Purpose: To evaluate the impact of dose reduction in multidetector computed tomography pulmonary angiography (CTPA) for detection of pulmonary embolism (PE).
Materials and methods: After induction of PE in 6 anesthetized pigs, a 64-detector-row CTPA was performed at 3 different dose protocols: A (120 kV/120 mAseff.), B (120 kV/80 mAseff.), and C (80 kV/80 mAseff.). Images were evaluated by 2 radiologists independently. A high-dose CTPA (120 kV/250 mAseff.) served as a reference standard. Sensitivity, specificity, and positive and negative predictive values were calculated and compared using the Wilcoxon test. Interobserver agreement was determined by calculation of κ values. Radiation exposure and objective image parameters were assessed and compared with a 2-sided t test.
Results: In the reference scan, a total of 94 emboli were detected: 17 in the main and lobar pulmonary level (category A), 47 in the segmental level (category B), and 30 in the subsegmental level (category C). All protocols reached high diagnostic accuracy in the detection of PE in category A. No significant difference was observed between protocols A and B in the detection rate of segmental and subsegmental PE (sensitivity: 93.6% and 91.5% vs 85.1% and 87.2%; positive predictive value: 100% and 97.7% vs 97.5% and 95.3%). Interobserver agreement was excellent at the segmental (κ=0.97 and 0.94) and subsegmental levels (κ=0.94 and 0.92). Using protocol C, the detection rate of segmental and subsegmental emboli was significantly impaired and interobserver agreement was significantly inferior (sensitivity: 46.8% and 44.6% at segmental and 56.7% and 50.0% at subsegmental level; κ=0.70 and 0.60, respectively; P<0.05). Vessel attenuation was significantly higher in protocol C, [710.3 Hounsfield units (HU) ±200.6 SD] than in protocols A (414.4 HU±82.5 SD) and B (428.8 HU±78.9 SD) (P<0.001). Signal-to-noise and contrast-to-noise ratios were significantly decreased in protocols B (55.6; 45.8) and C (44.3; 39.4) compared with protocol A (62.1; 52.3) (P<0.05). Radiation dose was significantly reduced between protocols A and B and protocols B and C [volume CT dose index (CTDIvol): 7.1 vs 4.7 vs 1.4 mGy; dose-length product: 199.5 vs 132.1 vs 39.4 mGy·cm and E: 3.39 vs 2.25 vs 0.6 mSv; P<0.05].
Conclusion: This animal study proves the diagnostic use of a low-dose CTPA protocol with 80 mAseff. at 120 kV for PE evaluation without significant loss of diagnostic accuracy compared with a standard protocol with 120 mAseff. Using an "ultra-low"-dose CTPA protocol (80 kV; 80 mAseff.), emboli detection at the segmental and subsegmental level is significantly impaired, whereas evaluation of central emboli is still feasible with high diagnostic accuracy.