AJR:197, December 2011
Edward Y. Lee Evan J.Zucker Jason Tsai Donald A.Tracy Robert H.Cleveland David Zurakowski Phillip M.Boiselle
OBJECTIVE. The purpose of this study was to determine whether the use of multiplanar reformatted (MPR) MDCT images in the diagnosis of pulmonary embolism (PE) in children by faculty pediatric radiologists and radiology residents affects reader performance param-eters and adds diagnostic value compared with the use of axial MDCT images alone.
MATERIALS AND METHODS. This retrospective study was conducted with the im¬ages of 60 children (28 boys, 32 girls; mean age, 14.7 ± 3.5 years; range, 3.2-18 years) who consecutively underwent pulmonary CT angiography (CTA) for clinically suspected PE. Two faculty pediatric radiologists and two radiology residents independently reviewed images from each study initially using only axial MDCT images and later using MPR MDCT im¬ages in any x-, y-, or z-axis. Diagnostic accuracy, confidence level (1-5 ordinal scale), and interpretation time for MPR MDCT images were compared with those for axial MDCT im-ages by use of the McNemar test and paired Student t test. The kappa coefficient was calcu¬lated to assess interobserver agreement. Diagnostic accuracy was compared between faculty pediatric radiologists and radiology residents by logistic regression analysis, and confidence level, interpretation time, and added diagnostic value were evaluated by analysis of variance.
RESULTS. Nine of 60 pulmonary CTA studies (15%) were found to show PE. Diagnos¬tic accuracy in detection of PE ranged from 91.7% to 100% (mean, 96.7%) with no significant differences between axial and MPR MDCT images (McNemar test for matched binary pairs, p > 0.50 for each reviewer). Logistic regression showed no significant difference between fac¬ulty pediatric radiologists and radiology residents in diagnostic accuracy in detection of PE on axial MDCT images (p = 0.48) or MPR MDCT images (p = 0.24). Confidence level and interobserver agreement were significantly higher and average interpretation time was lon¬ger in the evaluation of PE with MPR MDCT images than with axial MDCT images for all reviewers (p < 0.001). Compared with faculty pediatric radiologists, radiology residents had significantly greater increases in confidence level, interobserver agreement, interpretation time, and added diagnostic value using MPR MDCT images than they did using axial MDCT images to diagnose PE (p < 0.001).
CONCLUSION. Use of MPR MDCT images for pulmonary CTA in the diagnosis of PE in children significantly increases confidence, interobserver agreement, and interpretation time among faculty pediatric radiologists and radiology residents. Because use of MPR MDCT im¬ages results in significantly greater improvements in reading parameters for residents than for faculty members, the routine use of this technique by trainees should be encouraged.