This paper presents a curve evolution model for 3D slice-by-slice image segmentation and its application in medical imaging. It is an iterative process based on the dual front evolution and the morphological dilatation to iteratively deform the initial contour towards the segmentation result. The dual front evolution model is proposed to form the new boundary by the contact position of two (or more) curves evolving in opposite directions. The fast sweeping evolution scheme is introduced for the contour evolution and the velocities for the propagation of the different curves are defined in accordance with the region-based characteristics. This model can achieve the global energy minimum and solves the disadvantages of classical level set evolution methods. Experimental results are given to illustrate the robustness of the method and its performance in precise region boundary localization and medical imaging.