CoChat
Sign Up
Cited by 43Privacy-Preserving Technologies in DataMedical Image Segmentation TechniquesArtificial Intelligence in Healthcare and Education

Privacy-Net: An Adversarial Approach for Identity-Obfuscated Segmentation of Medical Images

Open access
Kim, Bach Ngoc, Dolz, Jose, Jodoin, Pierre-Marc, Desrosiers, Christian·2021·IEEE Transactions on Medical Imaging
DOI 10.1109/tmi.2021.3065727
Open this paper in CoChat — free accountOpen PDF

Abstract

This paper presents a client/server privacy-preserving network in the context of multicentric medical image analysis. Our approach is based on adversarial learning which encodes images to obfuscate the patient identity while preserving enough information for a target task. Our novel architecture is composed of three components: 1) an encoder network which removes identity-specific features from input medical images, 2) a discriminator network that attempts to identify the subject from the encoded images, 3) a medical image analysis network which analyzes the content of the encoded images (segmentation in our case). By simultaneously fooling the discriminator and optimizing the medical analysis network, the encoder learns to remove privacy-specific features while keeping those essentials for the target task. Our approach is illustrated on the problem of segmenting brain MRI from the large-scale Parkinson Progression Marker Initiative (PPMI) dataset. Using longitudinal data from PPMI, we show that the discriminator learns to heavily distort input images while allowing for highly accurate segmentation results. Our results also demonstrate that an encoder trained on the PPMI dataset can be used for segmenting other datasets, without the need for retraining. The code is made available at: https://github.com/bachkimn/Privacy-Net-An-Adversarial-Approach-forIdentity-Obfuscated-Segmentation-of-MedicalImages.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1109/tmi.2021.3065727

Cites

5 of 5 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Deidentified medical images could be reidentified in medical image analysis because of the massive amount of image data used in training.

    “One example is medical image analysis, as deidentified medical images could be reidentified in medical image analysis because of the massive amount of image data used in training [39].”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [39]

    The Introduction says that removing patient-related image tags does not prevent identification: the raw image content can reveal a person’s identity. The paper attributes that risk to information in the images, not to the massive amount of image data used in training; that causal explanation is unsupported. (LLM, conf=0.88)

    Experimental — accuracy not yet measured

  • The identity-obfuscated segmentation framework integrates an encoder, a binary classifier, and a segmentation analysis network to safeguard medical data privacy while preserving the integrity and efficacy of medical image segmentation.

    “This framework integrates an encoder, a binary classifier, and a segmentation analysis network to form a robust approach to safeguard medical data privacy while preserving the integrity and efficacy of medical image segmentation.”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [39]

  • An innovative adversarial training method focused on identity-obfuscated segmentation has been proposed to address exposure of individual identities.

    “An innovative adversarial training method focused on identity-obfuscated segmentation has been proposed to address this challenge [39].”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [39]

  • Noise-to-image and image-to-image GANs have been used to synthesize realistic training magnetic resonance imaging images to boost the performance of convolutional neural networks for image diagnostic AI.

    “Noise-to-image and image-to-image GANs have been used to synthesize realistic training magnetic resonance imaging images to boost the performance of convolutional neural networks for image diagnostic AI [39,72].”

    From Generative AI in medical practice: in-depth exploration of privacy and security challenges (2024) · cites this paper as [39]

    The Introduction describes an adversarial encoder that obfuscates patient identity in MRI images while preserving information for a segmentation CNN. The Related Works section discusses adversarial learning for privacy, and the robustness analysis concerns distorted encoded images. Searches for noise-to-image and image-to-image GANs, synthetic MRI training images, and improved diagnostic performance found no passage addressing the attributed claim. (LLM, conf=0.76)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 52

  1. S. Zhou, H. Greenspan, and D. Shen, Deep Learning for Medical Image Analysis. Elsevier Science, 2017. 1
  2. G. Litjens et al., "A survey on deep learning in medical image analysis," MedIA, vol. 42, pp. 60-88, 2017. 1
  3. K. Kumar, M. Toews, L. Chauvin, O. Colliot, and C. Desrosiers, "Multi-modal brain fingerprinting: A manifold approximation based framework," NeuroImage, vol. 183, pp. 212 -226, 2018.
  4. H. McMahan, E. Moore, D. Ramage, S. Hampson, and B.A., "Fast trust region for segmentation," in Proc of ICAIS, 2017, pp. 1273-1282. 1
  5. N. Dowlin, R. Gilad-Bachrach, K. Laine, K. Lauter, M. Naehrig, and J. Wernsing, "A reproducible evaluation of ANTs similarity metric performance in brain image registration." in proc of ICML, 2016. 1
  6. E. Hesamifard, H. Takabi, and M. Ghasemi, "CryptoDL: Deep neural networks over encrypted data," arXiv preprint arXiv:1711.05189, 2017. · arXiv
  7. K. Nandakumar, N. Ratha, S. Pankanti, and S. Halevi, "Towards deep neural network training on encrypted data," in proc of CVPR-W, 2019, pp. 0-0. 1
  8. B. Rouhani, S. Riazi, and F. Koushanfar, "DeepSecure: Scalable provably-secure deep learning," in in proc of Design Auto. Conf. (DAC), 2018. 1, 3
  9. I. Goodfellow et al., "Generative adversarial nets," in proc NIPS, 2014, pp. 2672-2680. 2
  10. Y. Ganin et al., "Domain-adversarial training of neural networks," JMLR, vol. 17, no. 1, pp. 2096-2030, 2016. 2
  11. W. Oleszkiewicz, P. Kairouz, K. Piczak, R. Rajagopal, and T. Trzciński, "Siamese generative adversarial privatizer for biometric data," in proc of ACCV, 2018, pp. 482-497. 2, 3
  12. M. T. I. Ziad, A. Alanwar, M. Alzantot, and M. Srivastava, "CryptoImg: Privacy preserving processing over encrypted images," in proc of IEEE CNS, 2016, pp. 570-575. 2
  13. W. Wang, C.-M. Vong, Y. Yang, and P.-K. Wong, "Encrypted image classification based on multilayer extreme learning machine," MSSP, vol. 28, no. 3, pp. 851-865, 2017. 2
  14. P. Paillier, "Public-key cryptosystems based on composite degree resid- uosity classes," in proc ICTACT, 1999, pp. 223-238. 3
  15. C.-Y. Hsu, C.-S. Lu, and S.-C. Pei, "Homomorphic encryption-based secure SIFT for privacy-preserving feature extraction," in proc of MWSF- III, vol. 7880, 2011. 3
  16. J. Dai, B. Saghafi, J. Wu, J. Konrad, and P. Ishwar, "Towards privacy- preserving recognition of human activities," in proc of ICIP, 2015, pp. 4238-4242. 3
  17. J. Chen, J. Wu, K. Richter, J. Konrad, and P. Ishwar, "Estimating head pose orientation using extremely low resolution images," in proc of IEEE SSIAI, 2016, pp. 65-68. 3
  18. D. J. Butler, J. Huang, F. Roesner, and M. Cakmak, "The privacy-utility tradeoff for remotely teleoperated robots," in Proc of ACM/IEEE ICHRI, 2015, pp. 27-34. 3
  19. A. Jalal, M. Z. Uddin, and T.-S. Kim, "Depth video-based human activity recognition system using translation and scaling invariant features for life logging at smart home," IEEE TCE, vol. 58, no. 3, pp. 863-871, 2012. 3
  20. P. McClure et al., "Distributed weight consolidation: A brain segmen- tation case study," in proc of NIPS, 2018, pp. 4093-4103. 3
  21. P. Xie, M. Bilenko, T. Finley, R. Gilad-Bachrach, K. E. Lauter, and M. Naehrig, "Crypto-nets: Neural networks over encrypted data," CoRR, vol. abs/1412.6181, 2014. 3 · arXiv
  22. J. Konecný, H. B. McMahan, D. Ramage, and P. Richtárik, "Federated optimization: Distributed machine learning for on-device intelligence," CoRR, vol. abs/1610.02527, 2016. 3 · arXiv
  23. H. B. McMahan, E. Moore, D. Ramage, and B. A. y Arcas, "Fed- erated learning of deep networks using model averaging," CoRR, vol. abs/1602.05629, 2016. 3 · arXiv
  24. P. Vepakomma, T. Swedish, R. Raskar, O. Gupta, and A. Dubey, "No peek: A survey of private distributed deep learning," CoRR, vol. abs/1812.03288, 2018. 3 · arXiv
  25. Q. Yang, Y. Liu, T. Chen, and Y. Tong, "Federated machine learning: Concept and applications," ACM Transactions on Intelligent Systems and Technology (TIST), vol. 10, no. 2, p. 12, 2019. 3
  26. C. Xu, J. Ren, D. Zhang, Y. Zhang, Z. Qin, and K. Ren, "GANobfusca- tor: Mitigating information leakage under gan via differential privacy," IEEE TIFS, vol. 14, no. 9, pp. 2358-2371, 2019. 3
  27. N. Raval, A. Machanavajjhala, and L. P. Cox, "Protecting visual secrets using adversarial nets," in proc of CVPR-W, 2017, pp. 1329-1332. 3
  28. F. Pittaluga, S. Koppal, and A. Chakrabarti, "Learning privacy preserving encodings through adversarial training," in proc of IEEE WACV, 2019, pp. 791-799. 3
  29. Z. Wu, Z. Wang, Z. Wang, and H. Jin, "Towards privacy-preserving visual recognition via adversarial training: A pilot study," in proc of ECCV, 2018, pp. 606-624. 3
  30. T.-Y. Yang, C. Brinton, P. Mittal, M. Chiang, and A. Lan, "Learning informative and private representations via generative adversarial net- works," in proc of ICBD, 2018, pp. 1534-1543. 3
  31. P. C. Roy and V. N. Boddeti, "Mitigating information leakage in image representations: A maximum entropy approach," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 2586-2594. 3
  32. J. Chen, J. Konrad, and P. Ishwar, "VGAN-based image representation learning for privacy-preserving facial expression recognition," in proc of CVPR-W, 2018, pp. 1570-1579. 3, 11
  33. S. Wang, Z. Ding, and Y. Fu, "Cross-generation kinship verification with sparse discriminative metric," IEEE transactions on pattern analysis and machine intelligence, vol. 41, no. 11, pp. 2783-2790, 2018. 3
  34. H. Xia and Z. Ding, "Structure preserving generative cross-domain learning," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2020. 3
  35. J. Dolz, C. Desrosiers, and I. B. Ayed, "3D fully convolutional networks for subcortical segmentation in MRI: A large-scale study," NeuroImage, vol. 170, pp. 456-470, 2018. 4, 6, 11
  36. C. H. Sudre, W. Li, T. Vercauteren, S. Ourselin, and M. J. Cardoso, "Generalised Dice overlap as a deep learning loss function for highly unbalanced segmentations," in Deep learning in medical image analysis and multimodal learning for clinical decision support, 2017, pp. 240- 248. 4
  37. X. Chen, Y. Duan, R. Houthooft, J. Schulman, I. Sutskever, and P. Abbeel, "InfoGAN: Interpretable representation learning by infor- mation maximizing generative adversarial nets," in Advances in neural information processing systems, 2016, pp. 2172-2180. 5
  38. A. Kraskov, H. Stögbauer, and P. Grassberger, "Estimating mutual information," Physical review E, vol. 69, no. 6, p. 066138, 2004. 5
  39. G. Koch, R. Zemel, and R. Salakhutdinov, "Siamese neural networks for one-shot image recognition," in proc of ICML, 2015. 5
  40. G. Huang, Z. Liu, and K. Q. Weinberger, "Densely connected convolu- tional networks," in proc CVPR, 2017. 5
  41. K. Marek et al., "The Parkinson Progression Marker Initiative (PPMI)," Progress in neurobiology, vol. 95, no. 4, pp. 629-635, 2011. 6
  42. A. G. Roy, S. Conjeti, N. Navab, C. Wachinger et al., "QuickNAT: A fully convolutional network for quick and accurate segmentation of neuroanatomy," NeuroImage, vol. 186, pp. 713-727, 2019. 6
  43. A. M. Mendrik et al., "MRBrainS challenge: online evaluation frame- work for brain image segmentation in 3T MRI scans," Comp. Intel. and Neuro., vol. 2015, p. 1, 2015. 6
  44. B. Avants, N. Tustison, G. Song, P. Cook, A. Klein, and J. Gee, "A reproducible evaluation of ANTs similarity metric performance in brain image registration." Neuroimage, vol. 54, no. 3, pp. 2033-2044, 2011.
  45. K. Kumar, M. Toews, L. Chauvin, O. Colliot, and C. Desrosiers, "Multi- modal brain fingerprinting: a manifold approximation based framework," NeuroImage, vol. 183, pp. 212-226, 2018. 6
  46. J. Dolz, K. Gopinath, J. Yuan, H. Lombaert, C. Desrosiers, and I. Ben Ayed, "HyperDense-Net: A hyper-densely connected CNN for multi- modal image segmentation," IEEE TMI, vol. 38, no. 5, pp. 1116-1126, 2019. 7
  47. A. P. Zijdenbos, B. M. Dawant, R. A. Margolin, and A. C. Palmer, "Morphometric analysis of white matter lesions in MR images: method and validation," IEEE TMI, vol. 13, no. 4, pp. 716-724, 1994. 8
  48. K. Zou et al., "Statistical validation of image segmentation quality based on a spatial overlap index," Academic radiology, vol. 11, pp. 178-89, 02 2004. 8
  49. M. A. Gambacorta et al., "Clinical validation of atlas-based auto- segmentation of pelvic volumes and normal tissue in rectal tumors using auto-segmentation computed system," Acta Oncol, vol. 52, no. 8, pp. 1676-1681, Nov 2013. 8
  50. L. Anders, F. Stieler, K. Siebenlist, J. Schäfer, F. Lohr, and F. Wenz, "Performance of an atlas-based autosegmentation software for delin- eation of target volumes for radiotherapy of breast and anorectal cancer," Journal of the European Society for Therapeutic Radiology and Oncology, vol. 102, pp. 68-73, 09 2011. 8
  51. G. C. Mattiucci et al., "Automatic delineation for replanning in na- sopharynx radiotherapy: What is the agreement among experts to be considered as benchmark?" Acta Oncologica, vol. 52, pp. 1417 -1422, 2013. 8
  52. T. Miyato, S.-i. Maeda, M. Koyama, and S. Ishii, "Virtual adversarial training: a regularization method for supervised and semi-supervised learning," IEEE transactions on pattern analysis and machine intelli- gence, vol. 41, no. 8, pp. 1979-1993, 2018.

Cited by

Sorted by citations

Authors
Kim, B. N., Dolz, J., Jodoin, P.-M., & Desrosiers, C.
Affiliations
  • École de Technologie Supérieure (CA)
  • Université de Sherbrooke (CA)
ORCID
Venue
IEEE Transactions on Medical Imaging
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published
2021-07
Volume
40
Issue
7
Pages
1737-1749
ISSN
  • 0278-0062 (ISSN-L)
  • 1558-254X
DOI
10.1109/tmi.2021.3065727
OpenAlex
W3003567739
Cited by
43
References
5
Citations by year
2026: 5 · 2025: 11 · 2024: 12 · 2023: 8 · 2022: 3 · 2021: 4
FWCI
2.39
Type
journal article
Language
en
Open access
Licence
Locations
  • IEEE Transactions on Medical Imaging (publishedVersion)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • Espace ÉTS (ETS) (submittedVersion)
  • arXiv (Cornell University)
Funders
  • Natural Sciences and Engineering Research Council of Canada (RGPIN-2017-758170, RGPIN-2018-05715)
  • Reseau de BioImagrie du Quebec (247388)
  • NVIDIA Corporation through their GPU grant program
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Medical Image Segmentation Techniques
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
identity obfuscation, adversarial learning, medical image segmentation, brain MRI segmentation, Parkinson's disease, feature removal, domain adaptation
MeSH terms
Humans, Image Processing, Computer-Assisted, Magnetic Resonance Imaging, Privacy

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review