One of the most leading reason for death among women is breast cancer. As early as the cancer is detected the survival rate will be high. Detecting masses at the early stage using mammograms is the prime and difficult process as most of the cancer masses get obscured in normal breast tissues. This paper presents a new computational methodology that assists the radiologist in detecting masses in mammographic images. The proposed mass detection algorithm is first preprocessed by enhancing the image using histogram equalization and removing the objects surrounding the breast region using global thresholding. The second stage is extraction of shape and texture features for discriminating masses and non-mass regions. The discriminating power of the features extracted is done using classification approach. The third stage involves classification of cancerous tissues using wavelet neural network (WNN). The WNN classifier as evaluated in mammographic images acquired from mammographic analysis centers. Performance analysis of the proposed WNN classifier as based on the accuracy of the classifier to identify pathologies in the mammogram image. A classification accuracy of 90.19% as obtained and the proposed classifier was found to be efficient. © International Science Press.