Classifying Microcalcifications in Digital Mammograms using Machine Learning techniques

Abstract

This paper presents a Computer Aided Diag­nosis (CAD) of breast cancer from mammograms. The first part involves severa! image processing techniques, which extract a set of features from the microcalcifications (µCa) present in a mammo­gram. The second part applies different machine learning techniques to obtain an automatic diagno­sis. The Machine Learning (ML) approaches are: Case-Based Reasoning (CBR) and Genetic Algo­rithms (GA). We study the application of these al­gorithms as classification systems in order to dif­ferentiate benign from malignant µCa in mammo­grams, obtained from the mammography database of the Girona Health Area, and we compare the classification results to other classification tech­niques.