A Morphology-Based Method for Automated Detection of Clustered Microcalcifications

摘要

Mathematical morphology offers a unified and powerful approach to numerous image-processing problems. This paper presents a method for automated detection of clustered microcalcification in mammograms using mathematical morphology. The method employs only one structuring element to reduce processing time, together with a top-hat transformation acting as a high-pass filter. Moreover, in order to effectively and accurately detect real microcalcifications, operations of differentiation, square-curve transfer function and postprocessing are also involved. The proposed method is tested with several mammograms to validate its effectiveness. Our preliminary results suggest the potential use of the proposed method.