Development of a method for left ventricular shape evaluation based on surfaces obtained by real-time 3D echocardiographic images

Left ventricular (LV) shape is strictly related to LV function. Previously proposed shape indexes suffer from their intrinsic 2D nature. Our aim was to develop a method to quantify directly 3D LV shape from LV surfaces semi-automatically detected from real-time 3D echocardiographic (RT3DE) images. First, a monodimensional shape descriptor has been obtained by sampling through an helical pattern the extracted endocardial surfaces. Then, four shape indexes have been defined by comparing the LV with reference shapes. The developed method was applied to RT3DE datasets obtained from 12 normal subjects, 8 patients with dilated cardiomiopathy and 7 ischemic patients. Some of the proposed shape indexes were able to follow chamber deformation throughout the cardiac cycle and to discriminate between normal and pathologic conditions.