[A comparison of the monitor and alternator findings of digital thoracic images with the aid of a computer-supported procedure]

Abstract

Purpose: To compare softcopy and hardcopy reading of chest radiographs a software tool was designed for creating lesions with clearly defined size, location and contrast.

Method: An ROC study was performed using a set of chest radiographs with 300 simulated small nodules and linear details displayed on film and on a 1 K monitor. Six observers participated in the study. ROC curves were generated on the basis of 7200 decisions.

Results: There was no significant difference in observer performance overall (AUC = Area under curve). Monitor: 0.84 +/- 0.08 vs. AUC, Film: 0.84 +/- 0.08). Slight, but not statistical significant differences were found in the detection of low-contrast lung nodules and of linear lung details.

Conclusion: Soft-copy reading using a high quality 1 K CRT display provides the same accuracy for the detection of subtle computer-simulated lesions as hardcopy reading. A software tool could be used for assigning optimal characteristics of CRT- and hardcopy displays using clearly defined lesions as a gold standard.