Clinical experience with diffusion-weighted MR in patients with acute stroke.

PURPOSE Our purpose was to evaluate the clinical efficacy, sensitivity, and specificity of echo-planar diffusion-weighted MR imaging in patients with acute infarction. METHODS We retrospectively analyzed 194 cases of acute ischemic stroke diagnosed clinically within 24 hours of onset and studied with echo-planar diffusion-weighted MR imaging. Examinations were considered to be positive for infarction when an increase in signal was noted on images acquired at a high b value but absent on images with a low b value. A final clinical diagnosis of acute stroke was used as the standard of reference. A subset of 48 patients scanned within 6 hours was also analyzed. RESULTS Diffusion-weighted MR imaging studies were positive in 133 of 151 cases of infarction (88% sensitivity) and negative in 41 of 43 cases with no infarction (95% specificity). Two cases identified as positive on diffusion-weighted images had nonischemic diagnoses (1.5% false-positive rate). Diffusion-weighted imaging had a positive predictive value of 98.5% and a negative predictive value of 69.5%. Use of T2-weighted sequences as well as diffusion-weighted imaging produced no false-positive findings. Of the negative scans, 69.5% corresponded to transient ischemic attacks or infarcts (mostly small brain stem infarcts). When only cases scanned within 6 hours of onset were considered, the sensitivity rose to 94% and the specificity to 100%. CONCLUSION Despite bias due to dependence between diffusion-weighted imaging and the final diagnosis, this analysis suggests high sensitivity and specificity for echo-planar diffusion-weighted imaging in the diagnosis of acute cerebral infarction, although negative scans did not rule out an ischemic pathogenesis.