In this paper we propose a level-set segmentation for polyp candidates in Computer Tomography Colongraphy (CTC). Correct classification of the candidate polyps into polyp and non-polyp is, in most cases, evaluated using shape features. Therefore, accurate recovery of the polyp candidate surface is important for correct classification. The method presented in this paper, evolves a curvature and gradient dependent boundary to recover the surface of the polyp candidate in a level-set framework. The curvature term is computed using a combination of the Mean curvature and the Gaussian curvature. The results of the algorithm were run through a classifier for two complete data-sets and returned 100% sensitivity for polyps greater than 5mm.