A Novel Breast Mass Diagnosis System based on Zernike Moments as Shape and Density Descriptors

In this paper, a novel Computer-aided Diagnosis(CADx) system has been proposed for mass diagnosis inmammography images. Zernike moments are utilized asdescriptors of shape and density characteristics in order toimprove the overall accuracy. The input Regions of Interest(ROI) are segmented and subjected to some preprocessing stages.The outcome of preprocessing stage is a gray-scale imagecontaining co-scaled translated mass which contains both shapeand density characteristics of the mass. Two groups of Zernikemoments have been extracted from the preprocessed images.Considering the performance of the overall system the mosteffective moments have been chosen and applied to a Multi-layerPerceptron (MLP) classifier. The Receiver OperationalCharacteristics (ROC) plot and the performance of overall CADxsystem are analyzed for each group of features. The averageachieved area under ROC curve (Az) and False Positive Rate(FPR) for high-order moments are 0.872 and 18.34%,respectively. Besides, for low-order moments those are equal to0.824 and 15.44%, respectively.

کلیدواژه ها:

Computer aided diagnosis ، mammography ، multi layer Perceptron ، Zernike moments