A committee of experts from leading U.S. medical centers and research institutes, is leveraging NVIDIA-powered federated learning, to enhance AI models for tumour segmentation in medical imaging.
Isha Salian, writing for NVIDIA, reports that the Society for Imaging Informatics and Medicine (SIIM) Machine Learning Tools and Research Subcommittee is spearheading this initiative. The project aims to evaluate the impact of federated learning and AI-assisted annotation on training AI models for cancer detection.
John Garrett, associate professor of radiology at the University of Wisconsin–Madison, emphasised the necessity of federated learning in medical AI development. "Adopting federated learning to build and test models at multiple sites at once is the only way, practically speaking, to keep up. It's an indispensable tool," Garrett stated.
The team, which includes collaborators from institutions such as Case Western, Georgetown University, and the Mayo Clinic, is using NVIDIA FLARE (NVFlare), an open-source framework for federated learning. NVIDIA provided RTX A5000 GPUs to support the project through its Academic Grant Program.
The study focuses on renal cell carcinoma, with six participating medical centers each providing data from about 50 medical imaging studies. Yuankai Huo, assistant professor at Vanderbilt University, explained the federated learning process: "The idea with federated learning is that during training we exchange the model rather than exchange the data."
The project's outcomes, including methodology, annotated datasets, and pretrained models, will be published to support further research in the field. This collaborative effort represents a significant step forward in using AI and federated learning to improve cancer detection while maintaining data privacy and security.