Mass General Home


Principal Investigator Koroshetz, Walter J. , Neurology
Investigation of MRI Diffusion/Perfusion Mismatch in Acute Ischemic Human Stroke
Acute stroke is a major public health problem with 500,000 strokes per year and stroke as the leading cause of adult disability. Standard techniques do not show the region undergoing damage until hours or days after the stroke. The extent and location of stroke is variable and it is difficult to predict outcome without this knowledge that becomes apparent on scans days after stroke. It is therefore difficult to counsel patients regarding whether to accept new but potentially dangerous therapies. It is also difficult to test medications designed to decrease eventual stroke size and thereby decrease disability without knowing before the treatment the patient's expected stroke size. We have been working to develop new MRI techniques for clinical use, which demonstrate definite signature of stroke within hours of onset of symptoms. One called diffusion weighted imaging or DWI shows evidence of ischemic injury in over 90% of stroke patients seen acutely at the MGH and scanned within hours. The other called perfusion imaging delineates the regions of brain, which do not receive normal blood flow in stroke patients. Our work is focused on learning how to predict final infarct location and volume from the initial imaging. In the renewal of our NIH grant we will investigate new means of analyzing the MRI data and will also work to incorporate clinical data into our predictive schema. The most significant changes in the new protocol are : 1) time of follow-up imaging is now at time of discharge from hospital; 2) NIH stroke scale will be performed as part of the study at the time of the initial and follow-up MRI; 3) Stroke Service Stroke Research Registry will be used to identify potential patients for recruitment; 4) Stroke Service Database will be used to obtain relevant clinical data. A very favorable priority score was assigned to the R01 renewal (attached).

Key Personnel