The cover illustrates the difference between the ROC (left) and FROC (right) paradigms. The two images are of the same breast. The arrows point to real lesions and the crosses to suspicious regions found, by the radiologist's search mechanism. LEFT: in the ROC paradigm the radiologist summarizes the information in the 3-suspcious regions by assigning a single rating to the image. Assuming that the lesion near the center is highly suspicious for disease and assuming a 5-point ROC rating scale (1: definitely diseased, 5: definitely diseased), the ROC rating would likely be 5. The ROC rating applies to the image, not to any specific region of the image. RIGHT: the dark crosses are regions that were suspicious enough to be marked. The dark cross near the central real lesion is a lesion localization. Assuming a 4-point FROC rating scale with 4 being most suspicious for presence of lesion, this lesion would be rated 4. The other dark cross is a non-lesion localization - the radiologist mistook a normal tissue variant for a lesion. This was rated 1, as it just crossed the reporting threshold. The other light cross did not reach the reporting threshold and was consequently not marked. If none of the suspicious regions crosses the reporting threshold and / or there were no suspicious regions, the image would not be marked. The corresponding ROC rating would be 1. A 5-point ROC rating scale corresponds to a 4-point FROC rating scale.