DerMAE: Improving skin lesion classification through conditioned latent diffusion and MAE distillation
This work addresses class imbalance in dermatological datasets for clinical applications, but it is incremental as it combines existing methods like diffusion models and MAE.
The paper tackled class imbalance in skin lesion classification by generating synthetic images with diffusion models and using MAE pretraining to improve feature learning, then distilled knowledge to a smaller model for mobile deployment, achieving improved classification performance.
Skin lesion classification datasets often suffer from severe class imbalance, with malignant cases significantly underrepresented, leading to biased decision boundaries during deep learning training. We address this challenge using class-conditioned diffusion models to generate synthetic dermatological images, followed by self-supervised MAE pretraining to enable huge ViT models to learn robust, domain-relevant features. To support deployment in practical clinical settings, where lightweight models are required, we apply knowledge distillation to transfer these representations to a smaller ViT student suitable for mobile devices. Our results show that MAE pretraining on synthetic data, combined with distillation, improves classification performance while enabling efficient on-device inference for practical clinical use.