Meeting Banner
Abstract #3000

Segmentation-Based Quantification of Brain SWI for Predicting the Stroke Evolution

Ping-Huei Tsai1, 2, Chia-Yuen Chen2, Chin-I Chen3, Fong Y. Tsai1, Hsiao-Wen Chung4, Wing P. Chan2

1Imaging Research Center, Taipei Medical University, Taipei, Taiwan; 2Department of Radiology, Wan Fang Hospital, Taipei Medical University, Taipei, Taiwan; 3Department of Neurology, Wan Fang Hospital, Taipei Medical University, Taipei, Taiwan; 4Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan

The aim of this study is using an auto segmentation method based on data clustering to investigate the symmetry of brain SWI in normal subjects and facilitate the quantification of the asymmetric distribution of the deoxygenated vessels in patients with acute ischemic stroke for a better prediction of the evolution. Our preliminary finding demonstrates that the proposed method provides objective information for evaluation of the patients, and may have a potential to contribute to determining the penumbra and predicting of the stroke prognosis, as well as the following treatment.

Abstract PDF Presentation Video

Keywords

abnormal acquisition activator acute added addition adjusts although among around aspect asymmetric auto availability avoid basal beneficial blood brain calculating cell centroids channel chin chosen clearly clinical closed clustering clusters coil comparable considering constraint contribute controlled controversial correlation critical days degree dependent describing detecting determination determining discrepancy distribution electrical emphasize enrolled evaluation every evolution extraction facilitate fang fast feasibility feature fifteen finding flair fussiness ganglia good gradient growth head hemispheres hemorrhagic hospital hours index indicates infarct injury intensity intra intravenous involved lead mainly matrix medical memberships minutes mismatch moreover neighbors neurology normalized objective original outcome oxygen patients penumbra percentages perfusion ping position possibility potential predict predicting preliminary prognosis progressive prominent quantification radiology reduction reflect reflects relationship residual resolution respectively ruled segmentation segmented severe sharp side slice solutions sort spaces spatial stoke stroke subjective subjects suggest suitable supine susceptibility suspected symmetry symptom system tissue treatment twelve twenty vessel vessels volunteers whose wing