An Asynchronous Federated Learning Platform for Lung Adenocarcinoma Grading via Foundation Models and Structural Pruning
Paper · https://arxiv.org/abs/2611.14590
arXiv:2611.14590v1 Announce Type: new Abstract: Classification of lung adenocarcinoma into poorly differentiated (PD) and well-to-moderately differentiated (WMD) subtypes plays an important role in patient survival; however, consistent classification is difficult due to inter-observer variability among pathologists. Collecting data from multiple hospitals is also challenging due to patient privacy concerns. This study proposes a web-based federated learning system that shares only locally trained model weights without sharing patient data between institutions. The UNI2 foundation model and CLAM-SB were used to classify pathological whole slide images. A four-client federated learning system was constructed, and the NLST cohort was reserved exclusively for external validation. In centralized evaluation, joint training with CPTAC and TCGA outperformed single-institution training on external validation, confirming the necessity of multi-institutional data integration. FedAvg, FedProx, FedBN, and FedNova were compared as federated learning algorithms under identical experimental conditions. Test AUC ranked as FedBN, FedProx, FedNova, and FedAvg. FedNova showed the most balanced classification performance between PD and WMD. L1-norm structural pruning at 20%, 30%, and 40% sparsity was applied to reduce model transmission size. FedAvg, FedProx, and FedBN showed a tendency of decreasing Test AUC under pruning, while FedNova achieved improved Test AUC across all sparsity levels. A web-based application was developed supporting slide upload, local training, AI classification, and attention heatmap-based visualization. The combination of FedNova and 40% pruning was confirmed as a practical approach that improves communication efficiency and classification performance while protecting patient privacy.
First seen by Oriel on .
AI summary
The study presents a web-based federated learning platform for privacy-preserving lung adenocarcinoma grading, finding FedNova with 40% structural pruning improves communication efficiency and classification performance.
- The four-client system uses UNI2 and CLAM-SB to classify pathological whole slide images, with the NLST cohort reserved for external validation.
- Centralized joint training on CPTAC and TCGA outperformed single-institution training externally, demonstrating the benefit of multi-institutional data integration.
- Test AUC ranked FedBN, FedProx, FedNova, then FedAvg, while FedNova provided the most balanced performance across PD and WMD classes.
- L1-norm structural pruning at 20%, 30%, and 40% reduced transmission size; FedNova improved Test AUC at every sparsity level while other algorithms generally declined.
- The application supports slide uploads, local training, AI classification, and attention heatmap visualization without sharing patient data between institutions.
#federated learning #computational pathology #foundation models #structural pruning
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| arXiv AI Papers | oai:arXiv.org:2611.14590v1 | https://arxiv.org/abs/2611.14590 Read discussion |