An application of topological median filters on detection and clustering of microcalcifications in digital mammograms

Cihan I. K., Senel H. G.

31st IEEE International Conference on Acoustics, Speech and Signal Processing, Toulouse, France, 14 - 19 May 2006, pp.2384-2387 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • City: Toulouse
  • Country: France
  • Page Numbers: pp.2384-2387
  • Middle East Technical University Affiliated: No


Existence of microcalcification clusters on mammograms is one of the earliest signs of breast cancers. In this study, a method that is based on topological median filters is proposed for the automated detection of microcalcifications. The proposed algorithm consists of two steps. First, probable microcalcification pixels in the mammograms are segmented out by using topological top-hat transform. Then, individual microcalcifications are clustered by using a subtractive clustering algorithm. The method has been applied to Nijmegen database of 34 mammograms with a total of 72 microcalcification clusters. The results show that the proposed algorithm has a success rate of 93%.