Rapid Quantification of Stroke Infarct from Diffusion Weighted Images

Sumit K. Nath¹, Rakesh Mullick¹, Dattesh Shanbhag¹, Uday Patil²

A vital step in the evaluation of acute and chronic stroke involves volume quantification of the infarct as determined from diffusion-weighted images (DWI). The routine assessment of infarct size is by visual review of the approximate size reported along with the location of the defect with reference to the brain anatomy. Assessment of the infarct volume is critical to management decisions in acute stroke, especially for thrombolytic therapy, and calls for quantitative methods. There are multiple efforts to develop post-processing methods to segment and automatically measure the infarct from DWI images. Given T2 shine-through artifacts from pre-existing old infarcts in the DWI images make this task very challenging. In order to offer a robust workflow for this evaluation we present a hybrid (visualization and segmentation) method to address this. The core concept is to integrate the presentation of maximum intensity projections (MIP) of the DWI data to the user to select suggestive infarct regions and use an underlying level-set based (top-down) image segmentation approach for rapid quantification of suspect regions