STARFormer: A novel spatio-temporal aggregation reorganization transformer of FMRI for brain disorder diagnosis.

Researchers developed a new computer program called STARFormer that analyzes brain scans to help diagnose autism and ADHD. The program looks at both how different brain areas connect and how brain activity changes over time, which previous methods often missed. When tested on brain scan databases, it performed better than existing tools at correctly identifying these conditions, potentially offering more accurate diagnosis methods.

This study introduces STARFormer, a novel artificial intelligence method for analyzing brain scans (fMRI) to diagnose autism spectrum disorder (ASD) and ADHD. The approach addresses limitations of existing methods by better integrating spatial and temporal patterns in brain activity. STARFormer incorporates three key components: spatial analysis of brain regions using connectivity patterns, temporal feature extraction through systematic time series analysis, and fusion of spatial-temporal features using parallel transformer architecture. Testing on two public datasets demonstrated state-of-the-art performance across multiple evaluation metrics, suggesting improved accuracy and reliability for brain disorder diagnosis compared to existing approaches.

While showing promise for improved fMRI-based diagnosis of ASD and ADHD, clinical application requires further validation studies. The method may eventually contribute to more objective diagnostic tools, but current evidence represents technological advancement rather than established clinical utility. Implementation would require specialized neuroimaging facilities and expertise.

Abstract does not report sample sizes, specific accuracy metrics, validation methods, or comparison benchmarks. No information on clinical populations studied, demographic characteristics, or real-world diagnostic performance. The study appears to be primarily methodological rather than clinical validation.

Many existing methods that use functional magnetic resonance imaging (fMRI) to classify brain disorders, such as autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD), often overlook the integration of spatial and temporal dependencies of the blood oxygen level-dependent (BOLD) signals, which may lead to inaccurate or imprecise classification results. To solve this problem, we propose a spatio-temporal aggregation reorganization transformer (STARFormer) that effectively captures both spatial and temporal features of BOLD signals by incorporating three key modules. The region of interest (ROI) spatial structure analysis module uses eigenvector centrality (EC) to reorganize brain regions based on effective connectivity, highlighting critical spatial relationships relevant to the brain disorder. The temporal feature reorganization module systematically segments the time series into equal-dimensional window tokens and captures multiscale features through variable window and cross-window attention.

The spatio-temporal feature fusion module employs a parallel transformer architecture with dedicated temporal and spatial branches to extract integrated features. The proposed STARFormer has been rigorously evaluated on two publicly available datasets for the classification of ASD and ADHD. The experimental results confirm that STARFormer achieves state-of-the-art performance across multiple evaluation metrics, providing a more accurate and reliable tool for the diagnosis of brain disorders and biomedical research. The official implementation codes are available at: https://github.com/NZWANG/STARFormer.