csuzbw/fancnn

Code accompanying the paper 'Automatic Coronary Artery Plaque Quantification and CAD-RADS Prediction using Mesh Priors' (IEEE-TMI)

FanCNN for plaque quantification

Code for the paper Automatic Coronary Artery Plaque Quantification and CAD-RADS Prediction using Mesh Priors, van Herten et al. 2023, IEEE Transactions on Medical Imaging (IEE-TMI) (link to the paper, arXiv).

This code uses coronary artery centerline priors to create mesh segmentations of the coronary artery lumen, as well as calcified and non-calcified plaque. The mesh segmentations are used to compute plaque volumes and to predict the CAD-RADS score.

Optimization

The method optimizes the locations of vertices describing the mesh segmentations for the lumen, calcified and non-calcified plaque through a 3D-CNN operating on polar-transformed multi-planar reformatted images. Meshes are subsequently used to generate signals for lumen and plaque area measures along the coronary artery centerline, which are processed by a 1D-CNN to predict the CAD-RADS score.

Reference

If you use this code, please cite the accompanying paper:

@article{vanherten2023automatic,
  title={Automatic Coronary Artery Plaque Quantification and CAD-RADS Prediction using Mesh Priors},
  author={Van Herten, Rudolf LM and Hampe, Nils and Takx, Richard AP and Franssen, Klaas Jan and Wang, Yining and Such{\'a}, Dominika and Henriques, Jos{\'e} P and Leiner, Tim and Planken, R Nils and I{\v{s}}gum, Ivana},
  journal={IEEE Transactions on Medical Imaging},
  year={2023},
  publisher={IEEE}
}