FedDA-TSformer achieved a Dice Similarity Coefficient of 0.842 for left ventricular endocardium segmentation, demonstrating superior accuracy in a cohort of 150 subjects from multiple centers.
Cohort (n=150)
Yes
The proposed federated learning framework and adaptive normalization techniques improve model performance and generalization across various deep learning tasks, including cardiac SPECT image segmentation.
Effect estimate: null (95% CI null)
p-value: p=null
FedDA-TSformer provides a robust, privacy-preserving solution for multi-center left ventricular segmentation, outperforming traditional FedAvg in handling domain shifts. By leveraging the TimeSformer architecture and domain adaptation mechanisms, the framework ensures spatial-temporal consistency and data security across heterogeneous clinical sites. Despite current limitations regarding communication overhead and its focus on a small SPECT-only dataset, this study establishes a scalable foundation for collaborative cardiac diagnosis. Future work will prioritize model compression, asynchronous updates, and cross-modality generalization to CT and MRI to enhance its practicality in resource-constrained environments.
Huang et al. (Sun,) conducted a cohort in Left Ventricular Segmentation (n=150). FedDA-TSformer vs. FedAvg was evaluated on Dice Similarity Coefficient for left ventricular segmentation (null, 95% CI null, p=null). FedDA-TSformer achieved a Dice Similarity Coefficient of 0.842 for left ventricular endocardium segmentation, demonstrating superior accuracy in a cohort of 150 subjects from multiple centers.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: