By taking the advantages of the Voronoi-based segmentation method in detecting the thin edges and the region growing segmentation method in segmenting regions more completely, we proposed a novel segmentation algorithm by combining the Voronoi diagrams and region growing to deal with the medical images in this paper. Firstly, the region growing segmentation algorithm is used to produce a rough partition, and then the Voronoi-based method is applied to the initial partition for further iterative processing to get a clearer border. Segmentation results on CT and MR images show that our method can segment the target area completely and accurately. Especially, it can effectively preserve the high frequency information in the edge area. Furthermore, we found that the proposed method behaved better on MR images than on CT images.