4TU precision medicine
Prof.dr. Michel Versluis
Recorded during: The 2020 TechMed Event | more info
Date recording: 28 October 2020
Abstract
Medical imaging offers tools for screening, diagnosis, image-guided interventions, and patient follow-up. These tools should be fast, safe, and precise. Artificial intelligence (AI) has made a successful entry into medical imaging by offering greater image processing speeds and accuracy by using unprecedented computing power. AI is revolutionizing the clinical workflow with automation, faster screening, and improved early detection rates. However, data-driven AI is solely based on post-reconstruction data (the images), making insufficient use of the physical information that is contained in the raw data. Model-driven AI circumvents this black-box approach by introducing fundamental science into the training of neural networks as to constrain the learning on the imaging data. This unique combination of physics and computer science enables novel functional and multimodal precision imaging, that is a faster, safer and ultimately personalized tool for clinical decision support.