Diagnostic Test Accuracy of DWI-FLAIR Mismatch for the Identification of Patients with Hyperacute Ischemic Stroke: A Systematic Review and Meta-analysis
Hussain Alkhars1, Mahmoud Omar2, Paul Wang3, Dottie Yu4, Mohammed Almarkhan1, Michael DeGeorgia3, Wassim Malak2
1George Washington University School of Medicine, 2Radiology, 3Neurology, University Hospitals Cleveland Medical Center, 4Cleveland Clinic Lerner College of Medicine
Objective:
To determine the pooled diagnostic accuracy of two qualitatively assessed FLAIR signal interpretation thresholds (conservative vs liberal) for defining DWI/FLAIR mismatch in estimating stroke onset time in hyperacute ischemic stroke
Background:
Accurate estimation of stroke onset is critical in wake-up stroke protocols, as inappropriate thrombolysis may lead to catastrophic hemorrhage. DWI/FLAIR mismatch is widely used to approximate onset timing, however, substantial variability exists in how subtle FLAIR hyperintensity is interpreted. Some studies classify faint FLAIR signal as a positive mismatch (liberal threshold), whereas others regard it as negative (conservative threshold). To address this inconsistency, we conducted a clustering meta-analysis comparing the diagnostic performance of these two visual thresholds for identifying stroke onset within 4.5 hours.
Design/Methods:
This systematic review and meta-analysis was prospectively registered (PROSPERO: CRD420251107468). PubMed, Embase, and Scopus were searched for studies published between 2000 and 2025. Studies were required to include patients with known time from symptom onset (reference standard). Two independent reviewers performed screening and data extraction. Sensitivity and specificity were pooled using a random-effects model with 95% CIs.
Results:
Of 1,387 unique references, 28 studies met inclusion criteria. For identifying stroke onset <4.5 hours using the conservative FLAIR threshold, pooled sensitivity was 0.55 (95% CI, 0.47–0.64) and specificity was 0.84 (95% CI, 0.79–0.88) (15 studies; n=2,756). Using the liberal threshold, sensitivity increased to 0.74 (95% CI, 0.67–0.81) with a corresponding decrease in specificity to 0.71 (95% CI, 0.56–0.82) (8 studies; n=1,165).
Conclusions:
Compared with the conservative approach, the liberal FLAIR threshold demonstrated higher sensitivity with slightly lower specificity in identifying patients likely to be within the therapeutic window. These defined data can provide a framework for clinicians to optimize patient selection for reperfusion therapy.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.