COMPUTERIZED CLASSIFICATION OF MALIGNANT AND BENIGN MICROCALCIFICATIONS ON MAMMOGRAMS - TEXTURE ANALYSIS USING AN ARTIFICIAL NEURAL-NETWORK

We investigated the feasibility of using texture features extracted fr om mammograms to predict whether the presence of microcalcifications i s associated with malignant or benign pathology. Eighty-six mammograms from 54 cases (26 benign and 28 malignant) were used as case samples. All lesions had been recommended for surgical biopsy by specialists i n breast imaging. A region of interest (ROI) containing the microcalci fications was first corrected for the low-frequency background density variation. Spatial grey level dependence (SGLD) matrices at ten diffe rent pixel distances in both the axial and diagonal directions were co nstructed from the background-corrected ROI. Thirteen texture measures were extracted from each SGLD matrix. Using a stepwise feature select ion technique, which maximized the separation of the two class distrib utions, subsets of texture features were selected from the multi-dimen sional feature space. A backpropagation artificial neural network (ANN ) classifier was trained and tested with a leave-one-case-out method t o recognize the malignant or benign microcalcification clusters. The p erformance of the ANN was analysed with receiver operating characteris tic (ROC) methodology. It was found that a subset of six texture featu res provided the highest classification accuracy among the feature set s studied. The ANN classifier achieved an area under the ROC curve of 0.88. By setting an appropriate decision threshold, 11 of the 28 benig n cases were correctly identified (39% specificity) without missing an y malignant cases (100% sensitivity) for patients who had undergone bi opsy. This preliminary result indicates that computerized texture anal ysis can extract mammographic information that is not apparent by visu al inspection. The computer-extracted texture information may be used to assist in mammographic interpretation, with the potential to reduce biopsies of benign cases and improve the positive predictive value of mammography.