Minimal-path contours combined with speckle tracking to estimate 2D displacements of the carotid artery wall in B-mode imaging

We present here an enhancement of our Contour and Speckle Tracking (CST) method, dedicated to assess the bi-dimensional (radial and longitudinal) movement of the common carotid artery walls in ultrasound B-mode sequences, in an attempt to detect early-stage atherosclerosis. CST is based on a speckle tracking approach guided by contour segmentation. We introduce a convolution operator dedicated to find the interfaces of the intima-media complex using a minimal-path approach, and we propose a block matching framework with a simple update scheme. Our method has been evaluated on 12 sequences of healthy carotid arteries, versus manually traced reference contours and 2D trajectories. A classical block matching speckle tracking (BMST) algorithm has also been applied for comparison. The accuracy of the segmentation was 30 ± 17 μm and 29 ± 21 μm for the lumen-intima and the media-adventitia interfaces, respectively. The mean longitudinal and radial dispersion error was 36 ± 24 μm and 14 ± 8 μm vs 103 ± 102 μm and 60 ± 33 μm, for CST and BMST, respectively.