Resolution du probleme inverse en elastographie ultrasonore par une methode variationnelle.

摘要

Atherosclerotic plaque rupture is responsible for the majority of myocardial infarctions and acute coronary syndromes. In vivo assessment of the mechanical stress could provide valuable plaque rupture indices for the diagnosis and prognosis of those clinical outcomes. As for now, it is not yet possible to determine the internal distribution of stresses directly in vivo. However, medical imaging modality such as ultrasound imaging could help achieving this goal through processing of image sequences providing the strains of the tissue and its elasticity parameters distribution. The strains and elasticity parameters of an elastic tissue can help to estimate its mechanical stresses. A number of ultrasound images processing methods have been proposed for that purpose. However, often their underlying hypotheses prevent them from being clinically applicable to clinical data. For example, a method may assume a constant stress distribution within a region where therefore the strain map can be interpreted also as an elastic parameter map in this region. Other methods would use an iterative process to best match the measurable data obtained from images to the ones predicted from a finite element model (FEM) of the tissue. But, their implementation generally requires a priori knowledge on each analyzed plaque, as for example its boundary conditions that are not often identifiable from clinical images. This obviously limits the transferability of those methods to the analysis of a broad range of clinical data. Such model-based inversions of ultrasound images do not currently consider non-linear rheological laws of arteries, an important concern for stress analysis.;The proposed approach can serve to develop a medical imaging tool to characterize atherosclerotic plaques in view of their vulnerability to rupture.;In this project, we first re-examine assessment of plaque vulnerability and demonstrate that non linear mechanical effects are important for the analysis of the stability of the atherosclerotic plaques. We then develop an inversion approach to process ultrasound images, based on the optimization of functionals that minimize the mechanical energy of elastic tissues. Then, we evaluate our approach on several plaque models and several ultrasound imaging modalities: radio-frequency ultrasound data, envelope or B-Mode data, intra-vascular and extra-vascular ones. Due to its generality, this approach can be applied to a broad range of plaque geometry and can be extended to inversions for non linear mechanical cases.