Comparison of observer performance reading from a video CRT vs. reading from film

The developing trend for all medical images to be in digital form raises an important question: should diagnostic interpretations be made directly from a CRT or from some form of hard copy recording of the CRT? This question was studied by having two observers read a series of digital images of the thyroid including both conventional scintigrams and computer-generated tomograms. The CRT images were photographed in a carefully controlled fashion. Images were read once from film and a second time directly from the CRT. No significant difference in performance could be found for either display modality based on an ROC analysis. Although based on uncomplicated images and analysis, this result suggests that the decision to read from film or from the CRT can be made on other grounds than observer performance. This important result needs confirmation with more complex images such as ultrasound or CT.