Transformer-based structural connectivity networks for ADHD-related connectivity alterations.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Developmental Neuroscience

Background:

• Attention-deficit/hyperactivity disorder (ADHD) is a prevalent neurodevelopmental disorder impacting behavior and learning.
• Current ADHD diagnosis relies on subjective assessments, highlighting the need for objective, imaging-based diagnostic tools.
• This research explores the potential of magnetic resonance imaging (MRI)-derived structural connectivity networks for identifying ADHD-related alterations.

Purpose of the Study:

• To investigate whether structural connectivity networks from MRI data can reveal alterations associated with ADHD.
• To leverage Transformer-based deep learning models for constructing and analyzing brain structural connectivity.
• To support a data-driven understanding of the neurobiological underpinnings of ADHD.

Main Methods:

• Utilized brain MRI data from 947 individuals (aged 7-26 years) from the ADHD-200 dataset.
• Employed Transformer-based deep learning models to learn brain region relationships and build structural connectivity networks.
• Applied five-fold cross-validation and statistical analyses to assess model performance and group differences.

Main Results:

• The developed method accurately distinguished individuals with ADHD from healthy controls with 71.9% accuracy and an AUC of 0.74.
• Significant differences in structural connectivity patterns were observed (P < 10^-6), particularly in regions related to motor and executive functions.
• Brain regions such as the thalamus and caudate showed markedly different importance rankings between ADHD and control groups.

Conclusions:

• ADHD is associated with alterations in structural connectivity across multiple brain regions.
• Transformer-based methods for building brain structural connectivity networks show potential for ADHD diagnosis and research.
• Objective, network-based approaches can enhance our understanding and diagnosis of neurodevelopmental disorders like ADHD.