Seminar June 6 TH, 2023
Hours: 16:30 - 18:00
Qiang Zhang
Title: Deep learning in advancing contrast-free cardiac MRI
Abstract: Myocardial scarring is a common final pathway for most cardiac diseases. Clinically, they are assessed with MRI, using an imaging method called late gadolinium enhancement, or LGE. However, LGE requires contrast injection, which increases scan time and costs, and has contraindications. A new, AI-based technology called “virtual native enhancement” (VNE) offers a contrast-free alternative. This combines MR images that do not normally need contrast injections, and uses AI to train machines to generate a “virtual LGE” image. During this presentation, Dr Zhang will introduce the latest advancements in VNE, and discuss the next steps towards its application in clinical practice.
Bio: Qiang Zhang is a deep learning scientist and BHF CRE Intermediate Fellow at the RDM Division of Cardiovascular Medicine, University of Oxford. Working with clinicians and MR scientists on a day-to-day basis, his primary research objective is to bridge the gap between machine learning and clinical needs, and translate AI solutions into real-world clinical practice. His recent research focus has been on AI enhancement of MRI for faster, safer, and more informative heart imaging.
Guillermo Sahonero Alvarez
Title: Optimizing Low-Field MRI sequences for high-field contrast equivalent images
Abstract: Low-field MRI has recently gained significant interest due to its potential to democratize access to MR-based medical analysis. Nevertheless, the adoption of this technology faces several challenges which range from technical aspects, such as image reconstruction and denoising, to a medical reluctance at embracing relatively-new or not-so-known technologies. MRI systems at 1.5T and 3T are the most common, which is why current work has focused on developing and improving image processing techniques to denoise and enhance low-field images, so they can look like high-field images. Unfortunately, little work has been done in sequence optimization and adaptation to acquire low-field images that could potentially be as useful as images acquired in higher fields. In this ongoing work, we propose an approach to generate low-field sequences to reduce the relative difference between high-field and low-field images. In this way, we expect to systematically find adaptations to make low-field MRI systems more attractive to medical personnel.
Bio: Guillermo Sahonero is a Ph.D. Student at the Institute for Biological and Medical Engineering from UC Chile and iHEALTH. Previously, he was a full-time lecturer and researcher at UCB (Bolivia) where he worked on developing real-time EEG-based BCIs and gait recognition systems for smart surveillance using edge computing paradigms and deep learning. He also contributed to the development of the IEEE Standard for a Unified Terminology for Brain-Computer Interfaces. His research interests are focused on MRI, Deep Learning, BCI, and Computer Vision.
