Radial multiscale cyst segmentation in ultrasound images of kidney

Cysts are one of the most common lesions in kidneys and can be diagnosed exploiting ultrasound images. In this paper, we develop an automatic approach for cyst segmentation in ultrasound images. The proposed approach comprises three steps: finding a seed point in the object exploiting the Gibbs random field, detecting the boundary based on multiresolution signal processing and edge refining and finally shape model features extracting to verify whether the object is a cyst. The proposed approach is fast and less complex to suit ultrasound exploration applications.