Computer-aided diagnosis: potential usefulness in diagnostic radiology and telemedicine

Abstract

Computer-aided diagnostic (CAD) schemes are being developed for automated detection of lesions and characterization of normal and abnormal patterns, in an effort to improve the diagnostic accuracy and the consistency of radiologists' final decisions. CAD schemes include the automated detection of masses and clustered microcalcifications in mammograms, as well as the detection of nodules, interstitial infiltrates, cardiomegaly, and pneumothoraces in chest radiographs. In this paper, we demonstrate the current performance, in terms of sensitivity, specificity and speed, of a prototype intelligent radiology workstation which includes a laser digitizer, a high-speed computer, and a jukebox library. The radiology workstation is now being implemented in the mammography section of our department for initial clinical evaluation.