Olivier Saut (University of Bordeaux, France)

Title: Data assimilation in tumor growth modeling: towards patient calibrated models using imaging devices

2:30 pm / CH 445

Abstract: Numerous mathematical models exist to describe cancer growth. Their main purposes are to dissect the various mechanisms involved in the disease, to evaluate or predict the growth or the effects of a therapy. Yet, in most cases, the quantitative results obtained through these models are restraint to simple setups or in-vitro studies. Indeed, these models have so many parameters (which are e.g tuning the interplays between the various phenomena influencing the disease) that it is practically difficult or even

impossible to recover them from experimental data or clinical routine. Furthermore, for obvious ethical reasons, the parameter identification process should not require any invasive technique on patients. Hence to study a patient-specific case, one has a find a way to calibrate an adequate model exclusively with the information available to clinicians. 
Meanwhile, clinicians are routinely following the evolution of the disease of a patient thanks to imaging devices. These images are almost our only source of information when trying to recover the parameters of a mathematical model.This makes continuous spatial models particularly well adapted to be coupled with these images. 
In the talk, we are interested in evaluating the calibration of mathematical models on patient images thus allowing their use for clinical applications. Several challenges are raised: dealing with or understanding the outputs of medical imaging devices, adapting or designing new models with the calibration process in mind, recovering the modeling parameters... These challenges and ways of overcoming them in some particular cases will be presented in this talk.