Breast ultrasound images gland segmentation

This paper introduces a study for the segmentation of the breast ultrasound images. The objective is to separate the breast gland, which is the region of interest for the breast cancer diagnosis, from other tissues. Images are pre-processed with four different algorithms that consider the image surrounding: speckle reducing anisotropic diffusion, homomorphic filter, Perona and Malik non-linear diffusion and Moran index. For each image pixel a four bins descriptor is created composed by the corresponding pixels of each of these preprocessed images.