MHATC: Autism Spectrum Disorder Identification Utilizing Multi-Head Attention Encoder Along with Temporal...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Imaging

Background:

• Resting-state functional magnetic resonance imaging (fMRI) is utilized for Autism Spectrum Disorder (ASD) diagnosis via network-based functional connectivity.
• ASD is linked to alterations in brain regions and their inter-connections, posing challenges for accurate classification.
• Existing deep learning models face limitations in classifying ASD from brain imaging data.

Purpose of the Study:

• To develop a novel deep learning architecture for improved ASD classification using resting-state fMRI.
• To address the limitations of current deep neural network solutions in analyzing brain connectivity patterns for ASD diagnosis.

Main Methods:

• A novel deep learning architecture, MHATC, incorporating multi-head attention and temporal consolidation modules was proposed.
• The MHATC model was designed to classify individuals as ASD patients based on resting-state fMRI data.

Main Results:

• The MHATC architecture demonstrated robustness and computational efficiency.
• The proposed method offers potential for improved ASD classification in research and clinical settings.

Conclusions:

• The MHATC deep learning model presents a promising advancement for diagnosing Autism Spectrum Disorder using fMRI.
• The architecture's efficiency and robustness suggest broad applicability in neurological research and clinical diagnostics.