Dataset and codes for the paper "Leveraging Synthetic Degradation for Effective Training of Super-Resolution Models in Dermatological Images"

Dataset and codes for the work: "Leveraging Synthetic Degradation for Effective Training of Super-Resolution Models in Dermatological Images", Francesco Branciforti, Kristen M. Meiburger, and Massimo Salvi

Abstract:
Background: Teledermatology relies on digital transfer of dermatological images, but compression and resolution differences compromise diagnostic quality. Image en-hancement techniques are crucial to compensate for these differences and improve quality for both clinical assessment and AI-based analysis. Methods: We developed a customized image degradation pipeline simulating common artifacts in dermatological images, in-cluding blur, noise, downsampling, and compression. This synthetic degradation ap-proach enabled effective training of DermaSR-GAN, a super-resolution Generative Adversarial Network tailored for dermoscopic images. The model was trained on 30,000 high-quality ISIC images and evaluated on three independent datasets (ISIC Test, Novara Dermoscopic, PH2) using structural similarity and no-reference quality metrics. Results: DermaSR-GAN achieved statistically significant improvements in quality scores across all datasets, with up to 23% enhancement in perceptual quality metrics (MANIQA). The model preserved diagnostic details while doubling resolution and surpassed existing approaches including traditional interpolation methods and state-of-the-art deep learning techniques. Integration with downstream classification systems demonstrated up to 14.6% improvement in class-specific accuracy for keratosis-like lesions compared to original images. Conclusions: Synthetic degradation represents a promising approach for training effective super-resolution models in medical imaging, with significant potential for en-hancing teledermatology applications and computer-aided diagnosis systems.