Computer-aided measurement of solid breast tumor features on ultrasound images

Abstract

This paper presents a new approach in the application of computer vision techniques to the diagnosis of solid breast tumors on ultrasound images. Most works related to medical image analysis for breast cancer detection refer to mammography. However, radiologists have proved the significance of some aspects observed on ultrasound images, among which are spiculation, calcifications, ellipsoid shape, dimensions, echogenicity, capsule, angular margins, lobulations, shadowing and ramifications. We have developed a common framework for the analysis of these criteria, so that a series of parameters are available for the physicians to decide whether the biopsy is necessary or not. We present a set of mathematical methods to extract objective evidence of the presence or absence of the diagnostic criteria. This system is able to extract the relevant features for solid breast nodules with high accuracy and represents a very valuable help in the assessment of radiologists.