Meeting Banner
Abstract #4769

Automatic Two Stage Classification and Segmentation of Ischemic Stroke Lesions in Diffusion-Weighted MRI

Pieter C. Vos 1,2 , Steven Mocking 2 , Priya Garg 2 , Aurauma Chutinet 3 , William A. Copen 4 , Max A. Viergever 1 , and Ona Wu 2

1 Radiology, Image Sciences Institute, Utrecht, Utrecht, Netherlands, 2 Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, Massachusetts, United States, 3 Department of Neurorology, MGH, Massachusetts, United States, 4 Department of Radiology, MGH, Massachusetts, United States

DWI is a reliable and routinely-used modality in the acute setting of ischemic stroke. Automated approaches for outlining the DWI lesion have the potential to assist in the rapid assessment of lesion volumetry in the acute setting of stroke. Our results demonstrate that the segmentation performance of pixel classification approach can be significantly improved with regional analysis, i.e. using a supervised classifier that discriminates false detected regions from true lesion regions in a two-stage classification approach.

This abstract and the presentation materials are available to members only; a login is required.

Join Here