Real world federated learning with a knowledge distilled transformer for cardiac CT imaging

Federated learning is a renowned technique for utilizing decentralized data while preserving privacy. However, real-world applications often face challenges like partially labeled datasets, where only a few locations have certain expert annotations, leaving large portions of unlabeled data unused. Leveraging these could enhance transformer architectures' ability in regimes with small and diversely annotated sets. We conduct the largest federated cardiac CT analysis to date (n = 8, 104) in a real-world setting across eight hospitals. Our two-step semi-supervised strategy distills knowledge from task-specific CNNs into a transformer. First, CNNs predict on unlabeled data per label type and then the transformer learns from these predictions with label-specific heads. This improves predictive accuracy and enables simultaneous learning of all partial labels across the federation, and outperforms UNet-based models in generalizability on downstream tasks. Code and model weights are made openly available for leveraging future cardiac CT analysis.

Authors

Institutions

Publication Details

Journal
npj Digital Medicine
Published
2025-02-06
DOI
https://doi.org/10.1038/s41746-025-01434-3
Citations
13
Primary Topic
Privacy-Preserving Technologies in Data
Type
article
Field-Weighted Citation Impact
32.25

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Real world federated learning with a knowledge distilled transformer for cardiac CT imaging

Malte Tölle, Norbert Frey, Clemens Scherer, Stefan Simm et al.
13 citations
npj Digital Medicine
Privacy-Preserving Technologies in Data
32.25
article

Real world federated learning with a knowledge distilled transformer for cardiac CT imaging

Malte Tölle, Norbert Frey, Clemens Scherer, Stefan Simm, Anja Hennemuth, Andreas Leha, Alexander Meyer, Gerhard‐Paul Diller, Peter Bannas, Tim Seidler, Stefan Groß, Stefan Orwat, Lars Kaderali, Nina Krüger, Tim Friede, Halvar Kelm, Moritz Seiffert, Eike Nagel, Jan Moritz Seliger, Sandy Engelhardt, Simon S. Martin, Florian André, Sebastian Eble, Moritz Bednorz, Philipp Garthe
article en
13 citations

Abstract

Federated learning is a renowned technique for utilizing decentralized data while preserving privacy. However, real-world applications often face challenges like partially labeled datasets, where only a few locations have certain expert annotations, leaving large portions of unlabeled data unused. Leveraging these could enhance transformer architectures' ability in regimes with small and diversely annotated sets. We conduct the largest federated cardiac CT analysis to date (n = 8, 104) in a real-world setting across eight hospitals. Our two-step semi-supervised strategy distills knowledge from task-specific CNNs into a transformer. First, CNNs predict on unlabeled data per label type and then the transformer learns from these predictions with label-specific heads. This improves predictive accuracy and enables simultaneous learning of all partial labels across the federation, and outperforms UNet-based models in generalizability on downstream tasks. Code and model weights are made openly available for leveraging future cardiac CT analysis.

npj Digital MedicineVol. 8(1)
Goethe University Frankfurt (DE), Universität Hamburg (DE), Zimmer Biomet (Netherlands) (NL), Heidelberg University (DE), Universitätsmedizin Greifswald (DE), University Hospital Heidelberg (DE), LMU Klinikum (DE), Humboldt-Universität zu Berlin (DE), University Hospital Münster (DE), Deutsches Herzzentrum der Charité (DE), University Medical Center Hamburg-Eppendorf (DE), German Center for Infection Research (DE), Universitätsmedizin Göttingen (DE), German Centre for Cardiovascular Research (DE), Kerckhoff Klinik (DE), Fraunhofer Institute for Digital Medicine (DE), Freie Universität Berlin (DE), Charité - Universitätsmedizin Berlin (DE), Ludwig-Maximilians-Universität München (DE)
Deutsches Krebsforschungszentrum, Deutsches Zentrum für Herz-Kreislaufforschung, Klaus Tschira Stiftung
Partnerships for the goals
Openalex Percentile: Top 0%
Privacy-Preserving Technologies in Data
32.25
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.