Volumetric Medical Image Segmentation: A 3D Deep Coarse-to-Fine Framework and Its Adversarial Examples

• Published in: Deep Learning and Convolutional Neural Networks for Medical Imaging and Clinical Informatics pp. 69 - 91
• Main Authors: Li, Yingwei, Zhu, Zhuotun, Zhou, Yuyin, Xia, Yingda, Shen, Wei, Fishman, Elliot K., Yuille, Alan L.
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2019; Springer International Publishing
• Series: Advances in Computer Vision and Pattern Recognition
• ISBN: 9783030139681; 3030139689
• ISSN: 2191-6586; 2191-6594
• DOI: 10.1007/978-3-030-13969-8_4

Although deep neural networks have been a dominant method for many 2D vision tasks, it is still challenging to apply them to 3D tasks, such as medical image Image segmentation, due to the limited amount of annotated 3D data and limited computational resources. In this chapter, by rethinking the strategy to apply 3D Convolutional Neural Convolutional Neural Networks (CNNs) to segment medical images, we propose a novel 3D-based coarse-to-Coarse-to-fine framework to efficiently tackle these challenges. The proposed 3D-based framework outperforms their 2D counterparts by a large margin since it can leverage the rich spatial information along all three axes. We further analyze the threat of adversarial Adversarial attack on the proposed framework and show how to defend against the attack. We conduct experiments on three datasets, the NIH pancreas dataset, the JHMI pancreas dataset and the JHMI pathological cyst dataset, where the first two and the last one contain healthy and pathological pancreases, respectively, and achieve the current state of the art in terms of Dice-Sørensen Coefficient (DSC) on all of them. Especially, on the NIH pancreas dataset, we outperform the previous best by an average of over \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2\%$$\end{document}, and the worst case is improved by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$7\%$$\end{document} to reach almost \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$70\%$$\end{document}, which indicates the reliability of our framework in clinical applications.