Abstract W P327: Quantitative Localization and Predictive Performance of Intracranial Hemorrhage
Background and Purpose: Current studies of association between intracranial hemorrhage (ICH) localization and stroke severity present conflicting evidence. Using advanced registration techniques on computed tomography (CT) scans, we provide detailed quantification of ICH location and its association with stroke outcomes.
Methods: We analyzed 111 scans from 111 patients enrolled in the MISTIE (Minimally Invasive Surgery plus recombinant-tissue plasminogen activator for Intracerebral Evacuation) trial. We registered CT scans to a CT template, estimated a 3-dimensional map of where ICH engagement occurs, and derived regions predictive of 2 severity scores: the National Institutes of Health Stroke Scale (NIHSS) and Glasgow Coma Scale (GCS) scores. We compared the prediction performance of these regions compared to using location information determined by human readers using the adjusted R2 from linear regression models.
Results: In this sample, ICH is primarily located in deep brain nuclei and lobar white matter, including the superior corona radiata and the thalamus (Figure 1), is related to severity scores. The prediction performance was twice as good for the image-based model compared to the reader-based model (adjusted R2: 0.254 vs. 0.129 for NIHSS, 0.214 vs. 0.069 for GCS).
Conclusions: Measures of ICH location and engagement using advanced CT imaging processing provide finer, more quantitative, and more predictive anatomic information than that provided by expert human readers. These findings suggest that important anatomic relationships may play biologically plausible roles in determining the severity of a particular ICH and can be estimated using this analytic platform.
Author Disclosures: J. Muschelli: None. N. Ullman: None. D. Hanley: Research Grant; Significant; NINDS. C. Crainiceanu: None.
- © 2015 by American Heart Association, Inc.