Topology-Preserving Hard Pixel Mining for Tubular Structure Segmentation
Guoqing Zhang (Tsinghua-Berkeley Shenzhen Institute, Tsinghua University),* Caixia Dong (The Second Affilated Hospital of Xi'an Jiaotong University), Yang Li (Tsinghua-Berkeley Shenzhen Institute, Tsinghua University)
The 34th British Machine Vision Conference
Abstract
Maintaining accurate topology in image segmentation, especially for tubular structures such as borders of neurons, vessels or roads, is essential in many research fields but remains unsolved. Existing neural segmentation networks are prone to topological errors on these fine-scale structures. We propose a skeleton-based hard pixel mining strategy and incorporate it into training end-to-end segmentation networks. In particular, we penalize the most topology-relevant yet mis-segmented pixels in overall cost function. We mine skeletal hard pixels through several simple logical and morphological operations which is fast and scalable. We evaluate our method on six public datasets of different domains, including neurons, roads and coronary arteries. Our approach demonstrates training speed comparable to the base model, while achieving a significant reduction in the number of erroneous connected components by 17.14% and loops by 53.87%, respectively. Extensive experimental results demonstrate that our method outperforms state-of-the-art computational topology-based method in topological preservation while maintaining competitive volumetric accuracy.Video
Citation
@inproceedings{Zhang_2023_BMVC,
author = {Guoqing Zhang and Caixia Dong and Yang Li},
title = {Topology-Preserving Hard Pixel Mining for Tubular Structure Segmentation},
booktitle = {34th British Machine Vision Conference 2023, {BMVC} 2023, Aberdeen, UK, November 20-24, 2023},
publisher = {BMVA},
year = {2023},
url = {https://papers.bmvc2023.org/0846.pdf}
}
Copyright © 2023
The British Machine Vision Conference is organised by