An effective method for ultrasound thyroid nodules segmentation

Ultrasound thyroid nodules segmentation is a much tough task due to the speckle noise, intensity heterogeneity, and low contrast. In this paper, we introduce an effective segmentation method to make the problem better. Our works are composed of pre-processing, extracting boundaries, postprocessing. First, we suppress the speckle noise by speckle reducing anisotropic diffusion method. Second, we introduce a reasonable and effective speed stopping term based local phase symmetry feature which is theoretically intensity invariant, and the new speed stop term is incorporated into distance regularized level set evolution. Then, to remedy the emerging problems, we do some effective post-processing. Qualitative and quantitative comparative results demonstrate the superiority of our proposed method.