Sophisticated computer techniques are being applied to the interpretation o
f radiologic images with ever-increasing success. Computer vision and artif
icial intelligence form the basis for automated detection and characterizat
ion of lesions appearing on mammograms. Information of this kind can be sup
plied to radiologists to help improve their diagnostic performance. Compute
r-based methods have the potential to increase the yield of screening mammo
graphy for the detection of early breast cancer or to decrease the number o
f biopsies done for benign disease.