Defending Against Adversarial Attacks On Medical Imaging Ai System, Classification Or Detection?

Li, Xin; Deng, Pan; Zhu, Dongxiao

doi:10.1109/isbi48211.2021.9433761

Cited by 22 publications

(17 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In some studies, adversarial training improved DL model robustness for multiple medical imaging modalities such as lung CT and retinal optical coherence tomography. 40 , 42 - 44 By contrast, Hirano et al 45 found that adversarial training generally did not increase model robustness for classifying dermatoscopic images, optical coherence tomography images, and chest x-ray images. The difference in effectiveness of adversarial training can be attributed to differences in adversarial training protocols (eg, single-step v iterative approaches).…”

Section: Discussionmentioning

confidence: 98%

“…One reason for this behavior could be that medical images are highly standardized, and small adversarial perturbations dramatically distort their distribution in the latent feature space. 40 , 41 Another factor could be the overparameterization of DL models for medical image analysis, as sharp loss landscapes around medical images lead to higher adversarial vulnerability. 14 …”

Section: Discussionmentioning

confidence: 99%

“…One contributing factor is that medical images have fundamentally differently properties than nonmedical images. 14 , 40 Thus, adversarial defenses well suited for nonmedical images may not be generalizable to medical images.…”

Section: Discussionmentioning

confidence: 99%

See 2 more Smart Citations

Using Adversarial Images to Assess the Robustness of Deep Learning Models Trained on Diagnostic Images in Oncology

Joel

Umrao

Chang

et al. 2022

JCO Clinical Cancer Informatics

View full text Add to dashboard Cite

PURPOSE Deep learning (DL) models have rapidly become a popular and cost-effective tool for image classification within oncology. A major limitation of DL models is their vulnerability to adversarial images, manipulated input images designed to cause misclassifications by DL models. The purpose of the study is to investigate the robustness of DL models trained on diagnostic images using adversarial images and explore the utility of an iterative adversarial training approach to improve the robustness of DL models against adversarial images. METHODS We examined the impact of adversarial images on the classification accuracies of DL models trained to classify cancerous lesions across three common oncologic imaging modalities. The computed tomography (CT) model was trained to classify malignant lung nodules. The mammogram model was trained to classify malignant breast lesions. The magnetic resonance imaging (MRI) model was trained to classify brain metastases. RESULTS Oncologic images showed instability to small pixel-level changes. A pixel-level perturbation of 0.004 (for pixels normalized to the range between 0 and 1) resulted in most oncologic images to be misclassified (CT 25.6%, mammogram 23.9%, and MRI 6.4% accuracy). Adversarial training improved the stability and robustness of DL models trained on oncologic images compared with naive models ([CT 67.7% v 26.9%], mammogram [63.4% vs 27.7%], and MRI [87.2% vs 24.3%]). CONCLUSION DL models naively trained on oncologic images exhibited dramatic instability to small pixel-level changes resulting in substantial decreases in accuracy. Adversarial training techniques improved the stability and robustness of DL models to such pixel-level changes. Before clinical implementation, adversarial training should be considered to proposed DL models to improve overall performance and safety.

show abstract

Section: Discussionmentioning

confidence: 98%

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Using Adversarial Images to Assess the Robustness of Deep Learning Models Trained on Diagnostic Images in Oncology

Joel

Umrao

Chang

et al. 2022

JCO Clinical Cancer Informatics

View full text Add to dashboard Cite

show abstract

“…Defence Method. In [41], the authors perform both prevention and detection in their solution. For prevention, the DNN is trained on both normal and adversarial samples to make it more robust to attacks, an idea that has been shown to be effective in the past [8].…”

Section: Gmm Methodsmentioning

confidence: 99%

“…For binary classifiers we use accuracy and for multi-class classifiers we use average-accuracy. We compute GMM's accuracy using the inverse of the author's metric called 'adversarial risk' -a combined performance measure of the detector and the victim's classifier [41]. Additionally, for MGM [40] and GMM [41], which used an evaluation different from accuracy, we performed the exact same evaluation made in the original papers.…”

Section: Metricsmentioning

confidence: 99%

The Security of Deep Learning Defences for Medical Imaging

Levy¹,

Amit²,

Elovici³

et al. 2022

Preprint

View full text Add to dashboard Cite

Deep learning has shown great promise in the domain of medical image analysis. Medical professionals and healthcare providers have been adopting the technology to speed up and enhance their work. These systems use deep neural networks (DNN) which are vulnerable to adversarial samples; images with imperceivable changes that can alter the model's prediction. Researchers have proposed defences which either make a DNN more robust or detect the adversarial samples before they do harm. However, none of these works consider an informed attacker which can adapt to the defence mechanism. We show that an informed attacker can evade five of the current state of the art defences while successfully fooling the victim's deep learning model, rendering these defences useless. We then suggest better alternatives for securing healthcare DNNs from such attacks: (1) harden the system's security and (2) use digital signatures.

show abstract

How Effective is Adversarial Training of CNNs in Medical Image Analysis?

Xie

Fetit

2022

Medical Image Understanding and Analysis

View full text Add to dashboard Cite

Defending Against Adversarial Attacks On Medical Imaging Ai System, Classification Or Detection?

Cited by 22 publications

References 11 publications

Using Adversarial Images to Assess the Robustness of Deep Learning Models Trained on Diagnostic Images in Oncology

Using Adversarial Images to Assess the Robustness of Deep Learning Models Trained on Diagnostic Images in Oncology

The Security of Deep Learning Defences for Medical Imaging

How Effective is Adversarial Training of CNNs in Medical Image Analysis?

Contact Info

Product

Resources

About