Abstract WP133: Assessing Aneurysmal Rupture Outcomes Using Automated Quantification Of Hemorrhage Volume

Abstract

Introduction: Hemorrhage volume affects the risk of developing complications after aneurysmal subarachnoid hemorrhage (aSAH). This volume is quantified subjectively through the Fisher Scale. We developed an automated method to quantify intracranial hemorrhage and determine the risk of complications after aSAH. Methods: CT scans of patients with aSAH were analyzed. An image-processing algorithm was designed in MATLAB to quantify aSAH volume. The automated algorithm was developed with region-growing and 3D k-means clustering. Two independent raters manually measured hemorrhage volumes using PACS and correlated these measurements with the algorithm measurements in ten patients. Vasospasm and hydrocephalus were identified through chart and imaging review. Hydrocephalus was adjudicated as ventricular enlargement in the setting of aSAH. ROC analysis was performed to evaluate the association between blood volume and the incidence of vasospasm, hydrocephalus and need for a permanent ventriculoperitoneal (VP) shunt. Results: One hundred and forty-one patients were included in the analysis. The intraclass coefficient correlation (ICC) for inter-rater agreement in PACS was 0.988 (p < 0.001). The ICC for comparing automated versus manual volumes was 0.966 (p < 0.001). Bland-Altman analysis reported a mean difference of -3.14 mL between automated and manual measurements. Hemorrhage volume discriminated for the development of vasospasm (AUC = 0.692), hydrocephalus (AUC = 0.775), and need of permanent VP shunt (AUC = 0.669). Conclusion: Vasospasm and hydrocephalus were associated with higher volume of aSAH. Patients with higher volume of aSAH at presentation were more likely to require a permanent VP shunt. An objective method to automatically measure aSAH volume can aid in assessing risk of complications among patients with aSAH.