hcarrion/glomangioma

cgDDI: Controllable Generation of Diverse Dermatological Imagery

This repository contains a disease-conditioned LoRA checkpoint from cgDDI (Controllable Generation of Diverse Dermatological Imagery), a hybrid framework designed for fair and efficient malignancy classification.

For more details, please refer to the paper: Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification.

Resources

• Repository: GitHub - ControllableGenDDI
• Dataset: 🤗 hcarrion/ControllableGenDDI Dataset

Method Description

cgDDI is a hybrid framework designed to address the systematic lack of expertly annotated images in dermatology, especially for underrepresented skin tones and rare diseases. It operates by:

1. Synthesizing realistic healthy skin samples without disturbing other input properties.
2. Mapping single-sample rare lesions onto novel skin-tones and locations non-parametrically.
3. Allowing for efficient parametric generation with as few as 10 training samples via Textual Inversion and LoRA adapters.

For more details on the training and generation pipeline, please refer to the official GitHub repository.

Citation

@inproceedings{carrion2026cgddi,
  title     = {Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification},
  author    = {Carri{\'o}n, H{\'e}ctor and Norouzi, Narges},
  booktitle = {Medical Image Computing and Computer-Assisted Intervention (MICCAI)},
  year      = {2026},
  publisher = {Springer},
  series    = {Lecture Notes in Computer Science}
}