Background and purpose: In ischemic stroke, MR perfusion-weighted imaging (PWI) and diffusion-weighted imaging (DWI) mismatch represents tissue at risk for infarction. Infarct growth should only take place in the presence of mismatch, although there have been reports of this occurring. We hypothesized that this observation may be attributable to the presence of undetected "hidden mismatch," which may become obvious when coregistration techniques are used.
Methods: MR PWI/DWI was performed within 48 hours of stroke onset and a final T2-weighted image at ≈3 months. Volumetric-subtraction mismatch volume was defined as PWI minus DWI volume and infarct growth was defined as T2 minus DWI volume. Coregistration mismatch volume was PWI not overlapped by DWI. Mismatch salvage was the proportion of coregistered mismatch tissue that had not progressed to infarction.
Results: Thirty-four patients were studied with MR at a median of 4.9 hours (interquartile range, 2.9-21.1 hours). With the volumetric-subtraction technique, 5 patients (14.7%; 95% CI, 0.05%-0.31%) had infarct growth exceeding mismatch volume, 11 patients (32.0%) had no mismatch and, among these, 3 (27.3%) had infarct growth (median volume, 2.2 mL; interquartile range, 1.0-6.5 mL). All patients had mismatch volume identified by coregistration method that was greater than infarct growth volume. The proportion of this volume salvaged was 77.7% (interquartile range, 63.0%-98.9%).
Conclusions: The illogical finding of infarct growth volume being greater than the presence of mismatch volume can be explained by the presence of "hidden mismatch," which may be detected by coregistration methods.