Injecting and removing malignant features in mammography with CycleGAN: Investigation of an automated adversarial attack using neural networks

$\textbf{Purpose}$ To train a cycle-consistent generative adversarial network (CycleGAN) on mammographic data to inject or remove features of malignancy, and to determine whether these AI-mediated attacks can be detected by radiologists. $\textbf{Material and Methods}$ From the two publicly available datasets, BCDR and INbreast, we selected images from cancer patients and healthy controls. An internal dataset served as test data, withheld during training. We ran two experiments training CycleGAN on low and higher resolution images ($256 \times 256$ px and $512 \times 408$ px). Three radiologists read the images and rated the likelihood of malignancy on a scale from 1-5 and the likelihood of the image being manipulated. The readout was evaluated by ROC analysis (Area under the ROC curve = AUC). $\textbf{Results}$ At the lower resolution, only one radiologist exhibited markedly lower detection of cancer (AUC=0.85 vs 0.63, p=0.06), while the other two were unaffected (0.67 vs. 0.69 and 0.75 vs. 0.77, p=0.55). Only one radiologist could discriminate between original and modified images slightly better than guessing/chance (0.66, p=0.008). At the higher resolution, all radiologists showed significantly lower detection rate of cancer in the modified images (0.77-0.84 vs. 0.59-0.69, p=0.008), however, they were now able to reliably detect modified images due to better visibility of artifacts (0.92, 0.92 and 0.97). $\textbf{Conclusion}$ A CycleGAN can implicitly learn malignant features and inject or remove them so that a substantial proportion of small mammographic images would consequently be misdiagnosed. At higher resolutions, however, the method is currently limited and has a clear trade-off between manipulation of images and introduction of artifacts.

[1]  Feng Lin,et al.  Low-Dose CT With a Residual Encoder-Decoder Convolutional Neural Network , 2017, IEEE Transactions on Medical Imaging.

[2]  拓海 杉山,et al.  “Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告 , 2017 .

[3]  Hatem Alkadhi,et al.  Precise and Automatic Patient Positioning in Computed Tomography: Avatar Modeling of the Patient Surface Using a 3-Dimensional Camera , 2018, Investigative radiology.

[4]  Roger B. Myerson,et al.  Game theory - Analysis of Conflict , 1991 .

[5]  Leyla Bilge,et al.  Cutting the Gordian Knot: A Look Under the Hood of Ransomware Attacks , 2015, DIMVA.

[6]  Vincent Dumoulin,et al.  Deconvolution and Checkerboard Artifacts , 2016 .

[7]  E. Suchman,et al.  The American soldier: Adjustment during army life. (Studies in social psychology in World War II), Vol. 1 , 1949 .

[8]  Philipp A. Kaufmann,et al.  Automated detection of lung cancer at ultralow dose PET/CT by deep neural networks - Initial results. , 2018, Lung cancer.

[9]  E. DeLong,et al.  Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. , 1988, Biometrics.

[10]  Andrew Y. Ng,et al.  CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning , 2017, ArXiv.

[11]  Regina Barzilay,et al.  Using Machine Learning to Parse Breast Pathology Reports , 2016 .

[12]  Gunnar Rätsch,et al.  An Empirical Analysis of Topic Modeling for Mining Cancer Clinical Notes , 2013, bioRxiv.

[13]  Jaime S. Cardoso,et al.  INbreast: toward a full-field digital mammographic database. , 2012, Academic radiology.

[14]  Geraint Rees,et al.  Clinically applicable deep learning for diagnosis and referral in retinal disease , 2018, Nature Medicine.

[15]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[16]  A. Boss,et al.  Classification of breast cancer in ultrasound imaging using a generic deep learning analysis software: a pilot study. , 2017, The British journal of radiology.

[17]  Andrew L. Beam,et al.  Adversarial Attacks Against Medical Deep Learning Systems , 2018, ArXiv.

[18]  Miguel Ángel Guevara-López,et al.  An evaluation of image descriptors combined with clinical data for breast cancer diagnosis , 2013, International Journal of Computer Assisted Radiology and Surgery.

[19]  Yoshua Bengio,et al.  Generative Adversarial Networks , 2014, ArXiv.

[20]  Dorit Merhof,et al.  Radiomic versus Convolutional Neural Networks Analysis for Classification of Contrast-enhancing Lesions at Multiparametric Breast MRI. , 2019, Radiology.

[21]  Andy Kitchen,et al.  Chest Radiographs in Congestive Heart Failure: Visualizing Neural Network Learning. , 2019, Radiology.

[22]  Yuval Shahar,et al.  Know Your Enemy: Characteristics of Cyber-Attacks on Medical Imaging Devices , 2018, ArXiv.

[23]  Kouichi Sakurai,et al.  One Pixel Attack for Fooling Deep Neural Networks , 2017, IEEE Transactions on Evolutionary Computation.

[24]  Thomas Frauenfelder,et al.  Deep Learning in Mammography: Diagnostic Accuracy of a Multipurpose Image Analysis Software in the Detection of Breast Cancer , 2017, Investigative radiology.