IEEE Transactions on Medical Imaging 2021

DRONE

DRONE: Dual-Domain Residual-based Optimization NEtwork for Sparse-View CT Reconstruction

Weiwen Wu, Dianlin Hu, Chuang Niu, Hengyong Yu, Varut Vardhanabhuti, Ge Wang

Current or former lab member. Select a name to see all of their projects.

Method overview of DRONE
The DRONE architecture (Fig. 1 of the paper).

Abstract

Sparse-view CT remains difficult when only a few projections are available. DRONE addresses it with three modules. The embedding module extends the sparse sinogram and suppresses artifacts in the image domain. The refinement module recovers details in the residual image and data domains. The awareness module enforces consistency between measurements and images using compressed-sensing kernel awareness.

DRONE was trained, validated and tested on preclinical and clinical datasets, where it showed advantages in edge preservation, feature recovery and reconstruction accuracy.

What’s in the repository

  • A README describing the environment and the five folders of the release
  • The code, trained networks and test datasets themselves are downloaded from Zenodo

Citation

@article{wu2021drone,
  title   = {DRONE: Dual-Domain Residual-based Optimization NEtwork for Sparse-View CT Reconstruction},
  author  = {Wu, Weiwen and Hu, Dianlin and Niu, Chuang and Yu, Hengyong and Vardhanabhuti, Varut and Wang, Ge},
  journal = {IEEE Transactions on Medical Imaging},
  volume  = {40},
  number  = {11},
  pages   = {3002--3014},
  year    = {2021},
  doi     = {10.1109/TMI.2021.3078067}
}
Back to projects