作者: S. Golemati , J. S. Stoitsis , A. Gastounioti , A. C. Dimopoulos , V. Koropouli
DOI: 10.1109/TITB.2012.2193411
关键词: Optical flow 、 Image warping 、 Gaussian noise 、 Context (language use) 、 Mathematics 、 Speckle noise 、 Real image 、 Computer vision 、 Artificial intelligence 、 Motion analysis 、 Pixel
摘要: Motion of the carotid artery wall is important for quantification arterial elasticity and contractility can be estimated with a number techniques. In this paper, framework quantitative evaluation motion analysis techniques from B-mode ultrasound images introduced. Six synthetic sequences were produced using 1) real image corrupted by Gaussian speckle noise 25 15 dB, 2) simulation package Field II. both cases, mathematical model was used, which simulated layers surrounding tissue, in radial longitudinal directions. The performance four techniques, namely optical flow (OFHS), weighted least-squares (OFLK(WLS)), block matching (BM), affine (ABMM), investigated context framework. average warping indices lowest OFLK(WLS) (1.75 pixels), slightly higher ABMM (2.01 highest BM (6.57 pixels) OFHS (11.57 pixels). Due to its superior performance, used quantify selected regions artery. Preliminary results indicate that promising, because it efficiently quantified radial, longitudinal, shear strains healthy adults diseased subjects.