Robustifying Deep Networks for Medical Image Segmentation

被引:0
|
作者
Zheng Liu
Jinnian Zhang
Varun Jog
Po-Ling Loh
Alan B. McMillan
机构
[1] University of Wisconsin,Department of Electrical and Computer Engineering
[2] University of Wisconsin,Department of Statistics
[3] University of Wisconsin,Department of Radiology
[4] Centre for Mathematical Sciences,Department of Pure Mathematics and Mathematical Statistics
[5] University of Cambridge,Department of Medical Physics
[6] University of Wisconsin,undefined
来源
关键词
Deep learning segmentation; Robustness; Adversarial attacks; Defenses;
D O I
暂无
中图分类号
学科分类号
摘要
The purpose of this study is to investigate the robustness of a commonly used convolutional neural network for image segmentation with respect to nearly unnoticeable adversarial perturbations, and suggest new methods to make these networks more robust to such perturbations. In this retrospective study, the accuracy of brain tumor segmentation was studied in subjects with low- and high-grade gliomas. Two representative UNets were implemented to segment four different MR series (T1-weighted, post-contrast T1-weighted, T2-weighted, and T2-weighted FLAIR) into four pixelwise labels (Gd-enhancing tumor, peritumoral edema, necrotic and non-enhancing tumor, and background). We developed attack strategies based on the fast gradient sign method (FGSM), iterative FGSM (i-FGSM), and targeted iterative FGSM (ti-FGSM) to produce effective but imperceptible attacks. Additionally, we explored the effectiveness of distillation and adversarial training via data augmentation to counteract these adversarial attacks. Robustness was measured by comparing the Dice coefficients for the attacks using Wilcoxon signed-rank tests. The experimental results show that attacks based on FGSM, i-FGSM, and ti-FGSM were effective in reducing the quality of image segmentation by up to 65% in the Dice coefficient. For attack defenses, distillation performed significantly better than adversarial training approaches. However, all defense approaches performed worse compared to unperturbed test images. Therefore, segmentation networks can be adversely affected by targeted attacks that introduce visually minor (and potentially undetectable) modifications to existing images. With an increasing interest in applying deep learning techniques to medical imaging data, it is important to quantify the ramifications of adversarial inputs (either intentional or unintentional).
引用
收藏
页码:1279 / 1293
页数:14
相关论文
共 50 条
  • [1] Robustifying Deep Networks for Medical Image Segmentation
    Liu, Zheng
    Zhang, Jinnian
    Jog, Varun
    Loh, Po-Ling
    McMillan, Alan B.
    [J]. JOURNAL OF DIGITAL IMAGING, 2021, 34 (05) : 1279 - 1293
  • [2] Medical Image Segmentation Using Deep Neural Networks
    Aydin, Musa
    [J]. 29TH IEEE CONFERENCE ON SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS (SIU 2021), 2021,
  • [3] A Comprehensive Study on Medical Image Segmentation using Deep Neural Networks
    Dao, Loan
    Ly, Ngoc Quoc
    [J]. INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS, 2023, 14 (03) : 167 - 184
  • [4] RETRACTED: Deep Neural Networks for Medical Image Segmentation (Retracted Article)
    Malhotra, Priyanka
    Gupta, Sheifali
    Koundal, Deepika
    Zaguia, Atef
    Enbeyle, Wegayehu
    [J]. JOURNAL OF HEALTHCARE ENGINEERING, 2022, 2022
  • [5] POST TRAINING UNCERTAINTY CALIBRATION OF DEEP NETWORKS FOR MEDICAL IMAGE SEGMENTATION
    Rousseau, Axel-Jan
    Becker, Thijs
    Bertels, Jeroen
    Blaschko, Matthew B.
    Valkenborg, Dirk
    [J]. 2021 IEEE 18TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (ISBI), 2021, : 1052 - 1056
  • [6] MIScnn: a framework for medical image segmentation with convolutional neural networks and deep learning
    Mueller, Dominik
    Kramer, Frank
    [J]. BMC MEDICAL IMAGING, 2021, 21 (01)
  • [7] MIScnn: a framework for medical image segmentation with convolutional neural networks and deep learning
    Dominik Müller
    Frank Kramer
    [J]. BMC Medical Imaging, 21
  • [8] DEEP LEARNING ARCHITECTURES FOR MEDICAL IMAGE SEGMENTATION
    Subramaniam, Sudha
    Jayanthi, K. B.
    Rajasekaran, C.
    Kuchelar, Ramani
    [J]. 2020 IEEE 33RD INTERNATIONAL SYMPOSIUM ON COMPUTER-BASED MEDICAL SYSTEMS(CBMS 2020), 2020, : 579 - 584
  • [9] Medical Image Segmentation With Deep Atlas Prior
    Huang, Huimin
    Zheng, Han
    Lin, Lanfen
    Cai, Ming
    Hu, Hongjie
    Zhang, Qiaowei
    Chen, Qingqing
    Iwamoto, Yutaro
    Han, Xianhua
    Chen, Yen-Wei
    Tong, Ruofeng
    [J]. IEEE TRANSACTIONS ON MEDICAL IMAGING, 2021, 40 (12) : 3519 - 3530
  • [10] Sample pairing (mixup) as a data augmentation technique for deep medical image segmentation networks
    Isaksson, L. J.
    [J]. RADIOTHERAPY AND ONCOLOGY, 2021, 161 : S1415 - S1416