T-SYNTH: A Knowledge-Based Dataset of Synthetic Breast Images

This repository contains code used in the paper:

"T-SYNTH: A Knowledge-Based Dataset of Synthetic Breast Images"

Christopher Wiedeman*, Anastasiia Sarmakeeva*, Elena Sizikova, Daniil Filienko, Miguel Lago, Jana Delfino, Aldo Badano

(* - equal contribution)

International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) Open Data 2025

• Huggingface Data Repository: https://huggingface.co/datasets/didsr/tsynth

• Arxiv: https://arxiv.org/abs/2507.04038

• Poster: https://github.com/DIDSR/tsynth-release/blob/main/images/poster.pdf

The contributions of our work are:

• We release T-SYNTH, a public synthetic dataset of paired DM (2D imaging) and DBT (3D imaging) images derived from a KB model, with pixel-level segmentation and bounding boxes of a variety of breast tissues.
• We demonstrate how T-SYNTH can be used for subgroup analysis. Specifically, Faster-RCNN is trained for and evaluated for lesion detection in a balanced dataset; results reveal expected trends in subgroup performance in both DM and (C-View) DBT (e.g., less dense lesions are harder to detect).
• We train detection models on limited patient data in both DM and DBT (C-View), and show that augmenting training data with T-SYNTH can improve performance.

Citation

@article{t-synth,
  title={{T-SYNTH}: A Knowledge-Based Dataset of Synthetic Breast Images},
  author={Christopher Wiedeman, Anastasiia Sarmakeeva, Elena Sizikova, Daniil Filienko, Miguel Lago, Jana G. Delfino, Aldo Badano},
  journal={MICCAI Open Data},
  volume={},
  pages={},
  year={2025}
}