Breast cancer is the common form of cancer and leading cause of mortality among woman, especially in developed countries. In western countries about 53–92 % of the population has this disease. As with any form of cancer, early detection and diagnosis of breast cancer can increase the survival rate. Mammography is the current diagnostic method for early detection of breast cancer. Breast parenchymal patterns are not stable between patients, between left and right breasts, and even within the same breast from year to year in the same patient. Breast cancer has a varied appearance on mammograms, from the obvious spiculated masses, to very subtle asymmetries noted on only one view, to faint calcifications seen only with full digital resolution or a magnifying glass. The large volume of cases requiring interpretation in many practices is also daunting, given the number of women in the population for whom yearly screening mammography is recommended. It seems obvious that this difficult task could likely be made less error prone with the help of computer algorithms. Computer-aided detection (CAD) systems have been shown to be capable of reducing false-negative rates in the detection of breast cancer by highlighting suspicious masses and microcalcifications on mammograms. These systems aid the radiologist as a ‘second opinion’ in detecting cancers and the final decision is taken by the radiologist. A supervised machine learning algorithm is investigated—Differential Evolution Optimized Wavelet Neural Network (DEOWNN) for detection of abnormalities in mammograms. Differential Evolution (DE) is a population based optimization algorithm based on the principle of natural evolution, which optimizes real parameters and real valued functions. By utilizing the DE algorithm, the parameters of the Wavelet Neural Network (WNN) are optimized. To increase the detection accuracy a feature extraction methodology is used to extract the texture based features of the abnormal breast tissues prior to classification. Then differential evolution optimized wavelet neural network classifier is applied at the end to determine whether the given input data is normal or abnormal. The performance of the computerized decision support system is evaluated using a mini database from Mammographic Image Analysis Society (MIAS) and images collected from mammogram screening centres. © Springer International Publishing Switzerland 2015.