Objectives: We examined the utility of echoplanar magnetic resonance p
erfusion imaging and diffusion-weighted imaging (DWI) in predicting st
roke evolution and outcome in 18 patients with acute hemispheric infar
ction. Methods: Patients were studied within 24 hours (mean, 12.2 hour
s), subacutely (mean, 4.7 days), and at outcome (mean, 84 days). Compa
risons were made between infarction volumes as measured on perfusion i
maging (PI) and isotropic DWI maps, clinical assessment scales (Canadi
an Neurological Scale, Barthel Index, and Rankin Scale), and final inf
arct volume (T2-weighted MRI). Results: Acute PI lesion volumes correl
ated with acute neurologic state, clinical outcome, and final infarct
volume. Acute DWI lesions correlated less robustly with acute neurolog
ic state, but correlated well with clinical outcome and final infarct
volume. Three of six possible patterns of abnormalities were seen: PI
lesion larger than DWI lesion (65%), PI lesion smaller than DWI lesion
(12%), and DWI lesion but no PI lesion (23%). A pattern of a PI lesio
n larger than the DWI lesion predicted DWI expansion into surrounding
hypoperfused tissue (p < 0.05). In the other two patterns, DWI lesions
did not enlarge, suggesting that no significant increase in ischemic
lesion size occurs in the absence of a larger perfusion deficit. Concl
usions: Combined early PI and DWI can define different acute infarct p
atterns, which may allow the selection of rational therapeutic strateg
ies based on the presence or absence of potentially salvageable ischem
ic tissue.