Abstract #0466

An Automated Tool for Prediction of Secondary Hemorrhage in Stroke.

Matus Straka¹, Bruce C. Campbell², Maarten G. Lansberg³, Greg Zaharchuk¹, Michael Mlynash³, Stephanie M. Kemp³, Demi Thai³, Gregory W. Albers³, Roland Bammer¹

¹Radiology, Stanford University, Stanford, CA, USA; ²Neurology, Royal Melbourne Hospital, Melbourne, Australia; ³Stanford Stroke Center, Stanford University, Stanford, CA, USA

Possible occurrence of secondary hemorrhage is an important consideration in stroke treatment. Based on previous research, we present a tool that can fully-automatically and in short time provide hemorrhage prediction maps based on standard DWI and PWI images. Acquired PWI data are automatically postprocessed to obtain maps cerebral blood volume (CBV) and coregistered with the DWI. The CBV maps are then mirrored about the A/P axis to obtain estimates of contralateral CBV. Hemorrhage is ultimately predicted in regions that manifest diffusion lesions together with very low CBV. An initial test on 19 stroke cases showed 83% specificity and 86% specificity.