Stefan Dvoretskii
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Stefan Dvoretskii has a total of 93 co-authors. Here are the top co-authors by number of shared publications:
- Marco Nolden
German Cancer Research Center
3 shared publications - Hamideh Haghiri
German Cancer Research Center
2 shared publications - Lena Maier-Hein
German Cancer Research Center
2 shared publications - Rajesh Baidya
German Cancer Research Center
2 shared publications - Alex Michie
Division of Radiotherapy and Imaging Institute of Cancer Research London UK
1 shared publication Ambar Rana
Essential Software Inc 9711 Washingtonian Blvd Suite 550 20878 Gaithersburg MD
1 shared publication
Latest Publications
- 2 February 2027
- Published by Springer Nature
- p. 117-125
A Hybrid AI and Rule-Based Approach to DICOM De-identification: A Solution for the MIDI-B Challenge at MICCAI 2024
- 2 February 2027
- Published by Springer Nature
- p. 117-125
A Hybrid AI and Rule-Based Approach to DICOM De-identification: A Solution for the MIDI-B Challenge at MICCAI 2024
- Open Access
- 28 September 2026
- Published by MDPI AG in Preprints.org
Comparison of Metadata Catalog Schemas for Cross-Border Health Dataset Discovery
Introduction: The European Health Data Space (EHDS) Regulation (EU) 2025/327 requires Member States to establish national health dataset catalogs feeding into the HealthData@EU infrastructure. While the detailed technical specifications for catalog content are currently being developed through HealthDCAT-AP and forthcoming implementing acts, several national and domain-specific health data infrastructures have already established metadata catalogs based on (Health)DCAT-AP and related profiles. These early implementations provide a valuable opportunity to assess emerging convergence, metadata alignment challenges, and identify practical requirements for future harmonization. Objectives: In this paper, we compare a set of independently developed metadata catalog schemas in use across European health data infrastructures to identify areas of structural and semantic convergence and divergence. This aims to provide a basis for alignment consistent with the emerging HealthDCAT-AP specification. Methods: This work was initiated at the SWAT4HCLS 2026 BioHackathon, where representatives of the participating infrastructures jointly manually mapped their schemas, and continued through extended collaborative efforts. A total of eleven schemas, health-related metadata catalogs and application profiles, were compared along three axes: classes, properties and value sets. Results: The study shows that the schemas align well with DCAT-AP, and especially requirements of the HealthDCAT-AP for the health-specific schemas, though there are some divergences and expected extensions. For aligned content, we illustrate the practical payoff of these alignments with federated SPARQL queries enabled at each level. Conclusions: The national metadata schemas provide a solid foundation towards enabling interoperable metadata discovery, while complying with the EHDS regulation. The necessary alignments and suggestions for extensions of studied schemas have been outlined, communicated and are being discussed on the national level. Beyond the technical findings, this work established a cross-community forum for alignment, knowledge exchange, and coordinated evolution of metadata specifications. Maintaining such collaboration will be essential as HealthDCAT-AP and related EHDS requirements continue to mature.
- 28 September 2026
- Published by MDPI AG in Preprints.org
Open AccessComparison of Metadata Catalog Schemas for Cross-Border Health Dataset Discovery
Introduction: The European Health Data Space (EHDS) Regulation (EU) 2025/327 requires Member States to establish national health dataset catalogs feeding into the HealthData@EU infrastructure. While the detailed technical specifications for catalog content are currently being developed through HealthDCAT-AP and forthcoming implementing acts, several national and domain-specific health data infrastructures have already established metadata catalogs based on (Health)DCAT-AP and related profiles. These early implementations provide a valuable opportunity to assess emerging convergence, metadata alignment challenges, and identify practical requirements for future harmonization. Objectives: In this paper, we compare a set of independently developed metadata catalog schemas in use across European health data infrastructures to identify areas of structural and semantic convergence and divergence. This aims to provide a basis for alignment consistent with the emerging HealthDCAT-AP specification. Methods: This work was initiated at the SWAT4HCLS 2026 BioHackathon, where representatives of the participating infrastructures jointly manually mapped their schemas, and continued through extended collaborative efforts. A total of eleven schemas, health-related metadata catalogs and application profiles, were compared along three axes: classes, properties and value sets. Results: The study shows that the schemas align well with DCAT-AP, and especially requirements of the HealthDCAT-AP for the health-specific schemas, though there are some divergences and expected extensions. For aligned content, we illustrate the practical payoff of these alignments with federated SPARQL queries enabled at each level. Conclusions: The national metadata schemas provide a solid foundation towards enabling interoperable metadata discovery, while complying with the EHDS regulation. The necessary alignments and suggestions for extensions of studied schemas have been outlined, communicated and are being discussed on the national level. Beyond the technical findings, this work established a cross-community forum for alignment, knowledge exchange, and coordinated evolution of metadata specifications. Maintaining such collaboration will be essential as HealthDCAT-AP and related EHDS requirements continue to mature.
- Open Access
Medical Image De-Identification Benchmark Challenge
The de-identification (deID) of protected health information (PHI) and personally identifiable information (PII) is a fundamental requirement for sharing medical images, particularly through public repositories, to ensure compliance with patient privacy laws. In addition, preservation of non-PHI metadata to inform and enable downstream development of imaging artificial intelligence (AI) is an important consideration in biomedical research. The goal of MIDI-B was to provide a standardized platform for benchmarking of DICOM image deID tools based on a set of rules conformant to the HIPAA Safe Harbor regulation, the DICOM Attribute Confidentiality Profiles, and best practices in preservation of research-critical metadata, as defined by The Cancer Imaging Archive (TCIA). The challenge employed a large, diverse, multi-center, and multi-modality set of real de-identified radiology images with synthetic PHI/PII inserted. The MIDI-B Challenge consisted of three phases: training, validation, and test. Eighty individuals registered for the challenge. In the training phase, we encouraged participants to tune their algorithms using their in-house or public data. The validation and test phases utilized the DICOM images containing synthetic identifiers (of 216 and 322 subjects, respectively). Ten teams successfully completed the test phase of the challenge. To measure success of a rule-based approach to image deID, scores were computed as the percentage of correct actions from the total number of required actions. The scores ranged from 97.91% to 99.93%. Participants employed a variety of open-source and proprietary tools with customized configurations, large language models, and optical character recognition (OCR). In this paper we provide a comprehensive report on the MIDI-B Challenge's design, implementation, results, and lessons learned.
Open AccessMedical Image De-Identification Benchmark Challenge
The de-identification (deID) of protected health information (PHI) and personally identifiable information (PII) is a fundamental requirement for sharing medical images, particularly through public repositories, to ensure compliance with patient privacy laws. In addition, preservation of non-PHI metadata to inform and enable downstream development of imaging artificial intelligence (AI) is an important consideration in biomedical research. The goal of MIDI-B was to provide a standardized platform for benchmarking of DICOM image deID tools based on a set of rules conformant to the HIPAA Safe Harbor regulation, the DICOM Attribute Confidentiality Profiles, and best practices in preservation of research-critical metadata, as defined by The Cancer Imaging Archive (TCIA). The challenge employed a large, diverse, multi-center, and multi-modality set of real de-identified radiology images with synthetic PHI/PII inserted. The MIDI-B Challenge consisted of three phases: training, validation, and test. Eighty individuals registered for the challenge. In the training phase, we encouraged participants to tune their algorithms using their in-house or public data. The validation and test phases utilized the DICOM images containing synthetic identifiers (of 216 and 322 subjects, respectively). Ten teams successfully completed the test phase of the challenge. To measure success of a rule-based approach to image deID, scores were computed as the percentage of correct actions from the total number of required actions. The scores ranged from 97.91% to 99.93%. Participants employed a variety of open-source and proprietary tools with customized configurations, large language models, and optical character recognition (OCR). In this paper we provide a comprehensive report on the MIDI-B Challenge's design, implementation, results, and lessons learned.
- Open Access
- 17 May 2025
- Published by Center for Open Science in BioHackrXiv
Towards AI-Ready Datasets for the Life Sciences
At the SWAT4HCLS 2025 Hackathon, we continued our work on dataset interoperability andAI-readiness, extending our efforts from the 2024 Elixir Biohackathon. This report outlinesthe progress made in graph serialization, metadata embedding, and knowledge graph analysis,which further enhance machine learning workflows and data integration
- 17 May 2025
- Published by Center for Open Science in BioHackrXiv
Open AccessTowards AI-Ready Datasets for the Life Sciences
At the SWAT4HCLS 2025 Hackathon, we continued our work on dataset interoperability andAI-readiness, extending our efforts from the 2024 Elixir Biohackathon. This report outlinesthe progress made in graph serialization, metadata embedding, and knowledge graph analysis,which further enhance machine learning workflows and data integration
- Open Access
- 9 December 2024
- Published by Center for Open Science in BioHackrXiv
Publishing FAIR datasets from Bioimaging repositories
We assessed the implementation of the FAIR principles in the currentbioimaging and clinical imaging data repositories. Additionally, to make the RDF export triples from the IDR discoverable, we also explored the Fair Data Point interface (Silva Santos et al.,2023), as a mechanism to facilitate the exposure of machine-actionable metadata and how itcould be added to the Imaging Data Resource (IDR) portal.
- 9 December 2024
- Published by Center for Open Science in BioHackrXiv
Open AccessPublishing FAIR datasets from Bioimaging repositories
We assessed the implementation of the FAIR principles in the currentbioimaging and clinical imaging data repositories. Additionally, to make the RDF export triples from the IDR discoverable, we also explored the Fair Data Point interface (Silva Santos et al.,2023), as a mechanism to facilitate the exposure of machine-actionable metadata and how itcould be added to the Imaging Data Resource (IDR) portal.