Heidelberg University Hospital (AICM) (Toelle, Engelhardt et al.)
Model weights not public. Contact creators for more information.
Largest federated cardiac CT analysis to date (n=8,104 scans) across a real-world federation of German university hospitals, addressing partially-labeled data across sites via a two-step semi-supervised knowledge-distillation strategy: task-specific CNNs first predict on unlabeled data per label type, then a SWIN-UNETR transformer learns from these predictions with label-specific heads. Learns a single federated model that simultaneously predicts TAVI-relevant landmarks (aortic hinge points, coronary ostia, membranous septum) and calcification from cardiac CT, improving generalizability over UNet-based baselines on downstream tasks.
Architecture
Transformer
SWIN-UNETR (Swin Transformer encoder + UNet-style decoder) trained via federated learning across multiple hospital sites, with knowledge distillation from task-specific CNN teachers on unlabeled data to handle partial/heterogeneous label availability per site
Framework
PyTorch
Added to catalog
2026-08-10
Federated German Hospital Cardiac CT Cohort (pre-TAVI)
Real-world federated cardiac CT cohort (n=8,104 scans) from a federation of German university hospitals, with partial/heterogeneous label availability per site (aortic hinge points, coronary ostia, membranous septum, calcification) prior to TAVI.
Aortic hinge-point, coronary-ostia, and membranous-septum landmark detection from cardiac CT (TAVI planning)
Cardiac/aortic calcification segmentation from cardiac CT