Three-Dimensional Representation And Parametric Computation Of Cardiac Shapes Based On A NURBS Model

Heart diseases cause high mortality while the therapy of these diseases is still very difficult. Consequently, it is urgent to build up a cardiac three-dimensional model and calculate relevant parameters accurately, since these parameters (such as ejection fraction and left ventricular volume etc.) have the capability to improve the clinical diagnostic value. The work carried out in our research has values in both academic and practical aspects.In this paper, we present a realistic and flexible model of human heart surface, and this model is used to calculate cardiac functional parameters, the heart shape is represented with Non Uniform Rational B-Spline (NURBS) surface. With the flexibility of the NURBS, the NURBS model can fit heart shape more realistically than the cardiac model based on simple geometric shapes. At the aspect of assessment for cardiac parameters, we present an integral algorithm for cardiac volumetric measurement of NURBS surface based on matrix representation. Then, we use the clouds of points from endocardium and epicardium of left ventricle obtained by three-dimensional active shape model, and select seven phases from a complete heartbeat cycle and acquire their volumes according to our integral algorithm. Finally, we analyze the other functional parameters (like ejection fraction, cardiac output) by volumes. Experimental results show that the NURBS surface can well fit the surface of human heart and the inner and external surfaces of left ventricle. Experiments also show the volumetric algorithm can get satisfactory measurement accuracy and efficiency. Besides, our volumetric algorithm can accurately calculate volumes of those conicoids which can be fitted accurately by NURBS surfaces, and it can get better results according to add more control points of these free shapes like heart and left ventricle.