Medical imaging, particularly in breast cancer screening, requires very skilled interpretation only carried out by specially trained radiologists. A key issue is how to train such skilled behaviour? Recent changes to breast imaging has seen the introduction of high resolution digital imaging which facilitates intelligent interactive training. It has also enabled potential computer aided detection of abnormalities. However, this also tends to increase false positive cancer detections. A series of experiments are reported which examine the role of eye gaze and expertise in inspecting these images. It is proposed that current training approaches could be augmented by including aspects of the eye gaze behaviour of expert screening radiologists together with computer aided detection in new practical interactive training systems.