INTERACTIVE SELECTIVE AND ADAPTIVE CLUSTERING FOR DETECTION OF MICROCALCIFICATIONS IN MAMMOGRAMS

This paper presents a clustering algorithm, called interactive selecti ve and adaptive clustering (Isaac), to assist radiologists in looking for small clusters of microcalcifications in mammograms. Isaac is deve loped to identify suspicious microcalcification regions which are miss ed by other classification techniques due to false positive samples in the feature space. It comprises two parts: (i) selective clustering a nd (ii) interactive adaptation. The first part reduces the number of f alse positives by identifying the microcalcification subspace or domai ns in the feature space. The second part allows the radiologist to imp rove results by interactively identifying additional false positive or true negative samples. Clinical evaluations of mammograms indicate th e potential of using this algorithm as an effective tool to bring micr ocalcification areas to the attention of the radiologist during a rout ine reading session of mammograms. (C) 1996 Academic Press, Inc.