Enhancing Assessment of Cardiovascular Hemodynamics using Global Federated Learning AI

Blood flow within the heart and vessels is optimally regulated in healthy individuals but can become inefficient and complicated in diseased states.

Time-resolved three-dimensional phase-contrast magnetic resonance imaging, commonly known as 4D flow MRI, has proven to be an essential tool for examining these changes in blood flow.

However, its widespread clinical adoption is limited by prolonged measurement times and labor-intensive processing requirements.Recent advancements in artificial intelligence (AI) hold significant promise for improving medical imaging, yet these advancements face challenges due to the diversity, site-specific nature, and data privacy regulations of medical imaging data.

Federated learning offers a solution by enabling collaborative training of AI models while keeping the data decentralized.This proposal aims to develop a multi-task federated-learning AI platform for medical imaging, initially focusing on AI-enhanced post processing of 4D flow MRI analysis.

Specifically, we will establish a federated-learning AI platform to facilitate AI-assisted automated image processing and flow quantification.This initiative and common federated-learning AI platform will foster a lasting bilateral collaboration between Sweden and Korea, concentrating on AI-powered 4D flow MRI and the exchange of expertise in handling sensitive medical data, medical imaging analysis, fluid mechanics, and AI.