New reasons for early use of MRI in stroke.

Abstract

Neurologists are familiar with MRI as T1-, T2-, and proton density–weighted images, and, recently, magnetic resonance angiography (MRA). Here, we draw attention to newer MRI methods—diffusion-weighted imaging (DWI) and perfusion imaging (PI)—that are unique among all imaging methods in their sensitivity to pathophysiologic changes underlying acute stroke. Documentation in clinical studies is now sufficiently compelling to warrant careful consideration of DWI and PI by all physicians who care for patients with acute stroke syndrome. Conventional MRI methods usually detect ischemic lesions after a few hours, well before clear changes appear on x-ray CT, and with much greater anatomic resolution. The increased sensitivity is particularly obvious in the posterior fossa, where imaging artifacts can mask acute CT changes. However, the advantages of MRI for stroke diagnosis have not yet been documented by clinical outcome studies comparing it directly with CT, which led Powers and Zivin to express well-reasoned reservations about whether MRI is so clearly superior as to justify the costs in money and procedural adjustment that its rapid general adoption would require.1 We believe that recent literature on DWI and PI has tipped the balance definitively in favor of accelerated implementation of MRI as the principal imaging modality for initial evaluation of acute stroke syndrome. This issue of Neurology contains two articles2,3 and a review4 providing new support for that position. Their independent submission to a single journal within a few weeks of each other accurately reflects the pace of development in the field. Diffusion-weighted imaging. Diffusion in liquids occurs by Brownian motion, the random process of molecular displacements caused by thermal agitation. DWI measures the diffusive motion of water molecules as an apparent diffusion coefficient (ADC), which varies with location in biologic tissue. The anatomic differences can be rendered as an image by standard …