The accuracy of MRI in diagnosis of suspected deep vein thrombosis: systematic review and meta-analysis

Fiona C Sampson, Steve W Goodacre, Steven M Thomas, Edwin J R van Beek

Research output: Contribution to journalArticlepeer-review


Magnetic resonance imaging (MRI) may be used to diagnose deep vein thrombosis (DVT) in patients for whom ultrasound examination is inappropriate or unfeasible. We undertook a systematic review of the literature and meta-analysis to estimate the diagnostic accuracy of MRI for DVT. We searched databases of medical literature and citation lists of retrieved articles. We selected studies that compared MRI with a reference standard in patients with suspected DVT or suspected pulmonary embolus, or high-risk asymptomatic patients. Data were analysed by random effects meta-analysis. We included 14 articles in the meta-analysis. Most compared MRI with venography in patients with clinically suspected DVT. The pooled estimate of sensitivity was 91.5% (95% CI: 87.5-94.5%) and the pooled estimate of specificity was 94.8% (95% CI: 92.6-96.5%). Sensitivity for proximal DVT was higher than sensitivity for distal DVT (93.9% versus 62.1%). However, pooled estimates should be interpreted with caution as estimates of both sensitivity and specificity were subject to significant heterogeneity (P<0.001). Individual studies reported sensitivity ranging from zero to 100%, while specificity ranged from 43 to 100%. MRI has equivalent sensitivity and specificity to ultrasound for diagnosis of DVT, but has been evaluated in many fewer studies, using a variety of different techniques.

Original languageEnglish
Pages (from-to)175-81
Number of pages7
JournalEuropean Radiology
Issue number1
Publication statusPublished - Jan 2007


  • Humans
  • Magnetic Resonance Imaging
  • Phlebography
  • Reproducibility of Results
  • Sensitivity and Specificity
  • Venous Thrombosis


Dive into the research topics of 'The accuracy of MRI in diagnosis of suspected deep vein thrombosis: systematic review and meta-analysis'. Together they form a unique fingerprint.

Cite this