Ultrasound image segmentation by Bhattacharyya distance with Rayleigh distribution

In this paper, we study the problem of ultrasound image segmentation of kidney images. We propose a new region based active contour algorithm. The energy function of our algorithm is based on Chan-Vese energy function and the Bhattacharyya distance. In our framework, a curve is evolved to partition the image into two parts. Our algorithm minimizes the differences between each part and maximizes the distance of density function between each part. In the experiments, we use images of kidney phantoms and real ultrasound medical images to validate the performance our method. The results show that by taking into account the similarity of two different regions, the proposed method is able to deal with blurry boundaries in an ultrasound image and provide good results on both phantom and patient US images.