Comparison of Block Matching and Differential Methods for Motion Analysis of the Carotid Artery Wall From Ultrasound Images

Abstract

Abstract-Motion of the carotid artery wall is important for the quantification of arterial elasticity and contractility and can be estimated with a number of techniques. In this paper, a framework for quantitative evaluation of motion analysis techniques from B-mode ultrasound images is introduced. Six synthetic sequences were produced using 1) a real image corrupted by Gaussian and speckle noise of 25 and 15 dB, and 2) the ultrasound simulation package Field II. In both cases, a mathematical model was used, which simulated the motion of the arterial wall layers and the surrounding tissue, in the radial and longitudinal directions. The performance of four techniques, namely optical flow (OF H S ), weighted least-squares optical flow (OF L K (W L S ) ), block matching (BM), and affine block motion model (ABMM), was investigated in the context of this framework. The average warping indices were lowest for OF L K (W L S ) (1.75 pixels), slightly higher for ABMM (2.01 pixels), and highest for BM (6.57 pixels) and OF H S (11.57 pixels). Due to its superior performance, OF L K (W L S ) was used to quantify motion of selected regions of the arterial wall in real ultrasound image sequences of the carotid artery. Preliminary results indicate that OF L K (W L S ) is promising, because it efficiently quantified radial, longitudinal, and shear strains in healthy adults and diseased subjects.

3D cloud detection and tracking system for solar forecast using multiple sky imagers

“…By incorporating the deformation model, OF-based methods can capture the dynamic motions of clouds between consecutive frames based on image registration techniques (Bernecker et al, 2012). In general, block-wise cloud tracking, as discussed in previous works Golemati et al, 2012), is an extension of the pixel-wise methodology. These two approaches are applied under different tracking conditions: in the block-based approach, the properties within a cloud block are integrated to mitigate noise and misrepresentations of color in the original images, whereas the pixel-wise approach focuses on the typical movements that occur at the pixel or sub-pixel scale.…”

Improved Tracking Performance of Lagrangian Block-Matching Methodologies Using Block Expansion in the Time Domain: In Silico, Phantom and in Vivo Evaluations

“…Though we have not seen any article that describes the use of one extra reference block in a blockmatching algorithm, the use of pixel information from prior blocks has been investigated, using Kalman-based methods, with promising results (Golemati et al 2012;Zahnd et al 2013). The main difference between their methods and that described here is the reference block used for tracking.…”

“…Several methods have been developed for estimating longitudinal movement and the resulting strains in the arterial wall (Cinthio et al , 2006(Cinthio et al , 2007Gastounioti et al 2011;Golemati et al 2003Golemati et al , 2012Hasegawa and Kanai 2009;Idzenga et al 2012;Larsson et al 2011;Nilsson et al 2009Nilsson et al , 2010Numata et al 2007;Persson et al 2002Persson et al , 2003Yli-Ollila et al 2013;Zahnd et al 2011Zahnd et al , 2012b. However, the methods for estimating longitudinal movements of the arterial wall are limited.…”

“…However, Zahnd et al (2012b) included VVI in a validation and found that VVI is less accurate and has larger variability than a full-search block-matching method using a large block (2.5 3 1.3 mm 2 ). The use of pixel information from blocks in previous frames has been investigated with promising results (Golemati et al 2012;Zahnd et al 2013). In these studies, the pixel values in the reference block are predicted by a Kalman filter using the previous reference blocks.…”

A Case Analysis on Different Registration Methods on Multi-modal Brain Images

“…The normalized cross-correlation between two images and is given by the equation: (20) Where is the coordinate of a window centered on the coordinate ( , ∈ 2 ) , ( ) and ( ) are the expected value of and in the window where it is being correlated. Normalized cross-correlation can only very accurately align transposed images, and it turns out to be one of the methods most used in the literature.…”