https://scholars.lib.ntu.edu.tw/handle/123456789/638633
標題: | ConDistFL: Conditional Distillation for Federated Learning from Partially Annotated Data | 作者: | Wang, Pochuan Shen, Chen WEICHUNG WANG Oda, Masahiro CHIOU-SHANN FUH Mori, Kensaku Roth, Holger R. |
關鍵字: | Abdominal CT | Federated learning | Multi-organ and tumor segmentation | Partially labeled datasets | 公開日期: | 1-一月-2023 | 卷: | 14393 | 來源出版物: | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | 摘要: | Developing a generalized segmentation model capable of simultaneously delineating multiple organs and diseases is highly desirable. Federated learning (FL) is a key technology enabling the collaborative development of a model without exchanging training data. However, the limited access to fully annotated training data poses a major challenge to training generalizable models. We propose “ConDistFL”, a framework to solve this problem by combining FL with knowledge distillation. Local models can extract the knowledge of unlabeled organs and tumors from partially annotated data from the global model with an adequately designed conditional probability representation. We validate our framework on four distinct partially annotated abdominal CT datasets from the MSD and KiTS19 challenges. The experimental results show that the proposed framework significantly outperforms FedAvg and FedOpt baselines. Moreover, the performance on an external test dataset demonstrates superior generalizability compared to models trained on each dataset separately. Our ablation study suggests that ConDistFL can perform well without frequent aggregation, reducing the communication cost of FL. Our implementation will be available at https://github.com/NVIDIA/NVFlare/tree/main/research/condist-fl. |
URI: | https://scholars.lib.ntu.edu.tw/handle/123456789/638633 | ISBN: | 9783031474002 | ISSN: | 03029743 | DOI: | 10.1007/978-3-031-47401-9_30 |
顯示於: | 資訊工程學系 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。