Segmentation of Breast Focal Lesions on the Ultrasound Image

Abstract

An image segmentation technique is proposed. The technique is based on the automated calculation of the sigmoid filter parameters from the characteristics of the brightness distribution function for the image under study. The technique provides effective segmentation of the following focal lesions on ultrasound images: hypoechoic, with a relatively clear contour against a background with higher echogenicity (mainly cysts and benign breast tumors); hypoechoic, with a less clear contour against a less echogenic background (mainly carcinomas, but also benign lesions with pronounced fibrosis and sclerosis). For isoechoic lesions, the effectiveness of the suggested technique is suboptimal, so that alternative approaches should be used.