Literature DB >> 33634027

Anterior Mediastinal Lesion Segmentation Based on Two-Stage 3D ResUNet With Attention Gates and Lung Segmentation.

Su Huang1, Xiaowei Han2, Jingfan Fan1, Jing Chen1, Lei Du3, Wenwen Gao3, Bing Liu3, Yue Chen3, Xiuxiu Liu3, Yige Wang3, Danni Ai1, Guolin Ma3, Jian Yang1.   

Abstract

OBJECTIVES: Anterior mediastinal disease is a common disease in the chest. Computed tomography (CT), as an important imaging technology, is widely used in the diagnosis of mediastinal diseases. Doctors find it difficult to distinguish lesions in CT images because of image artifact, intensity inhomogeneity, and their similarity with other tissues. Direct segmentation of lesions can provide doctors a method to better subtract the features of the lesions, thereby improving the accuracy of diagnosis.
METHOD: As the trend of image processing technology, deep learning is more accurate in image segmentation than traditional methods. We employ a two-stage 3D ResUNet network combined with lung segmentation to segment CT images. Given that the mediastinum is between the two lungs, the original image is clipped through the lung mask to remove some noises that may affect the segmentation of the lesion. To capture the feature of the lesions, we design a two-stage network structure. In the first stage, the features of the lesion are learned from the low-resolution downsampled image, and the segmentation results under a rough scale are obtained. The results are concatenated with the original image and encoded into the second stage to capture more accurate segmentation information from the image. In addition, attention gates are introduced in the upsampling of the network, and these gates can focus on the lesion and play a role in filtering the features. The proposed method has achieved good results in the segmentation of the anterior mediastinal.
RESULTS: The proposed method was verified on 230 patients, and the anterior mediastinal lesions were well segmented. The average Dice coefficient reached 87.73%. Compared with the model without lung segmentation, the model with lung segmentation greatly improved the accuracy of lesion segmentation by approximately 9%. The addition of attention gates slightly improved the segmentation accuracy.
CONCLUSION: The proposed automatic segmentation method has achieved good results in clinical data. In clinical application, automatic segmentation of lesions can assist doctors in the diagnosis of diseases and may facilitate the automated diagnosis of illnesses in the future.
Copyright © 2021 Huang, Han, Fan, Chen, Du, Gao, Liu, Chen, Liu, Wang, Ai, Ma and Yang.

Entities:  

Keywords:  anterior mediastinal lesion segmentation; attention gates; deep learning; lung segmentation model; two-stage 3D ResUNet

Year:  2021        PMID: 33634027      PMCID: PMC7901488          DOI: 10.3389/fonc.2020.618357

Source DB:  PubMed          Journal:  Front Oncol        ISSN: 2234-943X            Impact factor:   6.244


  16 in total

Review 1.  Reconciliation of the anatomic, surgical, and radiographic classifications of the mediastinum.

Authors:  S L Aquino; G Duncan; K H Taber; A Sharma; L A Hayman
Journal:  J Comput Assist Tomogr       Date:  2001 May-Jun       Impact factor: 1.826

2.  T-snakes: topology adaptive snakes.

Authors:  T McInerney; D Terzopoulos
Journal:  Med Image Anal       Date:  2000-06       Impact factor: 8.545

3.  Reliability of tumor segmentation in glioblastoma: Impact on the robustness of MRI-radiomic features.

Authors:  Florent Tixier; Hyemin Um; Robert J Young; Harini Veeraraghavan
Journal:  Med Phys       Date:  2019-07-05       Impact factor: 4.071

4.  Automatic image segmentation by dynamic region growth and multiresolution merging.

Authors:  Luis Garcia Ugarriza; Eli Saber; Sreenath Rao Vantaram; Vincent Amuso; Mark Shaw; Ranjit Bhaskar
Journal:  IEEE Trans Image Process       Date:  2009-06-16       Impact factor: 10.856

5.  Tumor recurrence and survival in patients treated for thymomas and thymic squamous cell carcinomas: a retrospective analysis.

Authors:  Philipp Ströbel; Andrea Bauer; Bernhard Puppe; Til Kraushaar; Axel Krein; Klaus Toyka; Ralf Gold; Michael Semik; Reinhard Kiefer; Wilfred Nix; Berthold Schalke; Hans Konrad Müller-Hermelink; Alexander Marx
Journal:  J Clin Oncol       Date:  2004-04-15       Impact factor: 44.544

Review 6.  Thymic tumors.

Authors:  Frank C Detterbeck; Alden M Parsons
Journal:  Ann Thorac Surg       Date:  2004-05       Impact factor: 4.330

7.  Joint registration and segmentation of serial lung CT images for image-guided lung cancer diagnosis and therapy.

Authors:  Zhong Xue; Kelvin Wong; Stephen T C Wong
Journal:  Comput Med Imaging Graph       Date:  2009-08-25       Impact factor: 4.790

8.  Preoperative CT findings of thymoma are correlated with postoperative Masaoka clinical stage.

Authors:  Yan-Juan Qu; Guo-Bing Liu; He-Shui Shi; Mei-Yan Liao; Gui-Fang Yang; Zhi-Xiong Tian
Journal:  Acad Radiol       Date:  2012-09-14       Impact factor: 3.173

9.  CT-based radiomics features analysis for predicting the risk of anterior mediastinal lesions.

Authors:  He Sui; Lin Liu; Xuejia Li; Panli Zuo; Jingjing Cui; Zhanhao Mo
Journal:  J Thorac Dis       Date:  2019-05       Impact factor: 2.895

10.  Automatic Multi-Organ Segmentation on Abdominal CT With Dense V-Networks.

Authors:  Eli Gibson; Francesco Giganti; Yipeng Hu; Ester Bonmati; Steve Bandula; Kurinchi Gurusamy; Brian Davidson; Stephen P Pereira; Matthew J Clarkson; Dean C Barratt
Journal:  IEEE Trans Med Imaging       Date:  2018-02-14       Impact factor: 10.048

View more

北京卡尤迪生物科技股份有限公司 © 2022-2023.