A ROC study of detection accuracy of single pulmonary nodules by image diagnostic monitors with different resolution

CHENG Yong,ZHANG Xue-lin,CHEN Wei-Guo,YANG Hui,WU Yuan-kui,g D,LIU Guo-qing
DOI: https://doi.org/10.3969/j.issn.1006-9011.2008.05.007
2008-01-01
Journal of Medical Imaging
Abstract:Objective:The purpose of this study was to assess the influence and diagnostic value of professional image diagnostic monitors with different resolution used in PACS by determining the rate of detection of single pulmonary nodules (SPNs) on soft copy images of chest direct digital radiography. Methods:100 chest radiographs showing a solitary noncalcified pulmonary nodule, 24 patients with multi-pulmonary nodules and 15 cases without pulmonary nodule were selecteded for this study, and CT scans of the same patient served as the gold standard for the presence of pulmonary nodules. Two experienced radiologists and three residents interpreted 139 chest images four times by using diagnostic workstation monitors of PACS in two brands: Barco MGD521 MKII (2,560×2,048 pixels,75Hz) 21 in professional RCT and EIZO RadiForce G20 (1,200×1,600 pixels,60Hz) 21 in professional LCD. Each time, the radiologists used one kind of display to interpret the images with image magnification or not and marked their confidence of the nodules, the presence or absence of SPNs was investigated with a five-point confidences rating scale: ① definitely positive; ② probably positive; ③ indeterminate;④ probably negative, and ⑤ definitely negative. The interpreting results were analyzed by SPSS 13.0 software-the receiver operating characteristic (ROC) curve. Results: ① The observer performance (Az values) of all radiologists increased with using magnification, especially for the residents. For experienced radiologists, the performance of Barco RCT with or without magnification was comparable. In all monitors, the observer performance of experienced radiologists was higher than that of residents (P<0.05); ② For residents, the performance on EIZO LCD with magnification (Az=0.674) was superior to Barco RCT withoutmagnification (Az=0.651) (P=0.044<0.05). But for experienced radiologists, the performance was opposite. To all observers, there was no significant difference between EIZO LCD with magnification (Az=0.702) and Barco RCT without magnification (Az=0.703) (P=0.475>0.05). Conclusion:① The usage of enlargement function may further boost the imaging resolution, and obviously enhances the diagnosis potency. Compared with experienced chest radiologists, the increment of image resolution shows relatively improved diagnostic accuracy among less experienced radiologists, in the certain degree, it can also decrease the influence of which the individual difference creates; ② Proceeding from the consideration of the cost efficiency, the administrative offices may arrange less experienced radiologists to use lower resolution professional monitor for primary diagnostic tasks, but must remain the conventional enlargement post-processing function. The experienced radiologists should use standard professional monitors in order to guarantee finally diagnostic reliability.
What problem does this paper attempt to address?