09 - Reconstruction 3D de vaisseaux sanguins par fusion de données à partir d'images angiographiques et échographiques

Three-dimensional reconstruction of vascular lesions is of great medical interest for diagnosis and prognosis of the atheromateous disease, as well as for a better understanding of restenosis phenomena related to the revascularisation techniques. In order to provide a quantitative description of the geometry of complex atheromateous lesions, we propose an original approach for 3D vessel surface reconstruction, which relies on fusion of data obtained with both digital angiography and intravascular echography. This method is based on a first step consisting of a geometrical modelization of acquisitions and data, which yields a first rough reconstruction by geometrical fusion of the parameters extracted from the vessel contours obtained from both modalities which provides, by interpolation of echographic contours, the 3D surface of the vessel lumen . The next step consists in taking into account the imprecision on the parameters estimated during the first step. For this task, we propose an original method based on a modelization of parameters as fuzzy numbers which allows to introduce, through fuzzy mathematical morphology, the imprecision on the acquisitions and on the 3D data. This provides a 3D reconstruction which takes into account all data about the problem . The first results prove clearly the feasibility of the method and the interest of using information issued from different modalities for improving the 3D reconstruction of blood vessels, without any a priori mathematical model of vessel morphology. The explicite introduction of imprecisions in the fusion process allows to eliminate ambiguities and contradictions we would obtain in a simple reconstruction from only one modality, and leads to a decision about the true morphology of vessels.