Cardiac left ventricular volume changes assessment by long axis echocardiographical image processing

The authors describe a new fully automatic algorithm for left ventricular (LV) volume change measurement and assessment from both 2-chamber and 4-chamber apical views of echocardiographic images based on the American Society of Echocardiography (ASE) recommendations. The method identifies the left ventricular endocardium (LVE) in a 'centrebased' approach. In this approach the left ventricular centre point (LVCP) is firstly approximated using a robust fuzzy based technique. The LVE edge points are then searched for on nonuniform distributed radial lines emanating from the LVCP using a multiresolution edge detection technique based on the global maximum of wavelet transform (GMWT). The extracted edge points are then spatially/temporally processed to identify the final smooth boundary using cubic-B-splines. The extracted LVEs in a complete cardiac cycle, together with the standard approximating formula, are then used to compute the LV volume evolution in a cardiac cycle. Experiments with real two-dimensional echocardiographic images are presented. The accuracy of the GMWT edge detection technique, which is the heart of the system, is compared to the conventional Sobel and Laplace edge operators.