Kobatake and Yoshinaga [91] used skeleton analysis to detect spicules. The b2 algorithm first extracts the skeleton in regions of interest using different structuring elements to enhance lines in different directions. Subsequently, b2 detects the main lines of the skeleton using a variation on the Hough transform. Using the skeleton information and the main lines, b2 calculates the probability that pixels belong to a mass. Although in the original work the authors proposed an iris filter in order to enhance the image and automatically detect the regions of interest [92], we have used the same philosophy of the use of skeletons over the whole image.