Counterfactual brain network augmentation improves autism diagnosis accuracy
CFBrainAug: Modularity-Induced Counterfactual Brain Network Augmentation for Interpretable fMRI-based Autism Diagnosis
Machine Learning
Summary
Diagnosing autism using brain scans is tricky because machine learning models often struggle with limited data and offer explanations that doctors find hard to understand. The authors present a new method called CFBrainAug, which creates altered brain networks that mimic the opposite diagnosis to help the model learn better and explain its decisions more clearly. This approach helps identify key brain connections linked to autism and improves diagnosis accuracy on a standard dataset. Their method uses less complex models but still performs well by learning from these ‘counterfactual’ examples.
What this means in practice
- •For clinical neuroimaging teams: Enhance diagnostic tools for autism by integrating counterfactual brain network augmentation to improve accuracy and interpretability on small patient datasets.
- •For medical image analysis developers: Develop lightweight brain network classifiers that leverage counterfactual augmented data to compete with complex graph deep learning models efficiently.
Tested on one dataset.
Authors
Yifan Qiao, Sheng Li
Abstract
Machine learning on resting-state functional magnetic resonance imaging (rs-fMRI) brain networks is promising for diagnosing Autism Spectrum Disorder (ASD), but existing methods are prone to overfitting on limited training data, and their post-hoc statistical attribution offers limited clinical interpretability. We propose CFBrainAug, a counterfactual brain network learning framework that unifies data augmentation and counterfactual explanation by learning label-altered brain networks. Unlike conventional label-preserving augmentation, CFBrainAug first retrieves a label-opposite yet topologically similar candidate and then refines it under a module-number constraint. Jointly training on factual and counterfactual networks enriches the training distribution around the decision boundary, enabling a lightweight feature-selection-based SVM to perform competitively with recent graph deep learning methods. Meanwhile, the counterfactuals provide intrinsic explanations by identifying the connectivity differences between each subject and its counterfactual. Experiments on the ABIDE I dataset show that CFBrainAug reaches an accuracy of 74.1% on the NYU site, suggesting that counterfactual augmentation can complement more complex models in small-sample brain network analysis.