Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification

2026-07-14

AI Summary

This study addresses the significant performance disparities of existing dermatological malignancy classification models across diverse skin tones and rare disease populations, primarily caused by the scarcity of expert-annotated, heterogeneous medical images. To overcome this limitation, the authors propose the cgDDI framework, which for the first time enables non-parametric, single-shot lesion transfer across skin tones without requiring pre-annotated masks. The approach integrates parametric few-shot generation with controllable image synthesis, efficiently scaling datasets by over 400× using only minimal real samples. Evaluated on the DDI benchmark, the method achieves 86.4% accuracy using solely synthetic data and reaches a state-of-the-art 90.9% after fine-tuning. It also improves cross-dataset accuracy on F17k by 13.9%, substantially enhancing model fairness and generalization. The authors publicly release over 266,000 synthetic images along with their models.

Abstract

Accurate dermatological diagnosis naturally necessitates equitable performance across diverse populations, yet a systematic lack of expertly annotated images, especially for underrepresented skin tones and rare diseases, impedes progress toward measurably fair methods. We introduce cgDDI (Controllable Generation of Diverse Dermatological Imagery), a hybrid framework that (1) synthesizes realistic healthy skin samples without disturbing other input properties, (2) maps single-sample rare lesions onto novel skin-tones and locations non-parametrically, and (3) allows for efficient parametric generation with as few as 10 training samples. The framework supports both human and automated segmentation masking, enabling scalability to datasets without pre-made lesion masks. We grow a 656-image dataset by more than 400x and validate across two datasets: biopsy-confirmed Diverse Dermatology Images (DDI) and expert-verified Fitzpatrick17k (F17k). On the DDI benchmark, we achieve malignancy classification accuracy of 86.4% under synthetic-only training and 90.9% state-of-the-art performance with real data fine-tuning, alongside leading fairness metrics. Cross-dataset experiments show +13.9% accuracy improvements on unseen F17k data despite minimal disease overlap. We openly release 266k+ synthetic images, code, and generative models to further support fairness research at https://github.com/hectorcarrion/ControllableGenDDI.

Problem

• fairness
• malignancy classification
• dermatological imagery
• data scarcity
• underrepresented populations

Innovation

• controllable generation
• dermatological imagery
• fairness in AI
• few-shot synthesis
• non-parametric mapping