IMAGING

Acute Stroke The Melbourne group reported further application of the positron-emission tomography (PET) hypoxia marker Flabeled fluoromisonidazole (F-MISO).1–3 In one article, they further developed and validated their novel imaging methodology to map the penumbra using this tracer.1 Applying this method, they elegantly showed that hypoxia affects white matter to a similar degree and extent as gray matter, suggesting the former has at least as high a resistance to ischemia than the latter and that its salvage should help to maximize benefit of treatment.2 In a third article,3 they report that the impact of hypoxic tissue escaping infarction on subsequent clinical recovery is similar whether the tissue is identified within 12 hours of, or in the 12and 48-hour interval after stroke onset, documenting that F-MISO identifies true penumbral tissue, and that, consistent with earlier evidence, appropriate interventions should improve outcome even beyond 24 hours. The year 2004 has seen the first, long-awaited, articles reporting direct PET and diffusion-weighted imaging (DWI)/ perfusion-weighted imaging (PWI) comparisons.4–6 Using stateof-the-art diffusion tensor imaging (DTI) and fully quantitative PET as gold standard, Guadagno et al4 documented that the acute DWI lesion not only contains irreversibly damaged, but also penumbral tissue, in agreement with studies showing potential reversibility of the DWI lesion, while even severe apparent diffusion coefficient decreases can be found in either tissue category. One therapeutic implication is that a matched DWI/PWI lesion may still represent, at least in part, salvageable tissue. Comparing the predictive value of DWI and CFlumazenil (FMZ) for final infarction, Heiss et al5 found that although both have similar overall predictive power (around 84% of the final infarct), false-positives occurred with DWI but not with FMZ, consistent with the Guadagno et al findings.4 Assessing the validity of PWI to assess the at-risk tissue by means of PET, Sobesky et al6 concluded that overall the simple DWI-PWI mismatch overestimates the penumbra, but the use of time-to-peak (TTP) delay maps helps toward solving this problem, with TTP delays 4 s being best suited. These results apply specifically to the TTP method of deriving magnetic resonance imaging (MRI) perfusion maps; other methods, such as mean transit time (MTT) maps, may be less prone to overestimate the region of symptomatic ischemia.7 Thijs et al8 found large variations in hypoperfusion lesion size with different arterial input function (AIF) locations used to derive MRI perfusion maps. They found that the AIF derived from the contralateral middle cerebral artery (MCA) gave ischemic volumes that most accurately predicted follow-up lesion volume. The sensitivity of MRI relative to computed tomography (CT) has now been established for acute hemorrhage diagnosis in patients with focal stroke symptoms of less than 6 hours duration. Susceptibility-weighted MRI, most commonly the gradient-recalled echo (GRE) sequence, is used for that purpose. Fiebach et al9 found near perfect discrimination of hemorrhagic from ischemic stroke on MRI in a sample containing 62 cases of each, obtained in 6 hours: 100% sensitivity among experts; 95% sensitivity among medical students after a brief tutorial. Kidwell et al10 prospectively investigated a broad sample of 200 stroke patients, in which MRI followed by CT was obtained in 6 hours. The consensus of 4 experts’ independent, blinded reads found MRI superior for detecting any hemorrhage (because of MRI sensitivity to microand other chronic bleeds) and equivalent for acute hemorrhage, which was diagnosed by both modalities in 25 patients. There were 8 discrepant reads, 4 in either direction, for acute hemorrhage. Three of the discrepant cases of acute hemorrhage on CT were also diagnosed by MRI but classified incorrectly as chronic hemorrhage. However, 4 cases of acute hemorrhagic transformation on MRI were missed on CT. Smaller retrospective series have also reported cases of hemorrhagic transformation evident on susceptibility-weighted MRI but not CT following thrombolytic therapy,11,12 including cases where CT findings were equivocal because of residual angiography contrast.11 As evidence continues to confirm that prethrombolysis severity of clinical or MRI parameters predict outcome with recanalization, so does evidence that resolution of perfusion deficits is predictive of clinical recovery. Singer et al13 reported that greater amounts of at-risk tissue did not progress to infarct among patients who had recanalized relative to those who had not in a sample of 17; 80% of the MTT defect and 78% of the TTP 2 s delayed region did not progress to infarct on follow-up imaging. Chalela et al14 reported in a sample of 42 patients that resolution of at least 30% of the volume of MTT defect by 2 hours after standard IV tissue plasminogen activator treatment was associated with excel-