Detection of myocardinal boundaries from ultrasound imagery using active contours

Tracking the motion of the myocardium of the murine heart from ultrasound image sequence aids in the assessment of therapies on human cardiovascular illness. In order to track myocardium, it is necessary to segment both the inner and outer boundaries of the myocardium in each frame. In this paper, we present a snake-based method to the problem of myocardium contouring by exploiting the features of the ultrasound imagery of the mouse heart and speckle reducing anisotropic diffusion. The proposed algorithm features an adaptive snake initialization technique. A sequence of 20 images is tested using the proposed approach and the obtained results are compared with manual outlines of the myocardium. It is shown that the method agrees well qualitatively and quantitatively in terms of the Pratt's figure of merit as stated in W.K. Pratt (1977).