Segmentation of B-scan images of gallstones based on mathematical morphology

A new method for the segmentation of ultrasonic B-scan images of gallstones using mathematical morphology is presented. It is based on an intuitive approach assuming the shape of the solitary stone to be convex. The front echo is separated from the inner echo by taking the first 20% of the signal. The mean gray levels of the two regions are compared. It turns out that this alone is not reliable. But, analysing the stone shape and the segments seems to be promising for improving the segmentation algorithm.