Abstract
An experiment on ability to detect pulmonary nodules was performed to study the relationship between diagnostic accuracy and the period of observers' experience in the interpretation of chest radiographs. A set of 310 chest radiographs, 192 with a malignant pulmonary nodule and 118 normal, were viewed in a random order by 27 physicians. The observers were asked to check for the presence and the site of a nodule on each radiograph, and to give one of four grades of confidence ratings; definitely positive, probably positive, possibly positive, and probably negative. These confidence ratings were scored. Observers were divided into four groups, A to D, according to their periods of experience in the interpretation of chest radiographs. There were significant differences in the nodule detection score between group D, with least experience, and other experienced three groups (p < 0.0001). Receiver operating characteristic (ROC) analysis of observer performance was performed, and the area under the ROC curve was compared in the four groups. The statistical difference was significant between each of the four ROC curves, suggesting a relationship between diagnostic accuracy and observers' experience of interpretation. Sensitivity and positive predictive value, which were calculated by the confidence rating of possibly positive, tended to increase according to the period of experience. However specificity showed no such tendency. © 1993, The Japan Lung Cancer Society. All rights reserved.
Cite
CITATION STYLE
Mori, M., Takabatake, H., Sasaoka, S., Natori, H., & Abe, S. (1993). Experimental Evaluation of Accuracy of Detection of Pulmonary Nodules on Chest X-ray in Relation to the Experience of the Film Reader. Japanese Journal of Lung Cancer, 33(3), 357–362. https://doi.org/10.2482/haigan.33.357
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.