Enhancing Intracranial Vessel Segmentation using Diffusion Models without Label for 3D Time-of-Flight Magnetic Resonance Angiography

Overview

This repository contains the code for Enhancing Intracranial Vessel Segmentation using Diffusion Models without Label for 3D Time-of-Flight Magnetic Resonance Angiography. The pipeline to find the threshold value and then get the threshold label is shown in the figure below:

Afterward, the diffusion model below can be trained and inferred using the threshold label.

Citation

@article{kim2025enhancing,
  title={Enhancing intracranial vessel segmentation using diffusion models without manual annotation for 3D Time-of-Flight Magnetic Resonance Angiography},
  author={Kim, Jonghun and Na, Inye and Chung, Jiwon and Song, Ha-Na and Kim, Kyungseo and Ju, Seongvin and Eun, Mi-Yeon and Seo, Woo-Keun and Park, Hyunjin},
  journal={Computerized Medical Imaging and Graphics},
  pages={102651},
  year={2025},
  publisher={Elsevier}
}