Abstract TP91: Automated Cerebral Hemorrhage Volume Calculation And Stability Detection Using RAPID Software

Abstract

Introduction: The measurement of intraparenchymal hemorrhage (IPH) volume is important for management, particularly in evaluating expansion on subsequent imaging. However manual volumetric analysis is time-consuming, especially in busy hospital settings. We aimed to use automated RAPID hyperdensity software to accurately measure IPH volume across repeated imaging. Methods: We identified primary IPH cases, without evidence of IVH, SDH, or SAH, from two randomized ICH trials where enrollment was not based on IPH volume. Scans were excluded if (1) there was CT acquisition artifact, (2) measured IPH volume was < 0.4ml, or (3) patient underwent a neurosurgical procedure. Repeat imaging, conducted within 12h, was included in a secondary stability analysis. Manual IPH measurements were conducted by one neuroimaging expert using MIPAV software and compared to the performance of the RAPID hyperdensity module. Results: 149 scans were included with median IPH volume manually measured at 13.3ml (IQR 3.2-27.5) compared to RAPID detection of 18.0ml (IQR 5.1-36.8). The two modalities were highly correlated (Figure 1, r = 0.948, p < 0.001) with a median absolute difference of 3.4ml (IQR 0.2-9.4) and standard error of 0.655. Repeat imaging was available in 65 patients, with a median absolute difference of 0ml (IQR: -2.6-1.5) compared to RAPID detection at 0.1ml (IQR: -2.3-2.7). These absolute differences were also highly correlated (r = 0.878, p < 0.001), with the ability of RAPID to detect a 5ml IPH expansion with a Sensitivity of 98.1% and Specificity 72.7%. Processing time for the software was < 3 minutes. Conclusion: In our pilot data, the RAPID hyperdensity module has high reliability in its ability to quickly automatically detect IPH volume and high sensitivity to detect expansion on subsequent imaging.