Sunday 02 February 2025
A new approach has been developed to help diagnose autism spectrum disorder (ASD) using functional magnetic resonance imaging (fMRI). The technique, known as ASD-HNet, uses a combination of machine learning and graph theory to analyze brain activity patterns in individuals with ASD.
The researchers used data from the Autism Brain Imaging Data Exchange (ABIDE) dataset, which includes fMRI scans from over 1,000 individuals with ASD and nearly 2,000 controls. They developed a hierarchical feature extraction model that uses graph convolutional networks to identify patterns of brain activity associated with ASD.
The model was trained on a subset of the data and then tested on the remaining samples. The results showed that ASD-HNet outperformed existing methods in accurately identifying individuals with ASD, achieving an accuracy rate of over 73%.
One of the key advantages of ASD-HNet is its ability to extract features from brain activity patterns at multiple scales. This allows it to capture subtle differences between individuals with and without ASD, which may not be apparent when analyzing individual brain regions or networks in isolation.
The researchers also found that the model was able to identify specific brain regions and networks that are associated with ASD. These include the default mode network, which is responsible for introspection and self-reflection, as well as the salience network, which is involved in attention and arousal.
The development of ASD-HNet has significant implications for the diagnosis and treatment of ASD. Currently, there is no single test or biomarker that can definitively diagnose ASD, and the disorder is often diagnosed based on behavioral symptoms. The ability to identify specific brain activity patterns associated with ASD could potentially lead to more accurate and earlier diagnoses.
The researchers are now planning to further validate the model using additional data from other studies. They also hope to use ASD-HNet to develop personalized treatment plans for individuals with ASD, tailored to their unique brain activity patterns.
In addition to its potential clinical applications, ASD-HNet may also have broader implications for our understanding of the human brain and behavior. The ability to analyze complex brain networks and identify subtle differences between healthy and disordered brains could potentially lead to new insights into the neural basis of cognition and behavior.
Cite this article: “New Brain Imaging Technique Shows Promise in Diagnosing Autism Spectrum Disorder”, The Science Archive, 2025.
Autism Spectrum Disorder, Functional Magnetic Resonance Imaging, Fmri, Machine Learning, Graph Theory, Brain Activity Patterns, Hierarchical Feature Extraction, Graph Convolutional Networks, Default Mode Network, Salience Network
