Name: Ivan Štajduhar
Affiliation: University of Rijeka – Faculty of Engineering, Rijeka, Croatia
Primary research interest: biomedical image processing, multimodal learning
Title of the lecture: “RadiologyNET: A large-scale multimodal medical radiology database for transfer learning”
Keywords: medical radiology imaging, transfer learning, foundation models
Summary: I will discuss the problem of developing foundation models for aiding building decision support systems in medical radiology: (1) developing a pipeline for annotating a large database of medical radiology images from CHC Rijeka; (2) using the annotated dataset to train select deep learning model architectures; (3) validating the use of transfer learning on several open benchmarks.