Purpose: Our objective was to asses the interobserver agreement in the detection of pulmonary embolism (PE) with contrast-enhanced helical CT at the main pulmonary, lobar and segmental arteries. A prospective study was carried out in 51 patients with suspected PE. Finally, 29 patients were diagnosed of PE.
Subjects and methods: All patients were studied with helical CT. Images (5 mm collimation, 1.5 pitch factor, 3 mm reconstruction interval) were obtained after bolus contrast injection (120 ml, 4 ml/s, 15 s delay time). All cases were blinded and independently interpreted in three ways: two radiologists with different level of expertise and two expert radiologists reading by consensus. Agreement was evaluated by means of the kappa test.
Results: Kappa values for thrombi detection expressed an excellent agreement at the main (between 0.802 and 0. 946), lobar (between 0.915 and 0.958) and segmental (between 0.879 and 0.718) levels. For all vessels, mean kappa values were similar and excellent for all three combinations of readers. Arteries with more discrepancies were located mainly at the anterior and posterior areas of the upper lobes.
Conclusions: The high degree of agreement shown in this study indicates that helical CT is a reproducible test in the diagnosis of PE to the segmental level. Isolated readings and levels of expertise do not influence agreement.