Sep 2026· Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence· 0 citations· 40 references
TL;DR
Experimental results on three large-scale fMRI datasets demonstrate that BrainCGT achieves consistently better performance than existing graph-based methods for neurological disorder classification, highlighting the importance of incorporating causal directionality into brain graph transformer architectures for robust and interpretable neuroimaging analysis.
Abstract
Brain connectivity analysis is a fundamental tool for identifying biomarkers and understanding of neurological disorders. Most existing approaches employ graph transformers over undirected functional connectivity networks, which are typically estimated using correlation statistics. Although effective for capturing statistical associations, these models do not represent directed interactions between brain regions that arise from causal relationships. As a result, direction-specific disease mechanisms are not explicitly modeled, and interpretability is often limited. To address this gap, we present BrainCGT, a brain graph transformer designed to model causal connectivity inferred from fMRI time-series data. In this framework, brain networks are modeled as directed graphs with a modular organization, where nodes correspond to individual brain regions and directed edges reflect causal flow of information between them. Direction-aware node representations together with direction-biased attention mechanisms allow the model to capture asymmetric interactions across regions. Experimental results on three large-scale fMRI datasets demonstrate that BrainCGT achieves consistently better performance than existing graph-based methods for neurological disorder classification. In addition, examination of the learned attention structures shows correspondence with established neurobiological pathways, suggesting improved interpretability. These results highlight the importance of incorporating causal directionality into brain graph transformer architectures for robust and interpretable neuroimaging analysis.
It is concluded that while GNNs have significantly advanced the field, challenges remain in handling data heterogeneity, improving model robustness and interpretability, and achieving clinical translation.
BrainLinear is proposed, a lightweight geometry-aware framework for mining disease-discriminative connectome patterns, and shows that it improves AUC and ACC over the best baseline for each metric by up to $3.54 and $1.39 percentage points, respectively.
Si-Jing Wu, Dong-Yuan Li, Miaoting Huang et al.· 0 citations
An edge-centric brain transformer (EBT) framework is proposed that reformulates rs-fMRI analysis as functional connection representation learning and consistently outperforms representative graph neural networks, brain transformers, and conventional connectivity-based approaches.
Deng-Yi Zhao, Zhi-Heng Zhou, Meng-Yao Zhou et al.· 0 citations
An intelligent computational system that integrates genomic data with functional magnetic resonance imaging to establish meaningful correlations between genetic variations in key schizophrenia-associated genes and altered brain connectivity patterns to deliver a comprehensive gene-brain analysis system capable of predi...
Aditya Pujari, Aditya Jagtap, Aditi Dixit et al.· International Journal for Re...· 0 citations
Autism spectrum disorder (ASD) is associated with heterogeneous alterations across distributed brain systems, creating challenges for whole-connectome inference. The difficulty arises not only from the large number of connections, but also from dependence among effects indexed by anatomically and functionally related r...
A population-guided dynamic optimization mechanism for individual brain graphs, which feeds the stable disease-discriminative information learned at the population level back into the structural updating of individual brain graphs, thereby enhancing disease-related connections while suppressing site-related noisy conne...
Xuan He, Ting Mei, Huan-Huan Du et al.· Neural Networks· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.