An Artifical Intelligence Based Software Application for Microcalcification Detection on Mammogram Images

Abstract

In this study, a software framework is proposed for detecting of microcalcification clusters which are abnormalities in mammogram images. Apart from the two well known methods two different feature extraction methods integrating to this software an artificial intelligence based detection system has been developed. In the study Gray Level Co-occrrance Matrix(GLCM), Equal Width Discritization(EWD2), Wavelet Transform and Multiple Window Based Analysis (MWBA) were used as for feature extraction Artifical Neural Network (ANN) was used as classifier. The highest AUC value was obtained by the MWBA method(AUC=0,91).