Multidetector computed tomography (MDCT) is frequently used as the first-line technique to directly visualize pulmonary embolism (PE) with a reported sensitivity and specificity of 83% to 100% and 89% to 97%, respectively (1). However, for rapid and profound clinical decision making and prognostication, accurate detection of right ventricular (RV) dysfunction and pulmonary perfusion impairment remains pivotal because such patients are at higher risk of an adverse outcome (2).

The present study used a single-session, multicomponent cardiovascular magnetic resonance (CMR) imaging protocol facilitating, first, the visualization of pulmonary thrombi; second, the precise evaluation of RV volumes/function; and third, pulmonary perfusion on a parenchymal level.

Twelve patients (5 men; mean age, 63.1 ± 19.8 years) with documented PE on contrast-enhanced MDCT (Somatom Sensation 64, Siemens Medical Solutions, Forchheim, Germany) and 14 healthy controls (7 men; mean age, 39.3 ± 16.0 years) underwent multicomponent CMR imaging. On MDCT, the presence of PE was defined by visualization of thrombi in >1 image slice in at least 2 imaging planes or if vessel truncation implied total occlusion.

Multicomponent CMR imaging (Achieva 1.5-T magnetic resonance [MR] system equipped with a 32-channel coil, Philips, Best, the Netherlands) was performed as follows: 1) standard cine imaging (repetition time/echo time [TR/TE]/flip angle, 3.4 ms/1.7 ms/60°; spatial resolution, 1.8 × 1.8 × 8 mm³) using multiple standard views (i.e., short-axis orientation with full ventricular coverage and 2-, 3-, and 4-chamber orientation); 2) high-resolution, contrast-enhanced dynamic pulmonary perfusion imaging in coronal slice orientation (multislice 2-dimensional, k-t accelerated magnetic resonance sequence; TR/TE/flip angle, 2.5 ms/0.9 ms/15°; spatial resolution, 1.4 × 1.4 × 12 mm³; k-t factor, 6.0; net acceleration factor, 5.0) during a peripheral intravenous injection of a gadolinium-diethylenetriaminepentaacetic acid bolus (0.1 mmol/kg bodyweight; injection rate, 4.0 ml/s) (3); 3) a 3-dimensional MR angiography scan (TR/TE/flip angle, 2.9 ms/1.0 ms/30°; spatial resolution, 1.4 × 1.4 × 1.6 mm³) during pulmonary arterial contrast passage of a second gadolinium-diethylenetriaminepentaacetic acid bolus (0.1 mmol/kg bodyweight; injection rate, 3.0 ml/s). All patients received supplemental oxygen via a nasal cannula (2 to 6 l/min).

RV volumes were determined following Simpson's rule. Pulmonary CMR perfusion was evaluated visually for the presence of lobar/segmental areas of relative hypoenhancement. Quantitative pulmonary CMR perfusion parameters were derived from the signal-intensity time curves in each of the 5 pulmonary lobes and in the peripheral 2-cm border area of both lungs: relative peak enhancement (%), maximum peak enhancement (%), wash-in rate (arbitrary unit), area under the curve (AUC), and time to peak enhancement (seconds). To detect PE on 3-dimensional MR angiography, the identical criteria used for MDCT were applied.

Group comparisons of continuous variables were performed using the Student t test or Kruskal-Wallis test, as appropriate. Categorical data were compared using the Fisher exact test. Receiver-operating characteristic analysis was done to determine the accuracy of pulmonary CMR perfusion testing. All statistical analyses were performed with SPSS version 16.0 (SPSS Inc., Chicago, Illinois).

PE patients demonstrated significantly reduced systolic RV ejection fraction compared with controls (47.1 ± 10.4% vs. 57.2 ± 2.9%, p = 0.002). The diagnosis of PE was equally accurate on a per-patient basis with both MDCT and MR angiography. In all PE patients (12/12, 100%), pulmonary CMR perfusion imaging clearly delineated lobar, segmental, or peripheral wedge-shaped areas of relative hypoperfusion (normal pulmonary perfusion: (Figure 63_gr1)A,Online Video 1; pathological pulmonary perfusion: (Figure 63_gr1)B and (Figure 63_gr1)C, Online Video 2). Hence, in PE patients, the sensitivity of pulmonary CMR perfusion imaging, on a per-patient, per-lung, and per-lobe basis, was 100%, 90.0%, 71.1%, respectively, with corresponding kappa values of 1.0, 0.80, and 0.69. Comparing CMR perfusion of pulmonary lobes being affected by PE with normal lobe relative peak enhancement (172 ± 157% vs. 542 ± 213%, p < 0.001), maximum peak enhancement (314 ± 198% vs. 691 ± 264%, p < 0.001), wash-in rate (89 ± 72 vs. 184 ± 77, p < 0.001), and AUC (3,212 ± 2,269 vs. 7,215 ± 3,199, p < 0.001) were significantly reduced and time to peak enhancement was significantly prolonged (median [interquartile range]: 17.1 s [12.0 to 23.6 s] vs. 8.8 s [8.0 to 10.1 s], p < 0.001).

Using the peripheral lung measurements, receiver-operating characteristic analysis was done to determine the perfusion parameter with the highest diagnostic value for the detection of PE: the AUC for maximum peak enhancement (0.88, 95% confidence interval [CI]: 0.79 to 0.98) was best, but relative peak enhancement (0.88, 95% CI: 0.77 to 0.99), wash-in rate (0.86, 95% CI: 0.75 to 0.97), AUC (0.83, 95% CI: 0.72 to 0.94), and time to peak enhancement (0.78, 95% CI: 0.63 to 0.93) demonstrated good discriminatory power as well.

Reportedly, CMR was considered highly sensitive for the diagnosis of central PE, but showed reduced sensitivity for the detection of segmental and especially subsegmental emboli.

In the current study, contrast-enhanced high-resolution CMR pulmonary perfusion used the signal enhancement resulting from pulmonary arterial blood supply with decreased or absent signal indicating pulmonary arterial obstructive disease. Hence, CMR pulmonary perfusion imaging primarily visualized the consequences of PE on a parenchymal level. Using the parameter of peak signal enhancement, PE could be diagnosed as lobar, segmental, or subsegmental with sharply delineated perfusion defects.

During a single-session examination, multicomponent CMR delivered all functional and morphological information (i.e., pulmonary arterial supply, RV function, and pulmonary perfusion) directly and essentially related to adequate and rapid clinical decision making. Such data may serve as a profound basis for determination of therapeutic options in patients suspected of having a PE.

The present study has some limitations. Because of limited access to the patient in the magnetic resonance scanner environment and related safety concerns, hemodynamically unstable (i.e., severely hypotensive) patients were not examined. Total CMR imaging time was ∼20 min with an additional 10 min for evaluation/interpretation of image data. Hence, although the duration of image data acquisition was prolonged compared with MDCT, the multicomponent CMR imaging approach accurately delivered RV functional parameters (i.e., volumes together with regional and global [dys]function) considered essential for further therapeutic decision making.

In conclusion, taking full advantage of the radiation-free and noninvasive multicomponent CMR approach, serial imaging could potentially be used to assess and compare the efficiency of medical measures designed to improve pulmonary perfusion and/or RV function.