Skip to content

T-SYNTH is a synthetic digital breast tomosynthesis (DBT) dataset with four breast fibroglandular density distributions imaged using Monte Carlo x-ray simulations with the publicly available Virtual Imaging Clinical Trial for Regulatory Evaluation (VICTRE) toolkit.

License

Notifications You must be signed in to change notification settings

DIDSR/tsynth-release

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 

Repository files navigation

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)

overview

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.

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

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

About

T-SYNTH is a synthetic digital breast tomosynthesis (DBT) dataset with four breast fibroglandular density distributions imaged using Monte Carlo x-ray simulations with the publicly available Virtual Imaging Clinical Trial for Regulatory Evaluation (VICTRE) toolkit.

Topics

Resources

License

Stars

Watchers

Forks