IVUS Image conditioning for in-vivo characterization of arterial tissue

Abstract

In-vivo characterization of arterial tissue is of paramount importance for culprit plaque identification, patient risk assessment and surgical planning. As culprit plaque histology is well known, the in-vivo tissue composition could be a valuable diagnostic and therapeutic tool. Currently, no in-vivo imaging technique provides direct histological information of the vessel wall. Different IVUS processing techniques have been used to infer the tissue composition from acoustic impedance variations. Most popular approaches for plaque characterization are based on classification schemes trained with ex-vivo tissue samples, which may lead to mischaracterization due to post-mortem changes in tissue properties. Hence, pure image based techniques are not capable of delivering the internal stress state of the vessel wall, which are consequence of strains produced by pressure loads. However, these images, under proper conditioning, provide fully detailed information about the wall kinematics, such as strains and rigid motions. In this manuscript we will present the necessary steps in IVUS image processing to have at hand well-conditioned data to accurately identify the wall kinematics and thus formulate a inverse problem to characterize arterial wall mechanical properties. The first step consists in the IVUS gating to extract different cardiac phases. Then, registration algorithms aim at removing transversal rigid motions. The so conditioned images are the input to different optical flow algorithms with the aim of describing the wall kinematics, that is, displacement vector field throughout the entire IVUS image. This displacement field is employed as input data for the inverse mechanical problem in which the material parameters of a constitutive model are determined by solving a nonlinear optimization problem. The focus of this work is on the conditioning stages of the IVUS images and the performance of optical flow algorithms in recognizing the displacement field. Concerning the inverse characterization, preliminary results will be presented using a state-of-the-art formulation of the optimization problem.