Medical image segmentation with minimal path deformable models

This paper presents an algorithm that segments medical images by extracting object contours. It delineates object boundaries by detecting a path with the minimum energy on the image. A worm algorithm based on deformable models is proposed to find the minimal path by using the dynamic programming technique. The proposed algorithm overcomes the shortcomings of traditional deformable models such as fastidious initialization and inefficiency on segmenting objects with complex shapes or topologies. After presenting the algorithm, its performance on various synthetic and medical images is shown. Experimental results indicate that the proposed algorithm is robust to noise and edge discontinuities.