Applications of intensive HIFU simulation based on surrogate models using the CIVA HealthCare platform
Résumé
High Intensity Focused Ultrasound (HIFU) uses the energy of an ultrasonic beam to cause localized heating to alter or destroy the targeted tissues. Due to its limited collateral effects, this minimally invasive therapy is, in its field of application, an alternative to surgical treatments and is rapidly expanding in terms of both clinical research and industrial development. In this context, the use of simulation is a crucial element to reduce both the development time and prototype costs, as well as the number of preclinical and clinical experiments. To this aim, CEA-LIST has been developing the simulation platform, CIVA HealthCare in partnership with INSERM for several years, which uses optimized and validated modules for the simulation of a whole HIFU therapy protocol. However, multi-parametric studies for treatment optimisation (probe design, protocol definition, etc.), sensitivity analysis, and tissue characterisation are very time-consuming, since they involve many simulations. We propose an alternative solution of replacing the computationally expensive direct solver with a surrogate model built from a database of simulation results. The procedure breaks down into two stages First, the database is adaptively constructed to maximise the fidelity of the associated interpolator computed from the pairs (Input/Output) of the database. Then in the second step, the interpolator, also called surrogate model or metamodel, enables the generation of results in near real time for configurations covered by the range of the database. This procedure and its related tools have been implemented in CIVA-HealthCare. To highlight the potential offered by such intensive HIFU simulation based on surrogate models strategy, two applications are presented: The first concerns real-time pressure field computation, and the second is based on thermal simulations and deals with sensitivity analyses according to uncertainties in tissues parameters.
Origine : Fichiers produits par l'(les) auteur(s)
licence : CC BY - Paternité
licence : CC BY - Paternité
