Discrete biomedical perception: efficiency, modularity & scalability - MISS 2014 (06 Pt02)

Prof. Nikos Paragios (Ecole Centrale Paris, France) was one of the speakers of Medical Imaging Summer School (Sicily, July 28-August 1, 2014), here you can find his lecture 1: "Discrete biomedical perception: efficiency, modularity & scalability".

Abstract. Perception, interpretation and understanding of images is a well studied problem. The process consists of defining a set of parameters that should be determined given the observations. Discrete modeling consists of associating this inference process as a graph optimization problem where parameters correspond to nodes, labels to a discrete sampling of the optimization space and connectivity on codependencies between variables. 

This lecture provides the theoretical foundations of a generic mathematical formalism exploiting low as well as higher order probabilistic graphical models for biomedical image analysis. Furthermore, efficient generic/scalable methods for their inference is presented demonstrating the extreme potentials of such an approach. 

Last but not least, concrete biomedical imaging examples are considered like shape registration, organ and tumor segmentation, high level reasoning to better illustrate the interest of such a perception paradigm.

Shooting: Lorenzo Di Silvestro, Mauro Sodano
Editing and post-production: Lorenzo Di Silvestro, Agata Ventura