Semi-automatic volumetric assessment of perihemorrhagic edema with computed tomography

Background and purpose: Magnetic resonance imaging (MRI) shows perihemorrhagic edema (PHE) after intracerebral hemorrhage (ICH) with high contrast, but the procedure is often difficult or not available for clinical use. The aim of the present study was to establish and validate an observer independent method for quantification of PHE on computed tomography (CT) by comparing with simultaneously performed MRI. Methods: Patients with spontaneous supratentorial ICH were included. Twenty-two patients received coregistered MRI and CT on day 1, and 27 patients on day 5 after admission. Volumes for PHE and ICH were measured (i) manually on CT, (ii) manually on MRI (fluid-attenuated inversion recovery sequence), and (iii) threshold based on CT. To identify optimal threshold values (Hounsfield units) for best correlation of CT with MRI, upper and lower thresholds were adjusted gradually until the PHE volume on CT best fitted the PHE volume on MRI. The established threshold range was prospectively validated in another 15 patients. Results: A threshold range 5-33 Hounsfield units (HU) resulted in best correlation both on days 1 and 5. Using these thresholds in the validation group, PHE volumes on CT and MRI were highly comparable (31 +/- 26 ml vs. 30 +/- 27 ml) with good correlation (R(2) = 0.96, P < 0.01) and high inter-(0.96) and intraobserver (0.96) reliability. Manually traced PHE on CT was significantly larger (37.3 +/- 37 ml vs. 30 +/- 27 ml, P < 0.01) with worse inter-(0.89) and intraobserver (0.90) reliability. Conclusions: Threshold-based CT volumetry of PHE with a threshold range 5-33 HU is a reliable and observer independent method for quantification of PHE after spontaneous ICH.