Access to large volumes of so-called whole-slide images-high-resolution scans of complete pathological slides-has become a cornerstone of the development of novel artificial intelligence methods in pathology for diagnostic use, education/training of pathologists, and research. Nevertheless, a methodology based on risk analysis for evaluating the privacy risks associated with sharing such imaging data and applying the principle "as open as possible and as closed as necessary" is still lacking. In this article, we develop a model for privacy risk analysis for whole-slide images which focuses primarily on identity disclosure attacks, as these are the most important from a regulatory perspective. We introduce a taxonomy of whole-slide images with respect to privacy risks and mathematical model for risk assessment and design . Based on this risk assessment model and the taxonomy, we conduct a series of experiments to demonstrate the risks using real-world imaging data. Finally, we develop guidelines for risk assessment and recommendations for low-risk sharing of whole-slide image data.

BBMRI at and Diagnostic and Research Center for Molecular BioMedicine Medical University of Graz Graz A 8010 Austria

BBMRI cz and Masaryk Memorial Cancer Institute Brno Czech Republic

BBMRI ERIC Graz Austria

Berlin Institute of Health Charité Universitätsmedizin Berlin Berlin Germany

Faculty of Informatics Masaryk University Brno Czech Republic

Institute of Computer Science Masaryk University Brno Czech Republic

TMF eV Berlin Germany

Visual and Data intensive Computing Group CRS4 Pula Italy

Zobrazit více v PubMed

Zelic R, et al. Interchangeability of light and virtual microscopy for histopathological evaluation of prostate cancer. Sci. Rep. 2021;11:3257. doi: 10.1038/s41598-021-82911-z. PubMed DOI PMC

Holzinger, A. et al. Machine Learning and Knowledge Extraction in Digital Pathology Needs an Integrative Approach in Towards Integrative Machine Learning and Knowledge Extraction (eds Holzinger, A., Goebel, R., Ferri, M. & Palade, V.) 13–50 (Springer International Publishing, Cham, 2017).

Artificial Intelligence and Machine Learning for Digital Pathology (eds Holzinger, A., Goebel, R., Mengel, M. & Müller, H.) (Springer Cham, 2020) 10.1007/978-3-030-50402-1.

BBMRI-ERIC Colorectal Cancer Cohort (CRC-Cohort). https://www.bbmri-eric.eu/scientific-collaboration/colorectal-cancer-cohort/.

Ehteshami Bejnordi B, et al. Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer. JAMA. 2017;318:2199–2210. doi: 10.1001/jama.2017.14585. PubMed DOI PMC

Bandi P, et al. From detection of individual metastases to classification of lymph node status at the patient level: the CAMELYON17 challenge. IEEE Trans. Med. Imaging. 2019;38:550–560. doi: 10.1109/TMI.2018.2867350. PubMed DOI

Gutman DA, et al. Cancer Digital Slide Archive: an informatics resource to support integrated in silico analysis of TCGA pathology data. J. Am. Med. Inform. Assoc. 2013;20:1091–1098. doi: 10.1136/amiajnl-2012-001469. PubMed DOI PMC

Carbone M, et al. BAP1 cancer syndrome: malignant mesothelioma, uveal and cutaneous melanoma, and MBAITs. J. Transl. Med. 2012;10:1–7. doi: 10.1186/1479-5876-10-179. PubMed DOI PMC

Foretová L, et al. BAP1 syndrome-predisposition to malignant mesothelioma, skin and uveal melanoma, renal and other cancers. Klin. Onkol.: Čas. České Slov. Onkol Společnosti. 2019;32:118–122. doi: 10.14735/amko2019S118. PubMed DOI

Pocevičiūtė, M., Eilertsen, G. & Lundström, C. Survey of XAI in Digital Pathology in Artificial Intelligence and Machine Learn-ing for Digital Pathology (eds Holzinger, A., Goebel, R., Mengel, M. & Müller, H.) 56­88 (Springer Cham, 2020).

Wulczyn, E. et al. Interpretable survival prediction for colorectal cancer using deep learning. npj Digital Med.4, 71 (2021). PubMed PMC

Wulczyn, E. et al. Predicting prostate cancer specific-mortality with artificial intelligence-based Gleason grading. Commun. Med.1, 10 (2021). PubMed PMC

Schaumberg, A., Rubin, M. & Fuchs, T. H&E-stained whole slide image deep learning predicts SPOP mutation state in prostate cancer. bioRxiv, 1–14 10.1101/064279 (2018).

Chang P, et al. Deep-learning convolutional neural networks accurately classify genetic mutations in gliomas. Am. J. Neuroradiol. 2018;39:1201–1207. doi: 10.3174/ajnr.A5667. PubMed DOI PMC

Xu, H. et al. Spatial heterogeneity and organization of tumor mutation burden with im-mune infiltrates within tumors based on whole slide images correlated with patient sur-vival in bladder cancer. Journal of Pathology Informatics13, 100105 (2022). PubMed PMC

Schmauch B, et al. A deep learning model to predict RNA-Seq expression of tumours from whole slide images. Nat. Commun. 2020;11:1–15. doi: 10.1038/s41467-020-17678-4. PubMed DOI PMC

Ohmann C, et al. Sharing and reuse of individual participant data from clinical trials: principles and recommendations. BMJ Open. 2017;7:e018647. doi: 10.1136/bmjopen-2017-018647. PubMed DOI PMC

El Emam, K. Guide to the de-identification of personal health information (CRC Press, 2013).

Rubinstein YR, et al. The case for open science: rare diseases. JAMIA Open. 2020;3:472–486. doi: 10.1093/jamiaopen/ooaa030. PubMed DOI PMC

Shokri, R., Stronati, M., Song, C. & Shmatikov, V. Membership inference attacks against machine learning models in IEEE symposium on security and privacy (SP) (2017), 3–18. 10.1109/SP.2017.41.

Papernot, N. et al. Scalable Private Learning with PATE 2018. arXiv: 1802.08908. 10.48550/arXiv.1802.08908.

Holub P, et al. Enhancing reuse of data and biological material in medical research: from FAIR to FAIR-health. Biopreserv Biobank. 2018;16:97–105. doi: 10.1089/bio.2017.0110. PubMed DOI PMC

Papernot, N. et al. The limitations of deep learning in adversarial settings in IEEE European symposium on security and privacy (EuroS&P) (2016), 372–387. 10.1109/EuroSP.2016.36.

Smith, G. On the Foundations of Quantitative Information Flow in Foundations of Software Science and Computational Structures (ed de Alfaro, L.) (Springer Berlin Heidelberg, 2009), 288­302.

Alvim, M. S. et al. The Science of Quantitative Information Flow (Springer, 2020) 10.1007/978-3-319-96131-6.

Malin B, Loukides G, Benitez K, Clayton EW. Identifiability in biobanks: models, measures, and mitigation strategies. Hum. Genet. 2011;130:383–392. doi: 10.1007/s00439-011-1042-5. PubMed DOI PMC

European Medicines Agency. External guidance on the implementation of the European Medicines Agency policy on the publication of clinical data for medicinal products for human use. https://www.ema.europa.eu/en/documents/regulatory-procedural-guideline/external-guidance-implementation-european-medicines-agency-policy-publication-clinical-data_en-1.pdf.

Feit J, Kempf W, Jedlicková H, Burg G. Hypertext atlas of dermatopathology with expert system for epithelial tumors of the skin. J. Cutan Pathol. 2005;32:433–437. doi: 10.1111/j.0303-6987.2005.00291.x. PubMed DOI

Ježová M, Múčková K, Souček O, Feit J, Vlašín P. Hypertext atlas of fetal and neonatal pathology. Diagn. Pathol. 2008;3:S9. doi: 10.1186/1746-1596-3-S1-S9. PubMed DOI PMC

Lianas, L. et al. Cy-TEST - A new platform for training and testing in cytopathology. Diagn. Pathol.2. https://www.diagnosticpathology.eu/content/index.php/dpath/article/view/185 (2016).

Veeling, B. S., Linmans, J., Winkens, J., Cohen, T. & Welling, M. Rotation equivariant –CNNs for digital pathology. arXiv: 1806.03962 [cs.CV] 10.48550/arXiv.1806.03962 (2018).

El Emam K. Risk-based de-identification of health data. IEEE Security Privacy. 2010;8:64–67. doi: 10.1109/MSP.2010.103. DOI

Deng, J. et al. ImageNet: A large-scale hierarchical image database in IEEE conference on computer vision and pattern recognition. 248–255 10.1109/CVPR.2009.5206848 (2009).

He, K., Zhang, X., Ren, S. & Sun, J.–Deep Residual Learning for Image Recognition in IEEE Conference on Computer Vision and Pattern Recognition (CVPR)45 (IEEE, June 2016), 770—778. 10.1109/CVPR.2016.90.

Simonyan, K. & Zisserman, A. Very deep convolutional networks for large-scale image recognition.rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings, 1–14 (2015).

Img2vec. https://pypi.org/project/img2vec-pytorch/.

Chen, T., Kornblith, S., Swersky, K., Norouzi, M. & Hinton, G. Big self-supervised models are strong semi-supervised learners. arXiv preprint arXiv:2006.1002910.48550/arXiv.2006.10029 (2020).

Krajňanský, V. et al. Shedding light on the black box of a neural network used to detect prostate cancer in whole slide images by occlusion-based explainability. bioRxiv. 10.1101/2022.03.31.486599 (2022). PubMed