Related Experiment Video
Updated: Oct 23, 2025
12:21
Probing the Brain in Autism Using fMRI and Diffusion Tensor Imaging
Published on: September 12, 2011
25.4K
Classification of Autism Spectrum Disorder From EEG-Based Functional Brain Connectivity Analysis.
Noura Alotaibi1, Koushik Maharatna2
1Department of Electronics and Computer Science, University of Southampton, Southampton, SO17 1BJ, UK nma1y17@soton.ac.uk.
Neural Computation
|August 19, 2021
Summary
Early autism diagnosis using electroencephalogram (EEG) functional brain connectivity shows promise. Machine learning accurately distinguished autism spectrum disorder (ASD) children from typical children, aiding early intervention strategies.
Area of Science:
- Neuroscience
- Computational Psychiatry
- Biomedical Engineering
Background:
- Autism Spectrum Disorder (ASD) diagnosis relies on behavioral assessments, with increasing prevalence necessitating earlier detection methods.
- Early diagnosis of ASD is crucial for timely intervention, potentially mitigating long-term health and socioeconomic impacts.
- Investigating objective biomarkers for ASD is essential to complement existing diagnostic approaches.
Purpose of the Study:
- To develop and validate a machine learning framework for classifying children with ASD using functional brain connectivity.
- To assess the efficacy of phase-based functional brain connectivity measures derived from electroencephalogram (EEG) for ASD detection.
- To identify specific patterns of brain connectivity alterations in children with ASD.
Main Methods:
- Utilized a dataset of 12 children with ASD and 12 typically developing children.
- Computed functional brain connectivity networks using three phase-locking value (PLV) based approaches.
- Characterized networks with graph-theoretic parameters and employed a cubic support vector machine (SVM) classifier.
- Focused analysis on the theta frequency band for identifying significant connectivity changes.
Main Results:
- Achieved high classification performance: 95.8% accuracy, 100% sensitivity, and 92% specificity using the trial-averaged PLV approach with SVM.
- Identified significant alterations in functional brain connectivity within the theta band in children with ASD.
- Demonstrated the effectiveness of aggregated graph-theoretic features for characterizing ASD-related connectivity changes.
Conclusions:
- Functional brain connectivity, particularly using EEG-derived measures, shows significant potential as an objective tool for classifying ASD in children.
- The developed machine learning approach offers a promising avenue for early ASD detection and subsequent intervention.
- Findings highlight specific theta band connectivity alterations as potential biomarkers for ASD.
