GeMix: Conditional GAN-Based Mixup for Improved Medical Image Augmentation
This work addresses the need for more realistic image augmentation in high-stakes medical applications like COVID-19 detection, offering an incremental improvement over existing mixup techniques.
The authors tackled the problem of unrealistic images from naive pixel-wise mixup in medical image augmentation by proposing GeMix, a conditional GAN-based method that synthesizes visually coherent images along a class manifold, resulting in increased macro-F1 scores and reduced false negative rates for COVID-19 detection on the COVIDx-CT-3 dataset.
Mixup has become a popular augmentation strategy for image classification, yet its naive pixel-wise interpolation often produces unrealistic images that can hinder learning, particularly in high-stakes medical applications. We propose GeMix, a two-stage framework that replaces heuristic blending with a learned, label-aware interpolation powered by class-conditional GANs. First, a StyleGAN2-ADA generator is trained on the target dataset. During augmentation, we sample two label vectors from Dirichlet priors biased toward different classes and blend them via a Beta-distributed coefficient. Then, we condition the generator on this soft label to synthesize visually coherent images that lie along a continuous class manifold. We benchmark GeMix on the large-scale COVIDx-CT-3 dataset using three backbones (ResNet-50, ResNet-101, EfficientNet-B0). When combined with real data, our method increases macro-F1 over traditional mixup for all backbones, reducing the false negative rate for COVID-19 detection. GeMix is thus a drop-in replacement for pixel-space mixup, delivering stronger regularization and greater semantic fidelity, without disrupting existing training pipelines. We publicly release our code at https://github.com/hugocarlesso/GeMix to foster reproducibility and further research.