Évaluation de la robustesse du recalage rigide multimodal entre échographie et imagerie par résonance magnétique dans les systèmes de neurochirurgie guidée par image

Abstract

In this research thesis, we developed and implemented a detailed methodological approach aimed at evaluating the robustness of multimodal registration between Magnetic Resonance Imaging (MRI) and Ultrasound (US) in the specific context of neuronavigation. MRI-US registration is a critical component for compensating intraoperative tissue deformations, which create discrepancies between preoperative data and the actual conditions in the operative field. The integration of intraoperative ultrasound is justified by its ability to update preoperative MRI images by adjusting the observed deformations. The robustness of MRI-US registration was assessed by simulating various perturbations that may arise in a clinical context, affecting the geometry and characteristics of brain tissue. These perturbations are intended to challenge the registration’s ability to maintain integrity and accuracy under adverse conditions. The quantification of registration error is based on the identification of precise anatomical landmarks, identified by surgeons both before and after the introduction of these perturbations, enabling a robust and objective evaluation of registration performance. It is important to emphasize that this study focused on a single registration algorithm, thus providing and validating an exhaustive methodology for assessing its robustness. Although centered on this specific case, the experimental approach presented here offers a more general framework—one that can be replicated for other registration methods—and contributes to a deeper understanding of the factors influencing multimodal registration reliability. This thesis provides an in-depth analysis of MRI-US registration robustness, crucial for neuronavigation, and offers significant prospects for advancing registration techniques. This methodological contribution aims to ensure high-precision surgical guidance, which is essential for successful neurosurgical interventions and for minimizing operative risks.

Date	21 Feb 2025
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Simon Drouin (Supervisor)