| Summary: | Computer-aided diagnosis (CAD) approach is presented as strong frameworks to solve the inaccuracy problems. The major purpose of this paper is to improve a CAD system depended on supervised classification that can be useful in diagnosing and detecting the changes of breast cancers in digitized mammograms earlier, accurately and faster than standard examination programs by applying CAD according to image processing techniques beginning with preprocessing step, segmentation, feature extraction and finally classification stage. The work presented in this study is based on the integration of different features such as shape, texture and invariant moment features. This integration achieved best results for sensitivity and specificity rather than using the one type of features in breast cancer classification. The accuracy of our integration system reached 96% in the automatic mode of ANN while best accuracy accomplished by features result according to invariant moments that reached 97% by ANN in an automatic way. Keywords: Computer-aided diagnosis (CAD), Mammograms, K-nearest neighbor classifier (KNN), Support vector machine (SVM), Artificial neural network (ANN)
|
|---|
