Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 8
Encontrada(s) 1 página(s)

UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Rethinking trust in synthetic health data: lessons from 7 European research initiatives

  • Declerck, Jens
  • Kalra, Dipak
  • Airola, Antti
  • Ali Amer, Ahmed Youssef
  • Chatzichristos, Christos
  • Mañu Pereira, Maria del Mar
  • de Brito Robalo, Bruno M.
  • Ghini, Francesco
  • Gutiérrez Torre, Alberto|||0000-0002-5548-3359
  • Reidel Lojo, Sara Isabel|||0000-0003-1887-9538
Synthetic data generation (SDG) structured health data is increasingly promoted as a solution to longstanding barriers in health data access. It is offering the promise of privacy-preserving data reuse for research, innovation, and policy. Despite rapid technical advances, the adoption of synthetic health data in real-world settings remains limited. Shaped by challenges around data quality, representativeness, infrastructure readiness, trust, and legal uncertainty, this viewpoint draws on experiences from 7 European research initiatives within the HealthData4EU cluster to reflect on how SDG is being operationalized in practice. It synthesizes cross-project insights to highlight recurring methodological and governance tensions and to examine their implications for trust and responsible use. The analysis argues that trustworthy SDG cannot be achieved through technical optimization alone but requires alignment between evaluation practices, upstream data stewardship, regulatory clarity, and sustained stakeholder engagement. Addressing these conditions is essential for moving synthetic data from experimental pilots toward a credible and sustainable component of European health research ecosystems., This paper was funded by the Innovative Health Initiative through the SYNTHIA project. The study draws on insights from 7 research projects that form part of the HealthData4EU cluster. Projects within the cluster and their funding sources are as follows. SYNTHEMA, PHASE IV AI, SECURED, FLUTE, and AISym4MED are funded by the Horizon Europe program. PHEMS is funded by Horizon Europe and UK Research and Innovation. SYNTHIA is funded through the Innovative Health Initiative., Peer Reviewed, Article signat per 18 autors/es: Jens Declerck 1,2, MSc; Dipak Kalra 1, Prof Dr; Antti Airola 3; Ahmed Youssef Ali Amer 4; Christos Chatzichristos 5; Maria del Mar Mañu Pereira 6; Bruno M. de Brito Robalo 7; Francesco Ghini 8; Alberto Gutierrez-Torre 9; Sem Hoogteijling 10; Susanne Hultsch 11; Jan Ramon 12; Sara Reidel 13,14; Francesco Regazzoni 15,16; Luís Silva 17; Inês Silveira 17; Tsekeridou Sofia 18, PhD; Christophe Maes 1,2 / 1- The European Institute for Innovation through Health Data, Oosterzele, Belgium; 2- Unit of Medical Informatics and Statistics, Department of Public Health and Primary Care, The University of Ghent, Ghent, Belgium; 3- Department of Computing, University of Turku, Turku, Finland; 4- Innovative Medicine, Johnson & Johnson, Beerse, Belgium; 5- Department of Electrical Engineering, KU Leuven, Leuven, Belgium; 6- University Hospital Vall d'Hebron, Barcelona, Spain; 7- Department of Obstetrics and Gynaecology, Erasmus MC, University Medical Center, Rotterdam, The Netherlands; 8- Laboratory of Translational Research, Reggio Emilia, Italy; 9- Barcelona Supercomputing Center, Barcelona, Catalonia, Spain; 10- Department of Neurology and Neurosurgery, University Medical Center Utrecht Brain Center, Utrecht, The Netherlands; 11- VEIL.AI, Helsinki, Finland; 12- Institut national de recherche en sciences et technologies du numérique, Lille, France; 13- Vall d'Hebron Research Institute, University Hospital Vall d'Hebron, Barcelona, Spain; 14- BarcelonaTech (UPC), Universitat Politècnica de Catalunya, Barcelona, Spain; 15- University of Amsterdam, Amsterdam, The Netherlands; 16- Università della Svizzera italiana, Lugano, Ticino, Switzerland; 17- Laboratory for Instrumentation, Biomedical Engineering and Radiation Physics, NOVA School of Science and Technology, Caparica, Portugal; 18- Netcompany BELUX, Athens, Greece
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

Towards generalizable Federated Learning in medical imaging: a real-world case study on mammography data

  • Tzortzis, Ioannis N.
  • Gutiérrez Torre, Alberto|||0000-0002-5548-3359
  • Sykiotis, Stavros
  • Agulló López, Ferran|||0000-0002-7276-2472
  • Bakalos, Nikolaos
  • Doulamis, Anastasios
  • Doulamis, Nikolaos
  • Berral García, Josep Lluís|||0000-0003-3037-3580
Federated Learning has been rapidly gaining in popularity in medical applications, due to the increased privacy offered, since medical data doesn't need to leave the hospitals' premises for AI model training. However, a direct translation of a classic experiment to a federated one is not always straightforward. In this work, we delve into the intricacies of federated learning for a breast cancer classification tool. We compare classic model training with a federated variant, and highlight the adaptations that need to be taken care of to ensure the equivalence between the two. Specifically, we introduce the Breast Area Detection tool as an essential component of the pre-processing pipeline to enhance the robustness of Federated Learning by offering data harmonization. On top of that, we present an end-to-end Federated Learning framework that is effective for real-world data and scenarios. Among the three real-world hospitals involved in the experimental procedure, the proposed framework significantly improves performance at the first hospital, providing consistent results similar to those achieved in the classic approach. Experimental results demonstrate that the interventions introduced improved model performance by approximately 35%, aligning federated learning and centralized model performance., This paper is funded by the EU funded projects SECURED “Scaling Up Secure Processing, Anonymization And Generation Of Health Data For EU Cross Border Collaborative Research And Innovation” with grant agreement no. 101095717 and INCISIVE “A multimodal AI-based toolbox and an interoperable health imaging repository for the empowerment of imaging analysis related to the diagnosis, prediction and follow-up of cancer” with grant agreement no. 952179. This paper has been partially financed by the Spanish Ministry of Science (MICINN), the Research State Agency (AEI) and European Regional Development Funds (ERDF/FEDER) under grant agreement PID2021-126248OB-I00, Severo Ochoa Center of Excellence CEX-2021-001148-S-20-3, 2021- SGR-00478, UE, also Generalitat de Catalunya (AGAUR) under grant agreement 2021-SGR-00478., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

FRIDA: Free-rider detection using privacy attacks

  • Garcia Recasens, Pol|||0009-0008-5535-0012
  • Horváth, Ádám
  • Gutiérrez Torre, Alberto|||0000-0002-5548-3359
  • Torres Viñals, Jordi|||0000-0003-1963-7418
  • Berral García, Josep Lluís|||0000-0003-3037-3580
  • Pejó, Balázs|||0000-0002-1825-9251
Federated learning is increasingly popular as it enables multiple parties with limited datasets and resources to train a machine learning model collaboratively. However, similar to other collaborative systems, federated learning is vulnerable to free-riders — participants who benefit from the global model without contributing. Free-riders compromise the integrity of the learning process and slow down the convergence of the global model, resulting in increased costs for honest participants. To address this challenge, we propose FRIDA: free-rider detection using privacy attacks. Instead of focusing on implicit effects of free-riding, FRIDA utilizes membership and property inference attacks to directly infer evidence of genuine client training. Our extensive evaluation demonstrates that FRIDA is effective across a wide range of scenarios., Project no. 145832, implemented with the support provided by the Ministry of Innovation and Technology from the NRDI Fund, financed under the PD_23 funding scheme. Also, partially financed by the Spanish Ministry of Science (MICINN), the Research State Agency (AEI), and European Regional Development Funds (ERDF/FEDER) under grant agreement PID2021-126248OB-I00, MCIN/AEI/10.13039/501100011033/FEDER, UE, and Generalitat de Catalunya (AGAUR) under grant 2021-SGR-00478. Lastly, also funded by the European Union (Grant Agreement Nr. 101095717, SECURED Project)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publicación Artículo científico (JournalArticle)

SynthVal: A framework for validating synthetic medical images

  • Guidotti, Dario
  • Pandolfo, Laura
  • Gutiérrez Torre, Alberto|||0000-0002-5548-3359
  • López Rúbio, Omar
  • Pulina, Luca
Synthetic data is increasingly used in medical imaging to overcome data scarcity and privacy constraints. However, assessing the fidelity of synthetic images remains a critical challenge for ensuring their safe and effective use in clinical and AI applications. We present SynthVal, a Python-based framework for validating the quality of synthetic medical images through statistical comparisons in deep feature space. SynthVal extracts semantic image embeddings using transformer-based models and computes similarity metrics – including Fréchet Distance, Wasserstein Distance, and Kullback-Leibler Divergence – between real and synthetic data distributions. The framework is designed for modularity, scalability, and ease of integration into existing workflows via pip installation. We evaluate SynthVal using real images from the CSAW-CC dataset, and synthetic images produced by a model developed by the Barcelona Supercomputing Center. Our experiments systematically benchmark the influence of different feature extraction models and similarity metrics, providing practical insights for selecting validation strategies in medical image synthesis. SynthVal offers a reproducible and extensible solution for quality control in synthetic data pipelines., The work of Dario Guidotti, Laura Pandolfo, and Luca Pulina was supported by EU Project SECURED, ‘‘Scaling Up Secure Processing, Anonymization and Generation of Health Data for EU Cross-Border Collaborative Research and Innovation,’’ under Agreement 101095717. The work of Alberto Gutierrez-Torre and Omar Lopez-Rubio was supported by SECURED under Agreement 101095717; in part by Spanish Ministry of Science (MICINN), the State Research Agency (AEI), and European Regional Development Funds (ERDF/FEDER) under Grant PID2021-126248OB-I00; in part by MCIN/AEI/10.13039/501100011033/FEDER, UE; and in part by the Generalitat de Catalunya (AGAUR) under Grant 2021-SGR-00478., Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Rethinking trust in synthetic health data: lessons from 7 European research initiatives

  • Declerck, Jens
  • Kalra, Dipak
  • Airola, Antti
  • Ali Amer, Ahmed Youssef
  • Chatzichristos, Christos
  • Mañu Pereira, Maria del Mar
  • de Brito Robalo, Bruno M.
  • Ghini, Francesco
  • Gutiérrez Torre, Alberto
  • Reidel Lojo, Sara Isabel
Synthetic data generation (SDG) structured health data is increasingly promoted as a solution to longstanding barriers in health data access. It is offering the promise of privacy-preserving data reuse for research, innovation, and policy. Despite rapid technical advances, the adoption of synthetic health data in real-world settings remains limited. Shaped by challenges around data quality, representativeness, infrastructure readiness, trust, and legal uncertainty, this viewpoint draws on experiences from 7 European research initiatives within the HealthData4EU cluster to reflect on how SDG is being operationalized in practice. It synthesizes cross-project insights to highlight recurring methodological and governance tensions and to examine their implications for trust and responsible use. The analysis argues that trustworthy SDG cannot be achieved through technical optimization alone but requires alignment between evaluation practices, upstream data stewardship, regulatory clarity, and sustained stakeholder engagement. Addressing these conditions is essential for moving synthetic data from experimental pilots toward a credible and sustainable component of European health research ecosystems., This paper was funded by the Innovative Health Initiative through the SYNTHIA project. The study draws on insights from 7 research projects that form part of the HealthData4EU cluster. Projects within the cluster and their funding sources are as follows. SYNTHEMA, PHASE IV AI, SECURED, FLUTE, and AISym4MED are funded by the Horizon Europe program. PHEMS is funded by Horizon Europe and UK Research and Innovation. SYNTHIA is funded through the Innovative Health Initiative., Peer Reviewed, Article signat per 18 autors/es: Jens Declerck 1,2, MSc; Dipak Kalra 1, Prof Dr; Antti Airola 3; Ahmed Youssef Ali Amer 4; Christos Chatzichristos 5; Maria del Mar Mañu Pereira 6; Bruno M. de Brito Robalo 7; Francesco Ghini 8; Alberto Gutierrez-Torre 9; Sem Hoogteijling 10; Susanne Hultsch 11; Jan Ramon 12; Sara Reidel 13,14; Francesco Regazzoni 15,16; Luís Silva 17; Inês Silveira 17; Tsekeridou Sofia 18, PhD; Christophe Maes 1,2 / 1- The European Institute for Innovation through Health Data, Oosterzele, Belgium; 2- Unit of Medical Informatics and Statistics, Department of Public Health and Primary Care, The University of Ghent, Ghent, Belgium; 3- Department of Computing, University of Turku, Turku, Finland; 4- Innovative Medicine, Johnson & Johnson, Beerse, Belgium; 5- Department of Electrical Engineering, KU Leuven, Leuven, Belgium; 6- University Hospital Vall d'Hebron, Barcelona, Spain; 7- Department of Obstetrics and Gynaecology, Erasmus MC, University Medical Center, Rotterdam, The Netherlands; 8- Laboratory of Translational Research, Reggio Emilia, Italy; 9- Barcelona Supercomputing Center, Barcelona, Catalonia, Spain; 10- Department of Neurology and Neurosurgery, University Medical Center Utrecht Brain Center, Utrecht, The Netherlands; 11- VEIL.AI, Helsinki, Finland; 12- Institut national de recherche en sciences et technologies du numérique, Lille, France; 13- Vall d'Hebron Research Institute, University Hospital Vall d'Hebron, Barcelona, Spain; 14- BarcelonaTech (UPC), Universitat Politècnica de Catalunya, Barcelona, Spain; 15- University of Amsterdam, Amsterdam, The Netherlands; 16- Università della Svizzera italiana, Lugano, Ticino, Switzerland; 17- Laboratory for Instrumentation, Biomedical Engineering and Radiation Physics, NOVA School of Science and Technology, Caparica, Portugal; 18- Netcompany BELUX, Athens, Greece, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

FRIDA: Free-rider detection using privacy attacks

  • Garcia Recasens, Pol
  • Horváth, Ádám
  • Gutiérrez Torre, Alberto
  • Torres Viñals, Jordi
  • Berral García, Josep Lluís
  • Pejó, Balázs
Federated learning is increasingly popular as it enables multiple parties with limited datasets and resources to train a machine learning model collaboratively. However, similar to other collaborative systems, federated learning is vulnerable to free-riders — participants who benefit from the global model without contributing. Free-riders compromise the integrity of the learning process and slow down the convergence of the global model, resulting in increased costs for honest participants. To address this challenge, we propose FRIDA: free-rider detection using privacy attacks. Instead of focusing on implicit effects of free-riding, FRIDA utilizes membership and property inference attacks to directly infer evidence of genuine client training. Our extensive evaluation demonstrates that FRIDA is effective across a wide range of scenarios., Project no. 145832, implemented with the support provided by the Ministry of Innovation and Technology from the NRDI Fund, financed under the PD_23 funding scheme. Also, partially financed by the Spanish Ministry of Science (MICINN), the Research State Agency (AEI), and European Regional Development Funds (ERDF/FEDER) under grant agreement PID2021-126248OB-I00, MCIN/AEI/10.13039/501100011033/FEDER, UE, and Generalitat de Catalunya (AGAUR) under grant 2021-SGR-00478. Lastly, also funded by the European Union (Grant Agreement Nr. 101095717, SECURED Project)., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

SynthVal: A framework for validating synthetic medical images

  • Guidotti, Dario
  • Pandolfo, Laura
  • Gutiérrez Torre, Alberto
  • López Rúbio, Omar
  • Pulina, Luca
Synthetic data is increasingly used in medical imaging to overcome data scarcity and privacy constraints. However, assessing the fidelity of synthetic images remains a critical challenge for ensuring their safe and effective use in clinical and AI applications. We present SynthVal, a Python-based framework for validating the quality of synthetic medical images through statistical comparisons in deep feature space. SynthVal extracts semantic image embeddings using transformer-based models and computes similarity metrics – including Fréchet Distance, Wasserstein Distance, and Kullback-Leibler Divergence – between real and synthetic data distributions. The framework is designed for modularity, scalability, and ease of integration into existing workflows via pip installation. We evaluate SynthVal using real images from the CSAW-CC dataset, and synthetic images produced by a model developed by the Barcelona Supercomputing Center. Our experiments systematically benchmark the influence of different feature extraction models and similarity metrics, providing practical insights for selecting validation strategies in medical image synthesis. SynthVal offers a reproducible and extensible solution for quality control in synthetic data pipelines., The work of Dario Guidotti, Laura Pandolfo, and Luca Pulina was supported by EU Project SECURED, ‘‘Scaling Up Secure Processing, Anonymization and Generation of Health Data for EU Cross-Border Collaborative Research and Innovation,’’ under Agreement 101095717. The work of Alberto Gutierrez-Torre and Omar Lopez-Rubio was supported by SECURED under Agreement 101095717; in part by Spanish Ministry of Science (MICINN), the State Research Agency (AEI), and European Regional Development Funds (ERDF/FEDER) under Grant PID2021-126248OB-I00; in part by MCIN/AEI/10.13039/501100011033/FEDER, UE; and in part by the Generalitat de Catalunya (AGAUR) under Grant 2021-SGR-00478., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publicación Artículo científico (JournalArticle)

Towards generalizable Federated Learning in medical imaging: a real-world case study on mammography data

  • Tzortzis, Ioannis N.
  • Gutiérrez Torre, Alberto
  • Sykiotis, Stavros
  • Agulló López, Ferran
  • Bakalos, Nikolaos
  • Doulamis, Anastasios
  • Doulamis, Nikolaos
  • Berral García, Josep Lluís
Federated Learning has been rapidly gaining in popularity in medical applications, due to the increased privacy offered, since medical data doesn't need to leave the hospitals' premises for AI model training. However, a direct translation of a classic experiment to a federated one is not always straightforward. In this work, we delve into the intricacies of federated learning for a breast cancer classification tool. We compare classic model training with a federated variant, and highlight the adaptations that need to be taken care of to ensure the equivalence between the two. Specifically, we introduce the Breast Area Detection tool as an essential component of the pre-processing pipeline to enhance the robustness of Federated Learning by offering data harmonization. On top of that, we present an end-to-end Federated Learning framework that is effective for real-world data and scenarios. Among the three real-world hospitals involved in the experimental procedure, the proposed framework significantly improves performance at the first hospital, providing consistent results similar to those achieved in the classic approach. Experimental results demonstrate that the interventions introduced improved model performance by approximately 35%, aligning federated learning and centralized model performance., This paper is funded by the EU funded projects SECURED “Scaling Up Secure Processing, Anonymization And Generation Of Health Data For EU Cross Border Collaborative Research And Innovation” with grant agreement no. 101095717 and INCISIVE “A multimodal AI-based toolbox and an interoperable health imaging repository for the empowerment of imaging analysis related to the diagnosis, prediction and follow-up of cancer” with grant agreement no. 952179. This paper has been partially financed by the Spanish Ministry of Science (MICINN), the Research State Agency (AEI) and European Regional Development Funds (ERDF/FEDER) under grant agreement PID2021-126248OB-I00, Severo Ochoa Center of Excellence CEX-2021-001148-S-20-3, 2021- SGR-00478, UE, also Generalitat de Catalunya (AGAUR) under grant agreement 2021-SGR-00478., Peer Reviewed, Postprint (published version)