Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Self-supervised Graph Transformer with Contrastive Learning for Brain Connectivity Analysis towards Improving Autism Detection

Yicheng Leng, Syed Muhammad Anwar, Islem Rekik, Sen He, Eung-Joo Lee

TL;DR

The paper tackles autism detection from rs-fMRI functional connectivity by introducing a contrastive self-supervised learning framework for a Brain Network Transformer (BNT). It proposes graph dilation and shrinkage as brain-aware augmentations and a MoCo-based CSSL objective to pretrain the BNT on correlation matrices $C \in \mathbb{R}^{V\times V}$, followed by finetuning for the downstream task. Empirical results on ABIDE show a peak AUROC of $82.6$ and accuracy of $74\%$, outperforming prior GNN-based methods with stable performance. This approach reduces reliance on labeled data, offers robust representations for brain connectivity analysis, and holds promise for scalable ASD biomarkers. The work demonstrates the practicality of CSSL in neuroimaging, combining domain-specific augmentations with transformer-based graph models to improve diagnostic performance.

Abstract

Functional Magnetic Resonance Imaging (fMRI) provides useful insights into the brain function both during task or rest. Representing fMRI data using correlation matrices is found to be a reliable method of analyzing the inherent connectivity of the brain in the resting and active states. Graph Neural Networks (GNNs) have been widely used for brain network analysis due to their inherent explainability capability. In this work, we introduce a novel framework using contrastive self-supervised learning graph transformers, incorporating a brain network transformer encoder with random graph alterations. The proposed network leverages both contrastive learning and graph alterations to effectively train the graph transformer for autism detection. Our approach, tested on Autism Brain Imaging Data Exchange (ABIDE) data, demonstrates superior autism detection, achieving an AUROC of 82.6 and an accuracy of 74%, surpassing current state-of-the-art methods.

Self-supervised Graph Transformer with Contrastive Learning for Brain Connectivity Analysis towards Improving Autism Detection

TL;DR

The paper tackles autism detection from rs-fMRI functional connectivity by introducing a contrastive self-supervised learning framework for a Brain Network Transformer (BNT). It proposes graph dilation and shrinkage as brain-aware augmentations and a MoCo-based CSSL objective to pretrain the BNT on correlation matrices , followed by finetuning for the downstream task. Empirical results on ABIDE show a peak AUROC of and accuracy of , outperforming prior GNN-based methods with stable performance. This approach reduces reliance on labeled data, offers robust representations for brain connectivity analysis, and holds promise for scalable ASD biomarkers. The work demonstrates the practicality of CSSL in neuroimaging, combining domain-specific augmentations with transformer-based graph models to improve diagnostic performance.

Abstract

Functional Magnetic Resonance Imaging (fMRI) provides useful insights into the brain function both during task or rest. Representing fMRI data using correlation matrices is found to be a reliable method of analyzing the inherent connectivity of the brain in the resting and active states. Graph Neural Networks (GNNs) have been widely used for brain network analysis due to their inherent explainability capability. In this work, we introduce a novel framework using contrastive self-supervised learning graph transformers, incorporating a brain network transformer encoder with random graph alterations. The proposed network leverages both contrastive learning and graph alterations to effectively train the graph transformer for autism detection. Our approach, tested on Autism Brain Imaging Data Exchange (ABIDE) data, demonstrates superior autism detection, achieving an AUROC of 82.6 and an accuracy of 74%, surpassing current state-of-the-art methods.

Paper Structure

This paper contains 14 sections, 1 equation, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methods
  4. Contrastive Self-supervised Learning (CSSL) with Brain Network Transformer
  5. Graph Dilation and Shrinkage
  6. Experiments
  7. Experimental Settings
  8. Datasets
  9. Evaluation Metrics
  10. Implementation Details
  11. Results
  12. Performance Analysis
  13. Ablation Studies
  14. Conclusion

Figures (3)

  • Figure 1: Overview of our proposed contrastive self-supervised learning framework.
  • Figure 2: Graph dilation and shrinkage on correlation matrix $\mathbf{C}$ with nodes $n_i$.
  • Figure 3: ROC curves for various experiments on the variation of number of nodes selected for graph dilation/shrinkage as well as noise types and levels. The red curve shows our selected settings with state-of-the-art performance.