Computer classification can improve diagnostic accuracy by helping radiologists to be more accurate. An effective diagnostic aid does not function the same way as a single, independent detection system. The ability of computer classification to improve the diagnosis ofbreast cancer depends on the accuracy of radiologists, the accuracy of computer analyses, and the ability of radiologists to incorporate the computer classification results into their own diagnostic decision-making. The clinical potential of a classification technique must be assessed with an observer performance study in which radiologists read mammograms with and without the computer aid and their diagnostic accuracy in both cases is compared. Because this type of study provides the most direct evidence to date that computer classification can help improve breast cancer diagnosis, we will describe the following experiments in more detail.
Was this article helpful?