Screening and early detection of breast cancer needs an automated system that identifies the breast cancer in the mammograms as early as possible. An automated system that segments the mammogram masses and identifies the defect in the mammograms is proposed. The mammogram images are preprocessed by using median filter and adaptive histogram equalization. The median filtering process removes the unwanted pixels (noises) in the mammogram images which helps in improving the performance of the process. Histogram equalization process helps to equalize the intensity of the image which will be helpful to identify the objects present in the image. The breast masses are segmented from the preprocessed images using Region Growing algorithm. The Region Growing segmentation algorithm segments the breast masses from the image based on the clustered result of group of pixels in the image. From the extracted breast mass the features are extracted using GLCM algorithm. From the extracted features best features are selected using PSO algorithm. The selected features were then classified using Feed forward neural network. Finally the performance of the classifier is measured by calculating accuracy, sensitivity and specificity. The performance of the process is compared with kNN classifier. The kNN classifier classifies the images based on the distance metrics between the classifiers. The distance is calculated between the test and the training feature vectors. The obtained performance metrics shows that the neural network classifier produces better results compared to the kNN classifier. ROC curve and confusion matrix were plotted to show the performance of the classifiers.