Abstract TP48: Prediction of Neurologic Outcome after Ischemic Stroke with Volumetric Measurements on CT and MRI
Background: CT perfusion (CTP) and MR diffusion-weighted imaging (DWI) interpretation is typically qualitative. Manual quantitative volumetric measurement is too labor-intensive to be practical in the setting of acute ischemic stroke (AIS).
Hypothesis: Compared to a manual technique, automated volumetric analysis of CTP and DWI in AIS patients will be a strong predictor of neurologic outcome at follow-up.
Methods: We reviewed 123 inpatients from 2010-2014 with AIS, admission CTP, and clinical follow-up. Using the automated Olea Sphere software, we calculated CTP maps of cerebral blood volume (CBV) and mean transit time (MTT) using oSVD and Bayesian deconvolution. We used the OsiriX software to manually draw regions-of-interest corresponding to CBV and time-to-drain lesion volume on CTP, and core volume on DWI (in 74%). A multivariable ordinal logistic regression model was fitted to the outcome of modified Rankin Scale (mRS) at follow-up with covariates that had a p<0.2 and the primary predictor variables of stroke core, hypoperfused area, or mismatch volume. Mismatch was defined as CBV subtracted from MTT or time-to-drain.
Results: Patients had mean (± SD) age of 64 ± 16 years; 54% were male; 91% of strokes were in the anterior circulation; mean NIHSS was 13 ± 8; mRS was 2.6 ± 2.1 at 82 ± 46 days to follow-up. Logistic regression found that stroke core was predictive of outcome with both automatic vs manual processes; the automated processes were predictive of outcome for the hypoperfused area measurements; and neither method of measuring mismatch was predictive for outcome (Table 1).
Conclusion: Our data suggests that automated volumetric analysis of the stroke core and hypoperfused area, but not mismatch, provides accurate neurologic prognosis. Compared to manual volumetric measurements, automated software is more time efficient and may have superior predictive ability.
Author Disclosures: A. de Havenon: None. B. Donleavy: None. H. Wang: None. L. Chung: None. J. Majersik: Research Grant; Significant; NIH/NINDS U10 NS086606.
- © 2016 by American Heart Association, Inc.