Automated lungs node segmentation by means of dynamic programming and classification based on em

摘要

Is provided a method for automatic segmentation of the lungs node in a three-dimensional (3 - d) of computer tomography (ct) - volume data set. An input is received, which one of a user selected point near a limit of a node corresponds to (210). A model of the node is formed of the selected point by a user from designed, whereby the model is a deformable circle, which has a set of parameters beta, which represent a shape of the node (230). Continuous parts of the limit and discontinuities of the limit estimated be, until the set of parameters beta converges, by the use of dynamic programming device and select the maximizing (em) (240). The node, on the basis of estimates of the continuous parts of the limit and the discontinuities of the limit of segmented (250).