Section: Overall Objectives

Challenges

Real-Time Accurate Biophysical Models

The principal objective of this scientific challenge is the modeling of the operative field, i.e. the anatomy and physiology of the patient that will be directly or indirectly targeted by a medical intervention. This requires to describe various biophysical phenomena such as soft-tissue deformation, fluid dynamics, electrical propagation, or heat transfer. These models will help to simulate the reaction of the patient's anatomy to the procedure, but also represent the behavior of complex organs such as the brain, the liver or the heart. A common requirement across these developments is the need for (near) real-time computation.

Multi-Model Simulations

The notion of multi-model simulation encompasses two ideas. First, it captures the idea that organs are not isolated in the body and therefore are constantly interacting with the surrounding anatomy through various types or constraints. Second it translates the need to build complex models from "simpler" ones that interact with each other at a functional level, forming coupled systems (of which vascularized organs or an electro-mechanical model of the heart are good examples). As we start building larger simulations or models, computational efficiency will become of prime importance. That is why a part of our research consist in developing new strategies for parallel computing that will be adapted for multi-model simulations.

Simulation-guided Therapy

Image-guided therapy is a recent area of research that has the potential to bridge the gap between medical imaging and clinical routine by adapting pre-operative data to the time of the procedure. Several challenges are related to image-guided therapy (e.g. fusion of multi-modality images, registration, segmentation, reconstruction, ...) but the principal one consists in aligning pre-operative images onto the patient. As most procedures deal with soft-tissues, elastic registration techniques are necessary to perform this step. Recently, registration techniques started to account for soft tissue deformation using physically-based methods [30] . Yet, several limitations still hinder the use image-guided therapy in clinical routine. First, as registration methods become more complex, their computation times increase, thus lacking responsiveness. Second, as we have seen in previous sections, many factors influence the deformation of soft-tissues, from patient-specific material properties to boundary conditions with surrounding anatomy. A typical illustration of this problem, in the field of neurosurgery, is the brain shift that takes place when the skull is opened and the intracranial pressure drops [37] . It is clear that several of the techniques we are developing for interactive simulation could be applied to pre-operative images in order to provide added feedback during a procedure. In particular, several aspects, besides modeling brain tissue deformation, come into play during brain shift, such as contact between the brain and the skull, the influence of the vascular network, etc. We have already illustrated this potential in the context of coil embolization [28] .