Detecting the endocardium and epicardium of the left ventricle is important for further quantitative analysis of cardiac functions and three‐dimensional reconstruction. However, to detect the boundaries automatically is difficult. In this paper, we proposed a novel, single frame method, based on binary morphology and region growing techniques, to detect the boundaries of the left ventricle automatically except an A.O.I. (Area Of Interest) adjustment operation must be performed during the process of getting image data from video tape and storing it onto the disk. The method is suited for boundary detection required in three‐dimensional reconstruction for the observation of congenital abnormalities of ventricular septum or for generating cross sectional images of arbitrary orientation. Moreover, the method may be improved by further processings for accurate endocardial and epicardial boundary estimation for quantitative cardiac function evaluations.