A new variational method for selective segmentation of medical images

•A new image smoothing model is defined, which can attenuate complicated background but preserve edges of targeted object, thus effectively facilitating medical image segmentation•A modified Gout model is given by using a robust gradient operator. It is simpler and more robust to noise and intensity inhomogeneity. We also give a new initialization method for the level set function used in segmentation.•Based on the first two contributions, a two-phase selective image segmentation method is presented. In the first phase, the original image is smoothed by using our proposed model. Then in the second phase, the modified Gout model is used on the smoothed image to detect the target boundary.•Extensive experiments on real medical images show that, our smoothing model can greatly facilitate the second phase, and our method can significantly improve two existing related methods in terms of either visual assessment or quantitative evaluation.