Learning of latent spaces dedicated to the segmentation of medical images
Speaker: Olivier Bernard
Title: Learning of latent spaces dedicated to the segmentation of medical images
Abstract: Deep learning methods have become very relevant approaches in the medical field, with significant success in different applications such as image segmentation.
In this context, variational auto-encoder techniques have proven to be powerful techniques for an efficient representation of problems in reduced dimensional spaces, allowing relevant processing / analysis. During my seminar, I will illustrate the power of VAE through the application of echocardiographic image segmentation. In particular, I will show you how we have exploited this formalism in order to 1) provide guarantees on the segmented anatomical structures; ii) provide guarantees on the temporal consistency of the segmented shapes; iii) model the uncertainty of the segmentation algorithm.
Bio: Olivier Bernard received the Electrical Engineering degree and the Ph.D. degree from the University of Lyon (INSA), France, in 2003 and 2006, respectively. In 2007, he was a Postdoctoral Fellow with the Biomedical Imaging Group at the Federal Polytechnic Institute of Lausanne, EPFL, Switzerland. He is currently an Associate Professor with the University of Lyon (INSA) and the CREATIS laboratory, France, where he is also head of a research team. His current research interests include image analysis through deep learning techniques with applications in cardiovascular imaging and blood flow imaging. Dr. Bernard has been Associate Editor of the IEEE TRANSACTIONS ON IMAGE PROCESSING.