This repository contains the textual inversion and fine-tuning weights for cutaneous T-cell lymphoma, trained on top of stabilityai/stable-diffusion-2-1-base as part of the cgDDI (Controllable Generation of Diverse Dermatological Imagery) framework.

Paper: Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification (MICCAI 2026)
GitHub Repository: hectorcarrion/ControllableGenDDI
Full Dataset: hcarrion/ControllableGenDDI

Model Description

This model uses textual inversion to learn the disease-specific concept <cutaneous_T-cell_lymphoma-class>. It can be used to generate diverse, controllable, and skin-tone-balanced synthetic imagery of cutaneous T-cell lymphoma to support fair machine learning classification in dermatology.

Usage

You can load and run this model using the diffusers library:

python

import torch
from diffusers import StableDiffusionPipeline

model_id = "hcarrion/cutaneous_T-cell_lymphoma"
pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
pipe = pipe.to("cuda")

prompt = "a photo of a skin lesion of <cutaneous_T-cell_lymphoma-class> on dark skin"
image = pipe(prompt).images[0]
image.save("synthetic_lesion.png")

Citation

If you use this model or the cgDDI framework in your research, please cite:

bibtex

@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}
}