTI data improved classification accuracy by approximately 10%.
Conclusions:
- Machine learning analysis of multimodal MRI data is effective for distinguishing low-functioning ASD preschoolers from TDC.
- Specific neuroimaging markers show promise for early ASD identification.
- Further large-scale, multimodal MRI studies are recommended for external validation.
Abstract:
Multimodal imaging studies targeting preschoolers and low-functioning autism spectrum disorder (ASD) patients are scarce. We applied machine learning classifiers to parameters from T1-weighted MRI and DTI data of 58 children with ASD (age 3-6 years) and 48 typically developing controls (TDC). Classification performance reached an accuracy, sensitivity, and specificity of 88.8%, 93.0%, and 83.8%, respectively. The most prominent features were the cortical thickness of the right inferior occipital gyrus, mean diffusivity of the middle cerebellar peduncle, and nodal efficiency of the left posterior cingulate gyrus. Machine learning-based analysis of MRI data was useful in distinguishing low-functioning ASD preschoolers from TDCs. Combination of T1 and DTI improved classification accuracy about 10%, and large-scale multi-modal MRI studies are warranted for external validation.
