A Edge Detection Method for Microcalfication Clusters in Mammograms
Yu Guang Zhang, Wen Bing Lu, Cheng Fu, Li Song · 2009
Edge is one of the most important characteristics of microcalcifications, edge detection of microcalcification clusters has a great significance in computer-aided diagnosis system for the automatic detection of clustered microcalcifications in digitized mammograms. A lot of algorithms have been suggested for extracting medical image edges, however, few of them are well suited for edge extraction of microcalcifications due to obtaining discontinuous edges, or continuous edges with more over-detection points. In this paper, we propose a new method for clustered microcalcifications edge detection by integrating kirsch edge operator, edge linking with Markov model. First, initial edges are extracted by employing kirsch edge operator. Then, we thin the initial edges and fill many gaps in the edge image using edge linking technique. Finally, closed boundaries of microcalcifications are obtained based on Markov model. The experiments demonstrate that our algorithm can obtain closed boundaries with less over-detection points.