Evaluating Adversarial Robustness of Low dose CT Recovery

Kanchana Vaishnavi Gandikota, Paramanand Chandramouli, Hannah Dröge, Michael Moeller
Medical Imaging with Deep Learning, PMLR 227:1545-1563, 2024.

Abstract

Low dose computer tomography (CT) acquisition using reduced radiation or sparse angle measurements is recommended to decrease the harmful effects of X-ray radiation. Recent works successfully apply deep networks to the problem of low dose CT recovery on benchmark datasets. However, their robustness needs a thorough evaluation before use in clinical settings. In this work, we evaluate the robustness of different deep learning approaches and classical methods for CT recovery.We show that deep networks, including model based networks encouraging data consistency are more susceptible to untargeted attacks. Surprisingly, we observe that data consistency is not heavily affected even for these poor quality reconstructions, motivating the need for better regularization for the networks. We demonstrate the feasibility of universal attacks and study attack transferability across different methods. We analyze robustness to attacks causing localized changes in clinically relevant regions. Both classical approaches and deep networks are affected by such attacks leading to change in visual appearance of localized lesions, for extremely small perturbations. As the resulting reconstructions have high data consistency with original measurements, these localized attacks can be used to explore the solution space of CT recovery problem.

Cite this Paper

BibTeX
@InProceedings{pmlr-v227-gandikota24a, title = {Evaluating Adversarial Robustness of Low dose CT Recovery}, author = {Gandikota, Kanchana Vaishnavi and Chandramouli, Paramanand and Dr\"oge, Hannah and Moeller, Michael}, booktitle = {Medical Imaging with Deep Learning}, pages = {1545--1563}, year = {2024}, editor = {Oguz, Ipek and Noble, Jack and Li, Xiaoxiao and Styner, Martin and Baumgartner, Christian and Rusu, Mirabela and Heinmann, Tobias and Kontos, Despina and Landman, Bennett and Dawant, Benoit}, volume = {227}, series = {Proceedings of Machine Learning Research}, month = {10--12 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v227/gandikota24a/gandikota24a.pdf}, url = {https://proceedings.mlr.press/v227/gandikota24a.html}, abstract = {Low dose computer tomography (CT) acquisition using reduced radiation or sparse angle measurements is recommended to decrease the harmful effects of X-ray radiation. Recent works successfully apply deep networks to the problem of low dose CT recovery on benchmark datasets. However, their robustness needs a thorough evaluation before use in clinical settings. In this work, we evaluate the robustness of different deep learning approaches and classical methods for CT recovery.We show that deep networks, including model based networks encouraging data consistency are more susceptible to untargeted attacks. Surprisingly, we observe that data consistency is not heavily affected even for these poor quality reconstructions, motivating the need for better regularization for the networks. We demonstrate the feasibility of universal attacks and study attack transferability across different methods. We analyze robustness to attacks causing localized changes in clinically relevant regions. Both classical approaches and deep networks are affected by such attacks leading to change in visual appearance of localized lesions, for extremely small perturbations. As the resulting reconstructions have high data consistency with original measurements, these localized attacks can be used to explore the solution space of CT recovery problem.} }
Endnote
%0 Conference Paper %T Evaluating Adversarial Robustness of Low dose CT Recovery %A Kanchana Vaishnavi Gandikota %A Paramanand Chandramouli %A Hannah Dröge %A Michael Moeller %B Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2024 %E Ipek Oguz %E Jack Noble %E Xiaoxiao Li %E Martin Styner %E Christian Baumgartner %E Mirabela Rusu %E Tobias Heinmann %E Despina Kontos %E Bennett Landman %E Benoit Dawant %F pmlr-v227-gandikota24a %I PMLR %P 1545--1563 %U https://proceedings.mlr.press/v227/gandikota24a.html %V 227 %X Low dose computer tomography (CT) acquisition using reduced radiation or sparse angle measurements is recommended to decrease the harmful effects of X-ray radiation. Recent works successfully apply deep networks to the problem of low dose CT recovery on benchmark datasets. However, their robustness needs a thorough evaluation before use in clinical settings. In this work, we evaluate the robustness of different deep learning approaches and classical methods for CT recovery.We show that deep networks, including model based networks encouraging data consistency are more susceptible to untargeted attacks. Surprisingly, we observe that data consistency is not heavily affected even for these poor quality reconstructions, motivating the need for better regularization for the networks. We demonstrate the feasibility of universal attacks and study attack transferability across different methods. We analyze robustness to attacks causing localized changes in clinically relevant regions. Both classical approaches and deep networks are affected by such attacks leading to change in visual appearance of localized lesions, for extremely small perturbations. As the resulting reconstructions have high data consistency with original measurements, these localized attacks can be used to explore the solution space of CT recovery problem.
APA
Gandikota, K.V., Chandramouli, P., Dröge, H. & Moeller, M.. (2024). Evaluating Adversarial Robustness of Low dose CT Recovery. Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 227:1545-1563 Available from https://proceedings.mlr.press/v227/gandikota24a.html.

Related Material