Study and Analysis of Two Segmentation Methods for Ultrasound Images

Ultrasound images contain strong speckle noise and attenuation artifacts such as intensity inhomogeneity which makes the segmentation process challenging. In this paper, two segmentation algorithms have been studied and analyzed. First method is multiplicative intrinsic component optimization (MICO) algorithm which is based on the minimization of bias fields. It decomposes the ultrasound images into two multiplicative components, the bias field and the true image. Bias field accounts for the intensity inhomogeneities present in the image space which is assumed to be smoothly varying and the true image defines a physical property of the tissues. The energy in the formulation of this method is convex in each of its variables. Active contour method has been used for further segmentation of the energy minimized image. Thereafter, this method is compared with the traditional, edge based active contour method. Index Terms Ultrasound Images, Image Segmentation, Multiplicative Intrinsic Component Optimization (MICO), Active Contour