[Performance study of teleradiology network systems with CRT monitors: ROC analysis of an observation study of simulated lung nodules].

PURPOSE

To evaluate the diagnostic performance of commercially available CRT monitors used in a teleradiology system by determining the rate of detection of simulated lung nodules.

MATERIALS AND METHODS

Three types of CRT monitors were tested in the observation study. They had matrix sizes of 1024 x 768(16 inches, color), 1024 x 768(20 inches, color) and 1600 x 1125(24 inches, black and white). Twenty chest radiographs were obtained by Fuji computed radiography(FCR) of an anthropomorphic chest phantom with ten simulated nodules on its surface. These FCR films were digitized by a film digitizer with 125 DPI(1024 x 1024 matrix sizes)and 12 bit gray scales, and the image data were transferred from Shinshu University Hospital to other hospitals where interpretation was carried out by the radiologists. Ten radiologists of three hospitals were asked to interpret independently both the original FCR films and the images shown on the CRT monitors and to indicate the presence or absence of simulated nodules on the images by using a five-category rating scale. Receiver operating characteristics(ROC)curves were generated, and the results of interpretation on the FCR films and CRT monitors were compared.

RESULTS

Performance of the all readers was slightly better with the CRT monitors than on the FCR films, although the differences were not statistically significant. There were no statistically significant differences in performance depending on the type of CRT monitor.

CONCLUSION

Performance of the CRT monitors was comparable to that of FCR radiography in terms of interpreting the simulated lung nodules.