AJR:194, January 2010
Leandro A. Espinosa, Aine M. Kelly, Colleen Hawley, Radha Inampudi,Aamer Chughtai, Prachi Agarwal, Shokoufeh Khalatbari, James Myles, Ella Kazerooni
OBJECTIVE. The purpose of this study was to determine whether the view used, mul-tiplanar or axial, for image interpretation at pulmonary CT angiography for suspected acute pulmonary embolism alters the diagnostic confidence, accuracy, and interpretation time of cardiothqracic radiology specialists and radiology residents.
MATERIALS AND METHODS. Patients who underwent 50 consecutive pulmonary 64-MDCT angiographic examinations formed the study group (18 men, 32 women; mean age, 53 years; range, 19-93 years). Three blinded cardiothoracic faculty radiologists and three blinded radiology residents reviewed each case independently initially using only ax-ial display mode and later using multiplanar reformation (MPR) in any x-, y-, or z-axis. The presence of pulmonary embolism in the main through subsegmental pulmonary arteries was scored on a 5-point scale; diagnostic confidence for the overall examination was scored on a 3-point scale; and interpretation time was recorded. A surrogate reference standard consist-ed of either faculty agreement or, in cases of disagreement, adjudication by another, senior faculty member. Statistical analysis included the Kendall coefficient (IV), receiver operating characteristics curves, and a univariate repeated measures model.
RESULTS. Interobserver agreement between specialists on the diagnosis of pulmonary embolism was good for axial viewing (W= 0.72) and for MPR viewing (W= 0.79). Interobserv-er agreement between residents was good for axial viewing (W = 0.62) and for MPR viewing (W= 0.70). Reader confidence improved among all readers with MPR viewing, but the differ-ence did not reach statistical significance. Interpretation time with MPR was significantly lon-ger for two of the three specialists and significantly shorter for two of the three residents.
CONCLUSION. Use of MPR for viewing increased the reader agreement and interpre-tation time of cardiothoracic specialists but increased reader agreement between residents and might have decreased interpretation time. All readers had a trend toward increased confidence.