Extraction of cavity boundary from ultrasound images

To extract a more reliable cavity boundary from ultrasound images, a three step algorithm is presented in this paper. In the first step, we transform the ultrasound image into an appropriate feature space, in which the spatial distribution of the cavity inside the ultrasound images is represented. Next, morphological filtering is used to get an initial contour of the cavity. The initial contour is finally input to the snake model to get the exact cavity boundary. Experimental results are also provided to demonstrate the effectiveness of our algorithm.