Network framework for PET tumor segmentation driven by geodesic image prior
10.3760/cma.j.cn321828-20240531-00186
- VernacularTitle:基于测地线图像先验驱动PET肿瘤分割的网络框架
- Author:
Lin YANG
1
;
Dan SHAO
;
Zhenxing HUANG
;
Dong LIANG
;
Hairong ZHENG
;
Zhanli HU
Author Information
1. 中国科学院深圳先进技术研究院医学人工智能研究中心，深圳 518055
- Publication Type:Journal Article
- Keywords:
PET imaging;
Tumor segmentation;
Image prior;
Geodesic distance;
Deep learning
- From:
Chinese Journal of Nuclear Medicine and Molecular Imaging
2025;45(4):234-239
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct a prior based on the inherent properties of PET to accurately segment the lesion areas.Methods:A network framework for PET tumor segmentation driven by geodesic priors was proposed (geodesic network for short). Specifically, partial differential equations were constructed to characterize the geodesic distances between different regions in PET images. Tumor marker points identified by CT labeling were used as the initial conditions for the equations. To enhance the contrast between areas of lung or breast tumors and normal tissues, a smooth Heaviside function was utilized to map the geodesic distances. The network framework adopted a dual-branch architecture, using geodesic priors to assist in PET image segmentation.Results:The proposed method achieved a Dice coefficient of 94.92% in lung cancer segmentation and 90.12% in breast cancer segmentation. With the addition of geodesic priors in the Unet, the Dice coefficient for breast cancer increased by 32.37% (from 42.50% to 74.87%).Conclusion:Geodesic priors can significantly improve segmentation outcomes and enhance the generalization capability of the network.
