Image-Driven Simulation – IDeaS

Design of a closed system where medical imaging and the radiologist gestures are used to refine the parameters of the numerical simulation in order to increase its predictive capability (and therefore to limit the acquisition rate of the fluoroscopy)

Many vascular pathologies can now be treated in a minimally invasive way thanks to interventional radiology. Instead of open surgery, it allows to reach the lesion of the arteries with therapeutic devices through a catheter. Navigation inside the vascular network is performed using fluoroscopy as image guidance. Fluoroscopy is a low-dose Xray acquisition protocol that enables real-time visualization of the devices progressing within the patient. This definition itself unveils the major limitations of fluoroscopy. Even if performed at low dose rates, the length of interventional radiology procedures may lead to a high absorbed dose for the patient. Furthermore, low-dose Xray cannot but provide a poor visual feedback, with a low signal to noise ratio and low contrast, and where blood vessels cannot be seen but at times at the cost of a contrast medium injection. And finally, the third dimension, the depth, is completely lost within such a 2D, planar, imagery. The next major improvement brought to interventional radiology will definitely both limit radiation and provide a full 3D and real-time visual intra-operative feedback. One important path of research investigates real-time MRI and associated material for the tools. However, we believe that coupling mechanically-based simulation and on-line medical image acquisition shows more and shorter-time promises to solve such an apparent paradox.

This proposal targets research and development on Image-driven simulation: a coupled system of interactive computer-based simulation and on-line medical image acquisitions. The main idea is to use the medical images as references to continuously refine the model parameters of the simulation. Our guideline is to follow a sequential statistical filtering approach to fuse such heterogeneous data. This approach calls for an improved knowledge of the statistical behavior of simulation, which will be addressed through experimental studies.

Constrained on one hand by a sophisticated physical behavior model and, on the other hand, to fit the actual images acquired online, the simulated interventional devices should faithfully depict the 3D aspect of the actual ones. Progress is therefore expected to solve the single-view reconstruction problem in a dynamic setup. A second expected outcome of this proposal is to increase our understanding of the main causes for simulation outputs to deviate from real data. Directions of research for improving the realism of computer-based simulation should thereby emerge. First steps will be made towards groundbreaking clinical applications. The simulation will provide more accurate predictions of interventional tools motion within arteries enabling to focus the image field of view or reduce the acquisition rate by using the simulation outputs as interpolating states, resulting in a drastic reduction of radiation exposure. We strongly believe this project is an enabler for practitioners to benefit from an interactive fully 3D visual feedback for their surgical gesture. Moreover additional data computed by the simulation (internal energy of the device or pressure on the arterial wall for instance) could be used to help surgeons during the procedure to prevent bad gestures.