Segmentation of coronary artery based on discriminative frequency learning and coronary-geometric refinement

Weili Jiang, Yiming Li, Yuheng Jia, Yuan Feng, Yi Zhang, Jianyong Wang, Mao Chen

Abstract: Highlights•A frequency module to select the informative components for segmentation.•A boundary consistency loss ensures accurate segmentation boundaries.•Coronary-geometric Refinement method refines segmentation via skeleton and radius.•Comprehensive experiments demonstrate the effectiveness of the proposed Ori-Net.