Accurate Airway Tree Segmentation in CT Scans via Anatomy-aware Multi-class Segmentation and Topology-guided Iterative Learning
Puyang Wang,Dazhou Guo,Dandan Zheng,Minghui Zhang,Haogang Yu,Xin Sun,Jia Ge,Yun Gu,Le Lu,Xianghua Ye,Dakai Jin
2023-06-15
Abstract:Intrathoracic airway segmentation in computed tomography (CT) is a prerequisite for various respiratory disease analyses such as chronic obstructive pulmonary disease (COPD), asthma and lung cancer. Unlike other organs with simpler shapes or topology, the airway's complex tree structure imposes an unbearable burden to generate the "ground truth" label (up to 7 or 3 hours of manual or semi-automatic annotation on each case). Most of the existing airway datasets are incompletely labeled/annotated, thus limiting the completeness of computer-segmented airway. In this paper, we propose a new anatomy-aware multi-class airway segmentation method enhanced by topology-guided iterative self-learning. Based on the natural airway anatomy, we formulate a simple yet highly effective anatomy-aware multi-class segmentation task to intuitively handle the severe intra-class imbalance of the airway. To solve the incomplete labeling issue, we propose a tailored self-iterative learning scheme to segment toward the complete airway tree. For generating pseudo-labels to achieve higher sensitivity , we introduce a novel breakage attention map and design a topology-guided pseudo-label refinement method by iteratively connecting breaking branches commonly existed from initial pseudo-labels. Extensive experiments have been conducted on four datasets including two public challenges. The proposed method ranked 1st in both EXACT'09 challenge using average score and ATM'22 challenge on weighted average score. In a public BAS dataset and a private lung cancer dataset, our method significantly improves previous leading approaches by extracting at least (absolute) 7.5% more detected tree length and 4.0% more tree branches, while maintaining similar precision.
Image and Video Processing,Computer Vision and Pattern Recognition