ABSTRACT
Background: The non-contrast computed tomography (CT) isa routine brain imaging modality for stroke diagnosis. However,the multiparametric magnetic resonance imaging (MRI) isincreasingly being used for acute stroke diagnosis owing to itsbetter sensitivity in detecting ischaemic brain infarcts.Objectives: To evaluate the best available evidence on thecomparative diagnostic test accuracies between CT andmultimodal MRI for the diagnosis of acute haemorrhagic andischaemic strokes in patients presenting to the emergencydepartment within 6 hours after the onset of stroke-likesymptoms.Methods: MEDLINE-Ovid, AMED, CINAHL and ENBASEelectronic bibliographic databases were searched in Englishfrom January 2004 to August 2014 and supplemented withmanual bibliographic hand search. Prospective andretrospective cohort studies, which directly compared thediagnostic test accuracies between CT (as reference standard)and DW-MRI or GRE-MRI in detecting ischaemic orhaemorrhagic stroke in the same patient population within 6hours after the onset of stroke-like symptoms, were consideredin the review. Only studies with diagnostic outcomes in termsof sensitivity and specificity of MRI versus CT in detectingacute stroke were considered. The quality of the selectedstudies in terms of risk of bias and clinical applicability wereappraised using the QUality Assessment of DiagnosticAccuracy Studies-2 (QUADAS-2) tool.Results: A total of 780 patients were evaluated in the fourselected cohort studies. Three and two studies contributeddata on the diagnosis of acute ICH and ischaemic stroke,respectively. Three studies were of good quality while one wasof poor quality. GRE-MRI exhibited high sensitivity of 83-100%(95% CI) in detecting acute ICH compared to 100% for CTused as the reference standard. CT exhibited poor sensitivity(12-81%) but better specificity (88-100%) in detecting acuteischaemic stroke as compared to DW-MRI, which had asensitivity of 73% (95% CI) and specificity of 92% (95% CI).Conclusions: It appears that GRE-MRI has comparablesensitivity as CT in detecting acute ICH. Similarly, DWI isexcellently sensitive and specific in detecting acute ischaemiclesions. These strongly suggest that MRI is sufficientlyaccurate for routine evaluation of patients with suspected acutestroke in the emergency setting. However, the studiesgenerally lack applicability aspects to the general populationand current clinical practice therefore, warranting furtherresearch. In the meantime, CT and/or MRI tests can be usedfor routine assessment of patients with suspected stroke in theemergency setting.