Diagnostic performance of magnetic resonance imaging for acute pulmonary embolism: a systematic review and meta-analysis

With ongoing technical developments, magnetic resonance imaging (MRI) has notably evolved for the assessment of the pulmonary vasculature. However, uncertainty persists about the performance of MRI for the diagnosis of acute pulmonary embolism (APE).

To clarify the comprehensive role of MRI in diagnosing APE.

Studies were identified through a search of Pubmed and Ovid databases, and the QUADAS-2 tool was applied for quality assessment of the included studies.

Fifteen studies based on patients and nine based on vessels were retrieved. The patient-based analysis yielded an overall sensitivity of 0.75 (95% confidence interval, 0.70-0.79) and 0.84 (0.80-0.87) for all patients and patients with technically adequate images, respectively, with an overall specificity of 0.80 (0.77-0.83) and 0.97 (0.96-0.98) and a pooled diagnostic odds ratio (DOR) of 51.07 (18.36-142.05) and 155.22 (86.83-277.47). On average, MRI was technically inadequate in 18.89% of patients (range, 2.10%-27.70%). A direct comparison of different MRI modalities showed that the combined MRI test had the highest pooled DOR and the lowest proportion of inconclusive images. Of note, heterogeneity and moderate quality were observed. On a vessel basis, the MRI had high sensitivity and specificity in larger-order vessels, but a significantly lower sensitivity of 0.55 (0.50-0.60) for subsegmental APE.

On a patient-based level, MRI yields high diagnostic accuracy for the detection of APE, especially in technically adequate images, and the inconclusive MRI examinations mainly result from motion artifact and poor arterial opacification. The combined MRI test appears to be a more promising diagnostic tool with greater power of discrimination than single techniques. From a vessel-based perspective, MRI exhibits a high diagnostic capability with proximal arteries, but lacks sensitivity for peripheral embolism.