The performance of four methods for feature enhancement of mammographic images—optimal adaptive neighborhood image processing, automatic spatial filtering, adaptive histogram equalization, and maximal contrast pixel to pixel—are compared. Each method is tested using a database of 50 mammographic images, which included simulated starshaped nodules. A contrast parameter was used to compare the performance of each processing method. It was found that the maximal contrast pixel-to-pixel method gives the highest contrast value. A mammogram including a real subtle lesion is used to illustrate the usefulness of this last contrast enhancement method in clinical diagnosis.