GATE: Graph CCA for Temporal Self-Supervised Learning for Label-Efficient fMRI Analysis.

Researchers developed a new computer program called GATE that can better identify autism and dementia by analyzing brain scans (fMRI). The program learns patterns from brain scans without needing many examples that doctors have already diagnosed. When tested, it performed better than existing methods at correctly identifying these conditions from brain imaging data.

This study introduces GATE (Graph CCA for Temporal Self-supervised learning), a novel machine learning framework for analyzing fMRI brain scans to diagnose neurological conditions including autism. The method uses graph convolutional neural networks with self-supervised learning to improve classification accuracy when labeled training data is limited. The approach first learns patterns from unlabeled fMRI data, then fine-tunes on smaller labeled datasets. Testing on two independent datasets demonstrated superior performance for autism and dementia diagnosis compared to existing methods.

The framework addresses key challenges in medical imaging where obtaining large amounts of labeled data is expensive and time-consuming.

This computational approach could potentially support autism diagnosis through brain imaging analysis. However, clinical validation and integration with existing diagnostic workflows would be necessary before practical implementation. The method's ability to work with limited data could make it more feasible for clinical settings.

Sample sizes not reported. Only tested on two datasets. No details provided about clinical validation, comparison with standard diagnostic methods, or real-world implementation considerations.

In this work, we focus on the challenging task, neuro-disease classification, using functional magnetic resonance imaging (fMRI). In population graph-based disease analysis, graph convolutional neural networks (GCNs) have achieved remarkable success. However, these achievements are inseparable from abundant labeled data and sensitive to spurious signals. To improve fMRI representation learning and classification under a label-efficient setting, we propose a novel and theory-driven self-supervised learning (SSL) framework on GCNs, namely Graph CCA for Temporal sElf-supervised learning on fMRI analysis (GATE).

Concretely, it is demanding to design a suitable and effective SSL strategy to extract formation and robust features for fMRI. To this end, we investigate several new graph augmentation strategies from fMRI dynamic functional connectives (FC) for SSL training. Further, we leverage canonical-correlation analysis (CCA) on different temporal embeddings and present the theoretical implications. Consequently, this yields a novel two-step GCN learning procedure comprised of (i) SSL on an unlabeled fMRI population graph and (ii) fine-tuning on a small labeled fMRI dataset for a classification task.

Our method is tested on two independent fMRI datasets, demonstrating superior performance on autism and dementia diagnosis. Our code is available at https://github.com/LarryUESTC/GATE.