Privacy-Preserving in Medical Image Analysis: A Review of Methods and Applications

Alan Wee-Chung Liew; Hui Tian; Xuefei Yin; Yanming Zhu

REVIEW 3 major objections 5 minor 1 cited by

Privacy-Preserving in Medical Image Analysis: A Review of Methods and Applications

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This review maps six privacy-preserving techniques onto the distinct challenges of medical imaging tasks, so that each practical problem has a matched solution.

desk verdict A genuinely useful task-oriented survey of privacy-preserving medical imaging, but the citation and classification errors in its load-bearing application-to-technique mapping need fixing before I'd trust it. read the letter →

arxiv 2412.03924 v1 pith:ITQPM3TB submitted 2024-12-05 cs.CV

Yanming Zhu , Xuefei Yin , Alan Wee-Chung Liew , Hui Tian This is my paper · ORCID

classification cs.CV

keywords Privacy-preservingMedicalimageanalysisFederatedlearningDifferentialprivacyEncryptionobfuscation/deformationHomomorphicGenerativeadversarialnetworks

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper is a review, not a new method. It claims that existing surveys of privacy-preserving medical image analysis concentrate on individual technologies, leaving practitioners without a task-level view, and it aims to close that gap by organizing the field according to the challenges that arise in disease diagnosis, foundational image processing, pathological image analysis, and telemedicine. For each application area it identifies the privacy techniques that have been demonstrated—encryption, differential privacy, homomorphic encryption, federated learning, generative adversarial networks, and image deformation or obfuscation—and summarizes their reported benefits and trade-offs. If the review is right, a researcher or clinician can choose a privacy approach by looking up the concrete problem they face, rather than starting from a technology and searching for a use.

What carries the argument

The organizing device is an application-to-technique mapping: four classes of medical image analysis tasks are each paired with the privacy techniques that have been shown to address their characteristic threats, together with the advantages and limitations of each pairing. This mapping, rather than any single algorithm, carries the review's argument, because it converts a list of cryptographic and learning-based tools into a decision aid for specific practical problems. The same structure frames the paper's future directions, since gaps in the mapping indicate where zero-knowledge proofs and secure multi-party computation could be added.

What would settle it

An independent structured literature search covering the same application areas and technique families, followed by a check of every in-text citation against the reference list, would settle whether the mapping is complete and accurate; one concrete test is verifying whether the Section 3.2 differential-privacy liver segmentation result belongs to reference [48] rather than [37].

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that a useful perspective on privacy-preserving medical image analysis comes from aligning technical solutions with the specific privacy challenges of each application. It supports this claim by surveying recent work in four application domains: disease diagnosis, where sharing data for model training is the main risk; foundational processing such as segmentation and registration, where preserving accuracy is critical; pathological image analysis, where extremely high resolution makes re-identification a serious concern; and telemedicine, where cross-institutional sharing and cloud processing create exposure. Within each domain it maps the techniques that have been used, such as differentially private training and federated learning for diagnosis, encryption and adversarial obfuscation for segmentation, federated learning and generative adversarial networks for pathology, and deformation or homomorphic encryption for telemedicine. The review also names shared limitations of these methods, including noise-induced accuracy loss, computational cost, and scalability problems, and identifies zero-knowledge proofs and secure multi-party computation as emerging directions.

Load-bearing premise

The load-bearing premise is that the selection of surveyed papers is representative and correctly cited; if key works were missed or misattributed, the application-to-technique mapping would be unreliable, as illustrated by the Section 3.2 attribution of a differential-privacy liver segmentation method to the image-registration reference [37], which appears to be a mismatch with the differential-privacy paper [48].

Editorial extensions

If this is right

Researchers and clinicians can, according to this review, start from the imaging task—diagnosis, foundational processing, pathology, or telemedicine—and select among the privacy techniques already demonstrated in that setting.
Federated learning and differential privacy are presented as complementary, with several cited systems combining them, such as secure aggregation plus differential privacy for chest X-ray classification.
Homomorphic encryption is described as workable for encrypted diagnosis and telemedicine, but its computational cost is a barrier to real-time use, so lighter-weight variants are stated as a research need.
GAN-based synthetic data and image deformation or obfuscation are positioned as ways to share usable images without exposing raw patient data, with the caveat that they may discard diagnostic detail.
The review's proposed future directions—personalized privacy solutions, dynamic noise adjustment, efficient federated communication, zero-knowledge proofs, and secure multi-party computation—follow directly from the limitations it identifies in current methods.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper: the same challenge-to-solution mapping could become a quantitative decision aid, scoring each technique family on privacy strength, diagnostic accuracy loss, and compute cost within each of the four application areas.
Beyond the paper: the four application categories are not shown to be exhaustive; emerging settings such as multimodal or longitudinal imaging could stretch the taxonomy, so the mapping is best read as an open scaffold rather than a closed list.
Beyond the paper: grouping methods by application rather than by formal privacy guarantee suggests a natural follow-up benchmark that measures privacy-utility trade-offs on the same medical dataset across encryption, differential privacy, federated learning, and GAN-based approaches.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

This review's application-first organization is its real contribution. Instead of the usual technique-by-technique survey, it arranges privacy methods around clinical tasks—diagnosis, pathology, telemedicine—and that is a genuinely useful angle for someone trying to pick a method for a concrete problem. The technique overviews are consistent with the field, and the discussion of trade-offs is fair. If you need a starting point for privacy-preserving medical image analysis, this is a reasonable entry.

The soft spots are real but fixable. The stress-test note checks out: in §3.2 the differential privacy framework for liver segmentation is attributed to Ziller et al. [37], but [37] is actually Taiello et al. on image registration; the DP paper is [48]. And in §3.4, Yu et al. [45] (POMIC, a cloud outsourcing scheme) is presented under federated learning, which is a technique misclassification, not just a wrong number. Because the whole point of the review is to map techniques to applications, these errors misdirect a reader trying to learn which tool fits which task. They are not mere typos.

There is also no described search or inclusion protocol, so the comprehensiveness claim is hard to verify and the observed errors raise doubt about how many more misclassifications might be hiding. The abstract's grammar glitch ("we organizes") is minor, but it fits a pattern of careless editing.

The central organizational claim still holds up—privacy techniques can indeed be organized by medical imaging application, and this is a useful framing. The paper does not pretend to offer new experiments or derivations, so I don't fault it for that. But a review whose value depends on reliable mapping needs accurate citations and correct technique labels.

Who is this for? A newcomer to privacy-preserving medical imaging, or a clinician wanting a task-oriented overview. It deserves a serious referee, but only after revision. The authors should correct the citation errors, re-check every technique label against the reference, describe how they selected the literature, and tone down the "comprehensive" claim. With those changes, I'd be comfortable pointing people to it.

Referee Report

3 major / 5 minor

Summary. This manuscript is a narrative review of privacy-preserving techniques for medical image analysis. It surveys six families of techniques—encryption, de-identification/anonymization, differential privacy, homomorphic encryption, federated learning, and GANs—and organizes the literature by four application areas: disease diagnosis, foundational medical image processing, pathological image analysis, and telemedicine. For each application area, the authors map concrete challenges to corresponding privacy-preserving solutions, and they close with a discussion of advantages, limitations, and future directions such as zero-knowledge proofs and secure multi-party computation. The stated main contribution is this application-first organization, which the authors argue directly aligns technical options with practical problems and fills a gap left by technology-centric surveys.

Significance. If the application-to-technique mapping were reliable, this review would be a genuinely useful entry point for researchers and practitioners wanting to know which privacy-preserving technique fits a given medical imaging task. The application-oriented organization is a real strength, and the paper includes recent literature up to 2024, explicit discussion of trade-offs, and a forward-looking treatment of emerging technologies. The paper does not provide machine-checked proofs or reproducible code, but as a survey its value would come from accurate and comprehensive coverage. That value is currently compromised by citation and classification errors in the central mapping, and by the absence of any described selection or verification methodology.

major comments (3)

[§3.2, 'Differential privacy' bullet] The text states that "Ziller et al. [37] introduced a differential privacy deep learning framework for liver segmentation" as part of the Medical Segmentation Decathlon, but reference [37] is Taiello et al., "Privacy preserving image registration" (Medical Image Analysis, 2024), which is not a differential-privacy liver-segmentation paper. The actual Ziller et al. differential-privacy work is reference [48], which is cited correctly earlier in §3.1. The same citation number is thus used for two different works, and the support for the DP-for-segmentation claim is misattributed. Because the paper's central contribution is precisely the mapping from techniques to cited works, this needs correction and, more importantly, a full audit of every citation-to-technique assignment.
[§3.4, 'Federated learning' heading] Under the heading "Federated learning," the manuscript attributes to Yu et al. [45] "a flexible outsourcing scheme for privacy-preserving medical image classification on the cloud using convolutional neural networks." Reference [45] is POMIC, which is exactly a cloud outsourcing scheme and is not a federated learning method. This is not merely a wrong number; the cited work is placed in the wrong technique category. The heading and surrounding text should be corrected, for example by moving this work to the encryption/cloud-computing discussion or by removing it from the federated learning subsection. This error, together with the misattribution in §3.2, indicates that the technique classification needs systematic verification.
[§1 and §3 (overall methodology)] The manuscript describes itself as a "comprehensive overview" and claims to address gaps in the current research landscape, but it does not describe any search strategy, inclusion/exclusion criteria, or protocol for validating the technique classification of each cited work. The two concrete misplacements noted above show that such a protocol is needed: without it, the selection of works and the per-application technique assignments are not independently verifiable, and the claimed comprehensiveness cannot be assessed. The authors should either add a methodology subsection that explains how papers were collected and classified, or explicitly scope the review as a narrative survey with more modest claims about comprehensiveness.

minor comments (5)

[Abstract] The sentence "we organizes the review" should read "we organize the review," and the final phrase "privacy-preserving in medical image analysis" should read "privacy-preserving techniques in medical image analysis."
[Introduction] The subsection heading "Aims and Conributions" contains a typo; it should be "Contributions."
[§3.4] The heading "Federated learningYu et al." is missing a space; it should read "Federated learning. Yu et al."
[Figure 1] The text includes a caption for Fig. 1, but the figure itself is not visible in the manuscript text provided. If the figure is intended, please ensure it is actually rendered; if it is not, remove the caption and any reference to it.
[References] The reference list has minor formatting inconsistencies, such as "IEEE Access" appearing as both "IEEE Access" and "IEEE access" (references [13] and [15]), and several arXiv preprints are listed without an explicit preprint designation. Please harmonize the reference style throughout.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: this is a literature review with no derivation, fitted parameters, or self-referential equations, and its only self-citation is a background citation for a standard definition.

full rationale

This paper is a narrative review of privacy-preserving techniques in medical image analysis. Its central contribution is a taxonomy organized by applications and corresponding technical solutions, and that taxonomy is descriptive rather than derived from any fitted model, equation, or definitional loop. There are no derivations, no experimental predictions, no parameters fitted to data, and no mathematical claims that could reduce to their own inputs. The only identifiable self-citation is reference [44], a federated-learning survey by co-author Xuefei Yin and others, cited in Section 2.5 to support the standard definition of federated learning as a distributed approach where institutions train locally and exchange model parameters. That citation is background support for a well-established concept, not load-bearing for the paper's central application-to-technique mapping, so it does not constitute circularity. The skeptic's examples of citation mismatches, such as Section 3.2 attributing a differential-privacy liver-segmentation framework to reference [37] when that reference is the Taiello et al. image-registration paper, and Section 3.4 labeling Yu et al. [45] as federated learning when it is a cloud outsourcing scheme, are genuine accuracy and verification concerns for a review. However, misattribution is not circularity: it means the support for a claim may be incorrect, but the claim itself is not being defined in terms of its conclusion or statistically forced by a fitted input. Similarly, the absence of a described systematic search protocol affects the reliability and completeness of the review, but it does not make the review's organizational claims self-referential. Under the stated criteria, a review with no derivation chain and no self-referential argument should receive a circularity score of zero.

Assumptions & free parameters 0 free parameters · 0 assumptions · 0 invented entities

This is a review paper; no free parameters, axioms, or invented entities are introduced. The central claim depends only on the accuracy and representativeness of the cited literature, not on any new mathematical assumptions or fitted quantities.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Privacy-Preserving in Medical Image Analysis: A Review of Methods and Applications." pith.science (2026). https://pith.science/paper/ITQPM3TB

@misc{pith2026241203924,
  author       = {Pith},
  title        = {Pith review of: Privacy-Preserving in Medical Image Analysis: A Review of Methods and Applications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ITQPM3TB}},
  note         = {Machine review of arXiv:2412.03924}
}

read the original abstract

With the rapid advancement of artificial intelligence and deep learning, medical image analysis has become a critical tool in modern healthcare, significantly improving diagnostic accuracy and efficiency. However, AI-based methods also raise serious privacy concerns, as medical images often contain highly sensitive patient information. This review offers a comprehensive overview of privacy-preserving techniques in medical image analysis, including encryption, differential privacy, homomorphic encryption, federated learning, and generative adversarial networks. We explore the application of these techniques across various medical image analysis tasks, such as diagnosis, pathology, and telemedicine. Notably, we organizes the review based on specific challenges and their corresponding solutions in different medical image analysis applications, so that technical applications are directly aligned with practical issues, addressing gaps in the current research landscape. Additionally, we discuss emerging trends, such as zero-knowledge proofs and secure multi-party computation, offering insights for future research. This review serves as a valuable resource for researchers and practitioners and can help advance privacy-preserving in medical image analysis.

Figures

Figures reproduced from arXiv: 2412.03924 by the authors.

**Figure 1.** Privacy-preserving techniques in different medical image analysis applications. 3.1 Disease Diagnosis Disease diagnosis, such as for cancer and heart disease, heavily depends on analyzing medical images like CT, MRI, and X-rays. These images contain both pathological information and sensitive personal health data, creating significant privacy challenges. The need for large datasets to train AI models increases the … view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Beyond Anonymization: Object Scrubbing for Privacy-Preserving 2D and 3D Vision Tasks
cs.CV 2025-04 conditional novelty 5.0 of 10

Scrubbing sensitive objects with inpainting preserves more downstream detection and 3D reconstruction utility than deleting images, but the paper's privacy guarantee rests on the same oracle detector that defines and ...

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages · cited by 1 Pith paper

[37]

Medical Image Analysis 94, 103129 (2024)

Taiello, R., Önen, M., Capano, F., et al.: Privacy preser ving image registration. Medical Image Analysis 94, 103129 (2024)

work page 2024
[48]

Scientiﬁc Reports 11(1), 13524 (2021)

Ziller, A., Usynin, D., Braren, R., et al.: Medical imagi ng deep learning with dif- ferential privacy. Scientiﬁc Reports 11(1), 13524 (2021)

work page 2021
[45]

Applied Sciences 13(6), 3439 (2023)

Yu, Q., Zhang, H., Xu, H., Kong, F.: Pomic: Privacy-prese rving outsourcing med- ical image classiﬁcation based on convolutional neural net work to cloud. Applied Sciences 13(6), 3439 (2023)

work page 2023
[1]

BioData Mining 17(1), 33 (2024)

Al Badawi, A., Faizal Bin Yusof, M.: Private pathological assessment via machine learning and homomorphic encryption. BioData Mining 17(1), 33 (2024)

work page 2024
[2]

Journal of Information Technology Managem ent 12, 50–67 (2020)

Andrew, J., et al.: An eﬃcient privacy-preserving deep le arning scheme for medical image analysis. Journal of Information Technology Managem ent 12, 50–67 (2020)

work page 2020
[3]

IEEE Open J

Bisht, A., et al.: Eﬃcient personal-health-records shar ing in internet of medical things using searchable symmetric encryption, blockchain and ipfs. IEEE Open J. Commun. Soc. (2023) Privacy-Preserving in Medical Image Analysis: A Review 11

work page 2023
[4]

Chen, Y., Yang, X.H., et al.: Generative adversarial netw orks in medical image augmentation: a review. Comput. Biol. Med. 144, 105382 (2022)

work page 2022
[5]

Chevrier, R., et al.: Use and understanding of anonymizat ion and de-identiﬁcation in the biomedical literature: scoping review. J. Med. Inter net Res. 21(5), e13484 (2019)

work page 2019

Show all 48 references

[6]

Journal of Engineering Science & Technolog y Review 12(6) (2019)

Churi, P.P., Pawar, A.V.: A systematic review on privacy p reserving data publish- ing techniques. Journal of Engineering Science & Technolog y Review 12(6) (2019)

2019
[7]

In: DSAA

Faisal, F., Mohammed, N., Leung, C.K., Wang, Y.: Generati ng privacy preserving synthetic medical data. In: DSAA. pp. 1–10. IEEE (2022)

2022
[8]

WIRES DATA MIN KNOWL 13(4), e1495 (2023)

Gaudio, A., et al.: Deepﬁxcx: Explainable privacy-prese rving image compression for medical image analysis. WIRES DATA MIN KNOWL 13(4), e1495 (2023)

2023
[9]

In: ISBI

Gkillas, A., Ampeliotis, D., Berberidis, K.: Privacy-pr eserving federated deep- equilibrium learning for medical image classiﬁcation. In: ISBI. pp. 1–4. IEEE (2024)

2024
[10]

Pattern Recognition p

Guan, H., Yap, P.T., Bozoki, A., Liu, M.: Federated learn ing for medical image analysis: A survey. Pattern Recognition p. 110424 (2024)

2024
[11]

In: TML4H

Guerra-Manzanares, A., Lopez, L.J.L., Maniatakos, M., Shamout, F.E.: Privacy- preserving machine learning for healthcare: open challeng es and future perspec- tives. In: TML4H. pp. 25–40. Springer (2023)

2023
[12]

IEEE Access 9, 47731–47742 (2021)

Hasan, M.K., Islam, S., Sulaiman, R., et al.: Lightweigh t encryption technique to enhance medical image security on internet of medical thing s applications. IEEE Access 9, 47731–47742 (2021)

2021
[13]

IEEE Access 10, 66345–66355 (2022)

Huang, Q.X., Yap, W.L., et al.: Privacy-preserving deep learning with learnable image encryption on medical images. IEEE Access 10, 66345–66355 (2022)

2022
[14]

BioMed research inter- national 2022(1), 5164970 (2022)

Hussain, S., Mubeen, I., Ullah, N., et al.: Modern diagno stic imaging technique applications and risk factors in the medical ﬁeld: a review. BioMed research inter- national 2022(1), 5164970 (2022)

2022
[15]

IEEE access 7, 61656–61669 (2019)

Jin, H., Luo, Y., Li, P., Mathew, J.: A review of secure and privacy-preserving medical data sharing. IEEE access 7, 61656–61669 (2019)

2019
[16]

Kaissis, G., Ziller, A., et al.: End-to-end privacy pres erving deep learning on multi- institutional medical imaging. Nat. Mach. Intell. 3(6), 473–484 (2021)

2021
[17]

Khalid, N., Qayyum, A., Bilal, M., et al.: Privacy-prese rving artiﬁcial intelligence in healthcare: Techniques and applications. Comput. Biol. Med. 158, 106848 (2023)

2023
[18]

arXiv preprint arxiv:1909.0408 7 (2019)

Kim, B.N., Desrosiers, C., et al.: Privacy-net: An adver sarial approach for identity- obfuscated segmentation. arXiv preprint arxiv:1909.0408 7 (2019)

2019
[19]

arXiv preprint arXiv:2011.1283 5 (2020)

Kim, B.N., Dolz, J., et al.: Privacy preserving for medic al image analysis via non- linear deformation proxy. arXiv preprint arXiv:2011.1283 5 (2020)

2020
[20]

A i 4(3), 706–720 (2023)

Kolhar, M., Aldossary, S.M.: Privacy-preserving convo lutional bi-lstm network for robust analysis of encrypted time-series medical images. A i 4(3), 706–720 (2023)

2023
[21]

TOIT 21(2), 1–22 (2021)

Kumar, A., et al.: A novel cloud-assisted secure deep fea ture classiﬁcation frame- work for cancer histopathology images. TOIT 21(2), 1–22 (2021)

2021
[22]

COMPUT MED IMAG GRAP 102, 102139 (2022)

Kumar, R., Kumar, J., Khan, A.A., et al.: Blockchain and h omomorphic encryption based privacy-preserving model aggregation for medical im ages. COMPUT MED IMAG GRAP 102, 102139 (2022)

2022
[23]

IEEE J BIOMED HEALTH (2024)

Lai, H., Luo, Y., Li, B., et al.: Bilateral proxy federate d domain generalization for privacy-preserving medical image diagnosis. IEEE J BIOMED HEALTH (2024)

2024
[24]

Applied Sciences 13(14), 8295 (2023)

Lata, K., Cenkeramaddi, L.R.: Deep learning for medical image cryptography: A comprehensive review. Applied Sciences 13(14), 8295 (2023)

2023
[25]

Neural Comput

Li, D., Wu, Y., Lü, Q., et al.: Pmir: an eﬃcient privacy-pr eserving medical images search in cloud-assisted scenario. Neural Comput. Appl. 36(3), 1477–1493 (2024)

2024
[26]

Annals of Data Science 11(2), 733–747 (2024) 12 Yanming Zhu, Xuefei Yin, Alan Wee-Chung Liew, and Hui Tian

Liu, W., Zhang, Y., Yang, H., Meng, Q.: A survey on diﬀeren tial privacy for medical data analysis. Annals of Data Science 11(2), 733–747 (2024) 12 Yanming Zhu, Xuefei Yin, Alan Wee-Chung Liew, and Hui Tian

2024
[27]

IEEE TIFS 17, 69–84 (2019)

Liu, Y., Ma, Z., Liu, X., et al.: Privacy-preserving obje ct detection for medical images with faster r-cnn. IEEE TIFS 17, 69–84 (2019)

2019
[28]

Medical image analysis 76, 102298 (2022)

Lu, M.Y., Chen, R.J., Kong, D., et al.: Federated learnin g for computational pathol- ogy on gigapixel whole slide images. Medical image analysis 76, 102298 (2022)

2022
[29]

IEEE Acces s 9, 148037–148047 (2021)

Montenegro, H., et al.: Privacy-preserving generative adversarial network for case- based explainability in medical image analysis. IEEE Acces s 9, 148037–148047 (2021)

2021
[30]

S.: Medical image en- cryption through chaotic asymmetric cryptosystem

Ningthoukhongjam, T.R., Heisnam, S.D., Khumanthem, M. S.: Medical image en- cryption through chaotic asymmetric cryptosystem. IEEE Ac cess (2024)

2024
[31]

In: AI for Health Equity and Fairness, pp

Panzade, P., Takabi, D., Cai, Z.: Medblindtuner: Toward s privacy-preserving ﬁne- tuning on biomedical images with transformers and fully hom omorphic encryption. In: AI for Health Equity and Fairness, pp. 197–208. Springer (2024)

2024
[32]

Bioengineering 10(12), 1435 (2023)

Pinto-Coelho, L.: How artiﬁcial intelligence is shapin g medical imaging technology: A survey of innovations and applications. Bioengineering 10(12), 1435 (2023)

2023
[33]

Journal of Healthcare Informatics Research 7(2), 203–224 (2023)

Riedel, P., von Schwerin, R., Schaudt, D., et al.: Resnet fed: federated deep learn- ing architecture for privacy-preserving pneumonia detect ion from covid-19 chest radiographs. Journal of Healthcare Informatics Research 7(2), 203–224 (2023)

2023
[34]

IEEE Transactions on Medical Imaging 42(7), 1969–1981 (2022)

Shen, Y., Sowmya, A., Luo, Y., et al.: A federated learnin g system for histopathol- ogy image analysis with an orchestral stain-normalization gan. IEEE Transactions on Medical Imaging 42(7), 1969–1981 (2022)

2022
[35]

IEEE Access 11, 28628–28644 (2023)

Sohan, M.F., Basalamah, A.: A systematic review on feder ated learning in medical image analysis. IEEE Access 11, 28628–28644 (2023)

2023
[36]

arXiv prepr int arXiv:2403.16473 (2024)

Sun, M., et al.: Plaintext-free deep learning for privac y-preserving medical image analysis via frequency information embedding. arXiv prepr int arXiv:2403.16473 (2024)

2024 arXiv
[38]

Frontiers in Medic ine 10, 1227515 (2023)

Verma, T., et al.: Privacy-preserving continual learni ng methods for medical image classiﬁcation: a comparative analysis. Frontiers in Medic ine 10, 1227515 (2023)

2023
[39]

ACM Comput

Wood, A., Najarian, K., et al.: Homomorphic encryption f or machine learning in medicine and bioinformatics. ACM Comput. Surv. 53(4), 1–35 (2020)

2020
[40]

ACM T MULTIM COMPUT 19(1s), 1–23 (2023)

Xiang, T., Zeng, H., Chen, B., Guo, S.: Bmif: Privacy-pre serving blockchain-based medical image fusion. ACM T MULTIM COMPUT 19(1s), 1–23 (2023)

2023
[41]

In: CRAE

Xiong, Y., et al.: Privacy preserving data distillation in medical imaging with multidimensional matching on pathmnist. In: CRAE. vol. 132 49, pp. 23–28. SPIE (2024)

2024
[42]

Yang, P., et al.: Study of medical image classiﬁcation wi th privacy-preserving fed- erated learning: Chest x-ray images classiﬁcation as an exa mple (2023)

2023
[43]

Medical image analysis 58, 101552 (2019)

Yi, X., Walia, E., Babyn, P.: Generative adversarial net work in medical imaging: A review. Medical image analysis 58, 101552 (2019)

2019
[44]

ACM Comput

Yin, X., et al.: A comprehensive survey of privacy-prese rving federated learning: A taxonomy, review, and future directions. ACM Comput. Surv. 54(6), 1–36 (2021)

2021
[46]

Zhou, J., et al.: Personalized and privacy-preserving f ederated heterogeneous med- ical image analysis with pppml-hmi. Comput. Biol. Med. 169, 107861 (2024)

2024
[47]

IEEE Transactions on Medical Imaging ( 2024)

Zhu, E., et al.: Mp-net: A multi-center privacy-preserv ing network for medical image segmentation. IEEE Transactions on Medical Imaging ( 2024)

2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [37]

Medical Image Analysis 94, 103129 (2024)

Taiello, R., Önen, M., Capano, F., et al.: Privacy preser ving image registration. Medical Image Analysis 94, 103129 (2024)

work page 2024

[2] [48]

Scientiﬁc Reports 11(1), 13524 (2021)

Ziller, A., Usynin, D., Braren, R., et al.: Medical imagi ng deep learning with dif- ferential privacy. Scientiﬁc Reports 11(1), 13524 (2021)

work page 2021

[3] [45]

Applied Sciences 13(6), 3439 (2023)

Yu, Q., Zhang, H., Xu, H., Kong, F.: Pomic: Privacy-prese rving outsourcing med- ical image classiﬁcation based on convolutional neural net work to cloud. Applied Sciences 13(6), 3439 (2023)

work page 2023

[4] [1]

BioData Mining 17(1), 33 (2024)

Al Badawi, A., Faizal Bin Yusof, M.: Private pathological assessment via machine learning and homomorphic encryption. BioData Mining 17(1), 33 (2024)

work page 2024

[5] [2]

Journal of Information Technology Managem ent 12, 50–67 (2020)

Andrew, J., et al.: An eﬃcient privacy-preserving deep le arning scheme for medical image analysis. Journal of Information Technology Managem ent 12, 50–67 (2020)

work page 2020

[6] [3]

IEEE Open J

Bisht, A., et al.: Eﬃcient personal-health-records shar ing in internet of medical things using searchable symmetric encryption, blockchain and ipfs. IEEE Open J. Commun. Soc. (2023) Privacy-Preserving in Medical Image Analysis: A Review 11

work page 2023

[7] [4]

Chen, Y., Yang, X.H., et al.: Generative adversarial netw orks in medical image augmentation: a review. Comput. Biol. Med. 144, 105382 (2022)

work page 2022

[8] [5]

Chevrier, R., et al.: Use and understanding of anonymizat ion and de-identiﬁcation in the biomedical literature: scoping review. J. Med. Inter net Res. 21(5), e13484 (2019)

work page 2019