3D echocardiography pre-processing for ventricular volume estimation

Abstract

A method for left ventricular volume estimation based on 3D echographic data is presented. Pre-processing of 3D data is performed by means of two different nonlinear diffusion techniques: anisotropic diffusion and geometrical diffusion (mean curvature motion). Anisotropic diffusion, which assumes a high diffusion velocity in region with low brightness gradient and vice-versa, produces a filtering effect without smoothing the edges. On the contrary, geometrical diffusion, which assumes a diffusion velocity according to local curvature, produces a smoothing of surfaces. The ventricular volume is estimated through a floodfill operation and divergence theorem.