Classification of mammogram images by using CNN classifier

Classification of the breast tissues into the benign and malignant classes is a difficult assignment. The experimental results are takes 40 input images from DDSM dataset. We extract the GLCM, GLDM and Geometrical features from the mammogram images. In this paper we apply Convolution Neural Network as a classifier on the mammogram images to enhance the accuracy rate of CAD. Performance of the different classifiers is measured on receiver operating characteristic. In training stage, overall classification accuracy of 73%, with 71.5% sensitivity and 73.5% specificity for dense tissue is achieved by our proposed method along with it, accuracy of 79.23%, 73.25% sensitivity and 74.5% specificity is achieved for fatty tissue. Convolution neural network classifier is used to boost the classification performance. This classifier performs better than previous classifiers in that it shows more accuracy than the other classifiers, the misclassification rate of normal mammograms as abnormal. This approach performs good on overlapping problem. This method is different from all other approaches, which are used to identify normal mammograms by detecting cancers. Overlapped tissues are also detected by this using this classifier.