DWI intensity values for the prediction of time from stroke onset in acute stroke

Introduction: In acute stroke, the DWI-FLAIR mismatch allows for the identification of patients eligible for thrombolysis. FLAIR-lesions, however, are difficult to analyze. In comparison, DWI alone may be a suitable biomarker. We analyzed whether a relative DWI intensity threshold (rSI) can identify stroke patients imaged within the thrombolysis time-window. Methods: We retrospectively included patients according to the following criteria: 1) proven stroke, 2) symptom-onset <12h, 3) confirmed lesion in DW-imaging. Patients were dichotomized into two groups (stroke-onset-time [SOT] <4.5h/>4.5h). MR-imaging hardware: 1.5T Intera Master (Philips Medical Systems). A DWI lesion-volume was created and a) mean (DWmean), b) minimum (DWmin) and c) maximum voxel-values (DWmax) of the volume was calculated. Values were normalized: [value/mean value of a representative slice from the unaffected hemisphere]%. DWI-rSIs were correlated with SOT. The ability of DWmin-, DWmean- and DWmax-rSI values to allocate patients to the thrombolysis window was analyzed using receiver operating characteristics (ROC) curve analysis. Results: 44 patients were included (in median: strokeonset-time=2.3 h; age=62 a; NIHSS=8 points; lesion volume=23 ml). 31 patients were imaged within 4.5h after stroke-onset. Correlation of SOT with DWmin, DWmean and DWmax was 0.05, 0.46 and 0.43. Area under the curve (AUC) for DWmean and DWmax was 0.75 and 0.81. DWmin performed poorly (AUC: 0.53). Optimal rSI-thresholds with sensitivity/specificity were: for DWmean 162% with 58%/85%; for DWmax 239% with 71%/93%. Conclusion: DWI-rSIs identified patients within the 4,5h time-window with high specificity. This finding is promising for the use of DW-rSI in acute stroke. Disclosure: Nothing to disclose