Ultrasound Image Segmentation with Localized Level Set Based Curve Evolution and Rayleigh Distribution

In this paper, we present a new region based active contour algorithm for ultrasound image segmentation. An energy function based on a localized region-based active contour and shifted Rayleigh distribution is formulated. In our active contour framework, the target and background are represented as small local regions and the energy optimization is calculated at each point separately. The proposed algorithm drives the evolving curve according to the shifted Rayleigh distribution in a small local area. Using this framework, our algorithm is able to handle noisy and blurry boundaries object in ultrasound image. We test our algorithm on several ultrasound images and the results show the efficiency of our algorithm.