Diffusion-Weighted MR Imaging and Ischemic Stroke

Magnetic resonance imaging (MR) of ischemic cerebrovascular events with standard T1and T2-weighted spin-echo techniques is widely employed. Standard MR demonstrates vascular changes such as the absence of a flow void and the presence of arterial enhancement within 2-4 hours of ischemic stroke onset in some cases ( 1 ). The earliest morphologic change in the brain parenchyma, swelling, is visible at the same timepoint, but obvious, quantifiable signal changes within the brain parenchyma are usually not detectible for 6-12 hours. Recent reports suggest that computerized tomography (CT) can demonstrate early parenchymal abnormalities with almost equal effectiveness (2, 3). Both standard MR and CT cannot reliably outline the extent of ischemic stroke injury during the first few hours. The ability to reliably image the location and extent of ischemic stroke within the first few hours would have an enormous impact on clinical diagnostic accuracy and might provide a useful adjunctive method to assign and assess therapy. Diffusion-weighted MR imaging (DWI) is a new methodology that can apparently quantitatively demonstrate the extent of ischemic injury very rapidly. DWI signals are generated by evaluating the translational movement (Brownian motion) of water molecules, abundantly present in cerebral tissue, as compared to standard T1and T2weighted MR where image contrast is related to differences in T1and T2-relaxation times, proton density, and flow (4). Diffusion weighting of an MR study can be accomplished by integrating strong, short dephasing and rephasing magnetic field gradients into the standard spin-echo imaging pulse sequence (5). DWI studies in experimental stroke models were initially performed by Moseley et al in a cat permanent middle cerebral artery occlusion (MCAO) model (6). Hyperintensity on DWI, corresponding to the ischemic region at postmortem, was observed as early as 45 minutes after stroke onset. The same group, using a reversible MCAO model in rats, observed DWI changes 14 minutes after vascular occlusion and demonstrated that these abnormalities on DWI reverted to normal when the occlusion was removed after 33 minutes (7). Standard T2weighted MR was unremarkable at these early timepoints, while 31 P and H MR spectroscopy demonstrated changes compatible with ischemic injury within the region of DWI abnormality . Minematsu et al, using the same rat MCAO occlusion model, quantitatively evaluated DWI in early ischemic stroke (8). They observed that DWI ischemic lesion areas at 30 minutes after stroke onset (Fig. 1) were highly correlated with the area of infarction demonstrable at postmortem in permanently occluded animals. In a second experiment, Minematsu et al reperfused their animals after 1 or 2 hours of temporary ischemia (9). DWI studies were performed during and after occlusion and the DWI lesion areas (hyperintense regions) were compared. Within 1 hour of temporary occlusion, 55% of the DWI lesion area observed during occlusion reverted to normal on the postreperfusion DWI studies (P < .01). With 2 hours of temporary occlusion, only 17% of the prereperfusion DWI lesion area was reversible. Postmortem infarct areas were highly correlated with the postocclusion DWI area of hyperintensity . These animal stroke model studies demonstrate that DWI can rapidly reveal the location and extent of ischemic brain injury and that DWI can be used to document the reversibility of ischemic injury by reperfusion. The study by Chien et al extends these animal DWI studies to stroke patients, although at a later timepoint ( 1 0). The authors demonstrate that