Virtual image correlation of magnetic resonance images for 3D geometric modelling of pelvic organs

Numerical simulation of pelvic system could lead to a better understanding of common pathology, through objective and reliable analyses of pelvic mobility, according to mechanical principles. In clinical context, patient-specific simulation has the potential for a proper patient-personalized cure. For this purpose, a simulable 3D geometrical model, well suited to patient anatomy, is required. However, the geometric modelling of pelvic system from medical images (MRI) is a complex operator-dependent and time-consuming process, not adapted to patient-specific applications. This paper is addressing this challenging computational problem. The objective is to develop a technique providing a smooth, consistent and readily usable 3D geometrical model, seamlessly from image to simulation. In this paper, we use a generic topologically-simplified B-Spline model to represent pelvic organs. The presented paper develops a Virtual Image Correlation (VIC) method to find the best correlation between the geometry and the image. The final reconstructed geometrical model is to be compatible with meshing and Finite Element (FE) simulation. Then, a variety of tests are performed to prove the concept, through both prototypical and pelvic models. Finally, since the pelvic system is complex, including structures hardly identifiable in MRI, some feasible solutions to introduce more complex pelvic models are also discussed.