2018/07/09 by Nripesh Parajuli, Allen Lu, Parajuli, Nripesh +21
Computer Science · Engineering · Medicine · #Cardiac Valve Diseases and Treatments #Cardiovascular Function and Risk Factors #Computer Vision and Pattern Recognition (cs.CV) #Elasticity and Material Modeling #FOS: Computer and information sciences #Medical Image Segmentation Techniques #Medical Imaging and Analysis
paper · pdf · doi:10.48550/arxiv.1807.02951
openalex publication_date 2018/07/09 · openalex created_date 2022/08/04 · openalex updated_date 2026/07/28
The accurate quantification of left ventricular (LV) deformation/strain shows\nsignificant promise for quantitatively assessing cardiac function for use in\ndiagnosis and therapy planning (Jasaityte et al., 2013). However, accurate\nestimation of the displacement of myocardial tissue and hence LV strain has\nbeen challenging due to a variety of issues, including those related to\nderiving tracking tokens from images and following tissue locations over the\nentire cardiac cycle. In this work, we propose a point matching scheme where\ncorrespondences are modeled as flow through a graphical network. Myocardial\nsurface points are set up as nodes in the network and edges define neighborhood\nrelationships temporally. The novelty lies in the constraints that are imposed\non the matching scheme, which render the correspondences one-to-one through the\nentire cardiac cycle, and not just two consecutive frames. The constraints also\nencourage motion to be cyclic, which is an important characteristic of LV\nmotion. We validate our method by applying it to the estimation of quantitative\nLV displacement and strain estimation using 8 synthetic and 8 open-chested\ncanine 4D echocardiographic image sequences, the latter with sonomicrometric\ncrystals implanted on the LV wall. We were able to achieve excellent tracking\naccuracy on the synthetic dataset and observed a good correlation with\ncrystal-based strains on the in-vivo data.\n