Abstract:
The goal of the present work is to establish a strategy to perform the three-dimensional reconstruction of the coronary vessels in different moments of the cardiac cycle, taking as inputs the IVUS and the angiographic studies. In the conception of that methodology are prioritized the automation, the accuracy and the robustness of the involved methods. This problem of reconstruction is decomposed in three stages: filtering of the IVUS study, segmentation of the IVUS study and the 3D reconstruction of the vessel aided by angiographies. In the first stage, the oriented speckle reduced anisotropic diffusion (OSRAD) method is employed to remove the speckle noise on IVUS images. Furthermore, the identification and removal of the image artefacts are treated to reduce the complexity of the posterior segmentation. In the second stage, segmentation is driven by an active contours approach properly endowed with a variety of methodological extensions that allow us to deal with the poor quality encountered in ultrasound images. In addition, novel methods are presented to extract movement informations from IVUS frames influenced by the heart beat motion. This increases the segmentation robustness in this kind of frames and permits more accuracy in the vessel reconstruction during the systolic phase. In the final stage, a reconstruction process aided by not necessarily synchronized angiographies is presented, generalizing the classic approach which is restricted to the use of biplane angiographies. In this way, the methodology allows the use of more traditional equipments albeit quality of reconstruction can experience some decrease. In conclusion, the proposed approach integrates well-established solutions from the literature with novel methods towards maximizing the extraction of data from the given studies (IVUS and angiographies), relax the equipment requirements and obtain a more accurate reconstruction of the coronary vessels.