Functional connectivity based machine learning approach for autism detection in young children using MEG signals
Kasturi Barik1, Katsumi Watanabe2, Joydeep Bhattacharya3
1Department of Electronics and Electrical Communication Engineering, Indian Institute of Technology Kharagpur, Kharagpur, India.
Journal of Neural Engineering
|February 22, 2023
Summary
This study identifies functional brain connectivity patterns as early autism biomarkers in children. Coherence-based analysis of magnetoencephalogram data achieved high accuracy in detecting autism spectrum disorder (ASD).
Area of Science:
- Neuroscience
- Biomarkers
- Developmental Disorders
Background:
- Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder.
- Early identification of autism biomarkers is crucial for improving outcomes.
- Functional brain connectivity patterns may serve as potential biomarkers.
Purpose of the Study:
- To identify novel biomarkers for autism spectrum disorder (ASD) in children.
- To investigate functional brain connectivity patterns using magnetoencephalogram (MEG) data.
- To assess the efficacy of coherence-based (COH) measures for autism detection.
Main Methods:
- Resting-state MEG signals were recorded from children with ASD and typically developing (TD) controls.
- Complex coherency-based functional connectivity analysis was applied to characterize neural activity.
- Artificial neural network (ANN) and support vector machine (SVM) classifiers were used for autism detection.
Main Results:
- High classification accuracy (91.66%) was achieved using COH features in the high gamma band (50-100 Hz).
- Combining delta (1-4 Hz) and gamma band features yielded accuracies of 95.03% (ANN) and 93.33% (SVM).
- Children with ASD exhibited significant hyperconnectivity, supporting the weak central coherency theory.
Conclusions:
- Functional brain connectivity patterns are effective biomarkers for early autism detection in children.
- Region-wise COH analysis demonstrates superior performance compared to sensor-wise analysis.
- These findings contribute to improved diagnostic approaches for autism spectrum disorder.
