Yield of diffusion-weighted MRI for detection of potentially relevant findings in stroke patients NEUROLOGY Albers, G. W., Lansberg, M. G., Norbash, A. M., Tong, D. C., O'Brien, M. W., Woolfenden, A. R., Marks, M. P., Moseley, M. E. 2000; 54 (8): 1562-1567


To determine whether diffusion-weighted imaging (DWI) could identify potentially clinically relevant findings in patients presenting more than 6 hours after stroke onset when compared with conventional MRI.MRI with both conventional (T2 and proton density images) and echoplanar imaging (DWI and apparent diffusion coefficient maps) was performed 6 to 48 hours after symptom onset (mean, 27 hours) in 40 consecutive patients with acute stroke. All acute lesions were identified first on conventional images, then on DWI, by a neuroradiologist who was provided with the suspected lesion location, based on a neurologist's examination before imaging. Abnormalities were rated as potentially clinically relevant if they were detected only on DWI and 1) confirmed the acute symptomatic lesion to be in a different vascular territory than suspected clinically, 2) revealed multiple lesions in different vascular territories suggestive of a proximal source of embolism, or 3) clarified that a lesion, thought to be acute on conventional imaging, was not acute.The initial clinical impression of lesion localization was incorrect in 12 patients (30%). Clinically significant findings were detected by DWI alone in 19 patients (48%). DWI demonstrated the symptomatic lesion in a different vascular territory than suspected clinically or by conventional MRI in 7 patients (18%) and showed acute lesions in multiple vascular distributions in 5 patients (13%). In 8 patients (20%), DWI clarified that lesions thought to be acute on conventional MRI were actually old.In patients imaged 6 to 48 hours after stroke onset, DWI frequently provided potentially clinically relevant findings that were not apparent on conventional MRI.

View details for Web of Science ID 000086642000007

View details for PubMedID 10762494