Real World Federated Learning with a Knowledge Distilled Transformer for Cardiac CT Imaging
This addresses the challenge of utilizing unlabeled data in federated learning for cardiac CT analysis, which is incremental as it builds on existing methods to handle partial labels in a real-world setting.
The paper tackled the problem of partially labeled datasets in real-world federated learning for cardiac CT imaging by proposing a two-step semi-supervised strategy that distills knowledge from CNNs into a transformer, improving predictive accuracy and outperforming UNet-based models in generalizability on downstream tasks across eight hospitals with n=8,104 cases.
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.