Background and purpose: The mismatch between Perfusion image (PWI) and diffusion weighted (DWI) images using magnetic resonance (MR) is increasingly being applied in patient selection for therapeutic trials. Two approaches to the calculation of the mismatch volume exist, the commonly used volumetric and the more precise co-registration method, the latter of which considers lesion topography. We hypothesized that there are differences in the mismatch volume analysed by each method and that these are time dependent.
Methods: Patients within 48 hours of ischaemic stroke onset had baseline MR PI /DWI mismatch and T2 outcome volumes at 3 months. Volumetric mismatch volume was defined as PI minus DWI lesion. Co-registration mismatch volume was defined as the PI defect lesion not overlapped by the co-registered DWI lesion.
Results: 72 patients of median age 74.0 years were studied. Median baseline MR was at 5.9 hours (IQR 3.0, 20.4 hours) after stroke onset. Consistent underestimation of the mismatch volume occurred using the volumetric method (volumetric median 9.3 ml, IQR 0, 63ml. co-registration median 20.1 ml, IQR 3.2, 69.8 ml p<0.0001). This difference increased with time from stroke onset (p=0.006).
Conclusions: Volumetric analysis consistently underestimates the PWI/DWI mismatch volume when compared to the more precise co-registration method. This effect increases with time.