Abstract: Ventricular wall skeleton and endocardial boundary were extracted from cardiac MRIs using medical image segmentation and morphological image processing methods. Then， the endocardial morphology was analyzed automatically by the key parameters that can describe the morphology of ventricular septum. An algorithm was proposed to automatically calculate the curvature of ventricular septum skeleton， and the concept of ventricular septum offset was presented to make up for the deficiency of describing the ventricular septum by the curvature，thus improving the imagebased evaluation method of ventricular septum morphology. Experimental results demonstrate that the curvature， offset and endocardial boundary can be used as key parameters to evaluate the shape of ventricular septum.

Key words: ventricular septum, image segmentation, morphology analysis, right ventricle, medical images