Section: New Results

Image-based diagnoses

In the context of the female pelvic medicine, image-based diagnoses of pelvic floor disorders like prolapse or endometriosis rely on mechanical indicators, such as mobilities of organs and shear displacements between organs. This information would be useful for both precise diagnoses and planning of surgical procedure. Involving numerical tools for diagnoses and surgery planing becomes increasingly interesting for physicians in clinical uses. The advantages of numerical models are not only in visualization, but also in quantitative measurements on a group of organs, such as their shapes and their relative movements. The processing pipeline includes patient data retrieval, image analysis, patient-specific modeling and biomechanical simulation. Our work consists in proposing new methods and algorithms for modeling the 3D anatomy of specific patients based on image data. This model should be compatible with the requirements of a biomechanical simulation. Moreover, we aim at developing new image processing tools for analyzing 2D dynamic MRI (to assess the mobilities of the pelvic system by extracting certain mechanical indicators from images) and for comparison with simulations.

Registration between geometric models and images remains a major challenge in these applications. We proposed a new model-to-image registration approach which was developed and tested for segmentation of organs in 2D images and for tracking the motion of pelvic organs from 2D dynamic MRI. Thanks to this technique, evaluation of the level of shear strain that is encountered by the fascias (connective tissues between organs) during the motion became possible. This tool could help in early diagnostic of prolapse. In the next step, our objective is to extend this method for adapting it to 3D reconstruction (with 3D geometric models and 3D MR images) and for the comparison of 3D simulations with deformable images.