An adaptive, multiscale method for mass detection using a circular ring template and a nonparametric test is presented. Training data corresponding to the local background intensity level is extracted from the outer ring of the template. Test data for mass detection is obtained from the inner disk. The statistical nature of the background, specifically the tail structure of the unknown distribution, is estimated via tolerance intervals. Mass detection is based on the fraction of test data within the tail. Empirical Receiver Operating Characteristic (ROC) curves show a Sensitivity of 1 and a False Positive Rate of 1.40 per image on 30 images.