Segmentation of Ultrasound Images Based on Active Contours for Image Registration

Abstract

This paper describes the implementation of the segmentation method applied on ultrasound images, which is based on active contour models. The physical meaning of these models is represented by energy minimalisation of initially defined curve. Energy could be written as a weighted sum of internal and external energy. Internal energy is strictly dependent on the properties of the curve and external energy si represented by a feature extracted from the image. First and second derivative of the curve is used to calculate internal energy and Laplacian of the image as a feature to calculate external energy. Total energy is minimised by iteratively moving points of the curve. Achieved results and persistent problems are evaluated, suggesting the way the future work should lead.