Quantitative evaluation of level set contour segmentation method applied to three dimensional images detecting image edge based on threshold is discussed. The integration of two popular level set methods, i.e., reinitialization method and geodesic active contour method, leads to success in development of robust, with high accuracy, level set based segmentation method. The applications of this method to both synthetic and clinical medical images are shown. Accuracy of the propose method, in the view of both averaged contour location and standard deviation of curvature in the vicinity of contour, calculated using obtained level set function is discussed.