The invention described provides a method of quantitatively evaluating one or more of the likelihood. severity and progression of a disease from medical images comprising processing medical images of a test subject to derive one or more feature space values characteristic of a disease-dependent image attributes, comparing the feature space values to those of a previously established database from medical images of known health} and known diseased subjects, wherein the comparing is based on feature space values that best discriminate between health and diseased subjects, summing a weighted distance of discriminant feature space values of the test subject to those of at least one of the mean feature space value of the healthy subjects and the mean feature space value of the diseased subjects, and providing from the summing a single number which is indicative of at least one of disease likelihood. severity and progression.