A fast algorithm for global minimization of maximum likelihood based on ultrasound image segmentation
Jie Wen Huang, Xiaoping Yang, Yunmei Chen · Inverse Problems and Imaging · 2011
This paper presents a novel variational model for ultrasoundimage segmentation that uses a maximum likelihood estimator based onFisher-Tippett distribution of the intensities of ultrasound images.A convex relaxation method is applied to get a convex model of the subproblem with fixed distribution parameters.The relaxed subproblem, which is convex, can be fast solved by using a primal-dual hybrid gradient algorithm. The experimentalresults on simulatedand real ultrasound images indicate the effectiveness of the method presented.