Skip to content. Skip to navigation
CIM Menus
 

Geometric Paradigms for MRI Analysis With Applications to Cardioimaging and Neuroimaging


Luc Florack
Eindhoven University of Technology

April 24, 2015 at  2:30 PM
McConnell Engineering Building, MC12

Abstract: Magnetic Resonance Imaging (MRI) is a non-invasive and highly versatile imaging modality, with unprecedented capabilities for disclosing in-vivo anatomical and functional information. Imaging protocols as well as image analysis paradigms are traditionally set up so as to convey this information to the radiologist in a format geared towards visual interpretation, supported by software platforms for interactive “data visualization” and add-ons for quantification of visual structures. The advent of increasingly sophisticated acquisition schemes, producing complex highcodimensional images, prohibits a direct visual approach, calling for a radically new paradigm, with a more compelling role for theory to address notorious inverse problems. Examples of these are “diffusion weighted MRI” and spatial modulation of magnetization, or “tagging MRI”. The data produced in these cases are “non-visual images” that admitquantification and visually exploration (“visual analytics”) only by virtue of an priori axiomatic framework for their interpretation. The necessary axiomatics emerges from the interplay of the physical sciences (mathematics, physics), and often leads to a revaluation of established theories originally developed in a different context. Image researchers worldwide are starting to embrace the new paradigm, trying to “see what one can understand” rather than trying to “understand what one can see”. In this presentation I will illustrate the above rationale in the context of a differential geometric framework for the analysis of diffusion and tagging MRI data with “off-the-shelf” physics and mathematics from early twentieth century.