Towards Data-Efficient Medical Imaging: A Generative and Semi-Supervised Framework
This work addresses annotation bottlenecks to improve robustness in medical image analysis, but it is incremental as it augments existing baselines rather than introducing a fundamentally new approach.
The paper tackled the problem of scarce and imbalanced annotated data in medical imaging by proposing SSGNet, a framework that combines generative modeling with semi-supervised pseudo labeling to augment training data and refine labels, resulting in consistent gains in classification and segmentation performance across multiple benchmarks.
Deep learning in medical imaging is often limited by scarce and imbalanced annotated data. We present SSGNet, a unified framework that combines class specific generative modeling with iterative semisupervised pseudo labeling to enhance both classification and segmentation. Rather than functioning as a standalone model, SSGNet augments existing baselines by expanding training data with StyleGAN3 generated images and refining labels through iterative pseudo labeling. Experiments across multiple medical imaging benchmarks demonstrate consistent gains in classification and segmentation performance, while Frechet Inception Distance analysis confirms the high quality of generated samples. These results highlight SSGNet as a practical strategy to mitigate annotation bottlenecks and improve robustness in medical image analysis.