Thoracic Lymph Node Segmentation in CT Imaging via Lymph Node Station Stratification and Size Encoding
Lists
Only you see this
Abstract
Visible lymph node (i.e., LN, short axis $$\ge 5$$ mm) assessment and delineation in thoracic computed tomography (CT) images is an indispensable step in radiology and oncology workflows. The high demanding of clinical expertise and prohibitive laboring cost motivate the automated approaches. Previous works focus on extracting effective LN imaging features and/or exploiting the anatomical priors to help LN segmentation. However, the performance in general is struggled with low recall/precision due to LN’s low contrast in CT and tumor-induced shape and size variations. Given that LNs reside inside the lymph node station (LN-station), it is intuitive to directly utilize the LN-station maps to guide LN segmentation. We propose a stratified LN-station and LN size encoded segmentation framework by casting thoracic LN-stations into three super lymph node stations and subsequently learning the station-specific LN size variations. Four-fold cross-validation experiments on the public NIH 89-patient dataset are conducted. Compared to previous leading works, our framework produces significant performance improvements, with an average $$74.2\%$$ ( $$9.9\%$$ increases) in Dice score and $$72.0\%$$ ( $$15.6\%$$ increases) in detection recall at 4.0 (1.9 reduces) false positives per patient. When directly tested on an external dataset of 57 esophageal cancer patients, the proposed framework demonstrates good generalizability and achieves $$70.4\%$$ in Dice score and $$70.2\%$$ in detection Recall at 4.4 false positives per patient.
BibTeX
@inproceedings{guo2022thoracic,
author = {Dazhou Guo and Jia Ge and Ke Yan and Puyang Wang and Zhuotun Zhu and Dandan Zheng and Xian-Sheng Hua and Le Lu and Tsung-Ying Ho and Xianghua Ye and Dakai Jin},
title = {{Thoracic Lymph Node Segmentation in CT Imaging via Lymph Node Station Stratification and Size Encoding}},
booktitle = {International Conference on Medical Image Computing and Computer-Assisted Intervention},
year = {2022},
doi = {10.1007/978-3-031-16443-9\_6},
}Loading related content...