Background and purpose: Reliable quantification of both intracerebral hemorrhage and intraventricular hemorrhage (IVH) volume is important for hemostatic trials. We evaluated the reliability of computer-assisted planimetric volume measurements of IVH.
Methods: Computer-assisted planimetry was used to quantify IVH volume. Five raters measured IVH volumes, total (intracerebral hemorrhage+IVH) volumes, and Graeb scores from 20 randomly selected computed tomography scans twice. Estimates of interrater and intrarater reliability were calculated and expressed as an intrarater correlation coefficient and an absolute minimum detectable difference.
Results: Planimetric IVH volume analysis had excellent intra- and interrater agreement (intrarater correlation coefficient, 0.96 and 0.92, respectively), which was superior to the Graeb score (intrarater correlation coefficient, 0.88 and 0.83). Minimum detectable differences for intra- and interrater volumes were 12.1 mL and 17.3 mL, and were dependent on the total size of the hematoma; hematomas smaller than the median 43.8 mL had lower minimum detectable differences, whereas those larger than the median had higher minimum detectable differences. Planimetric total hemorrhage volume analysis had the best intra- and interrater agreement (intrarater correlation coefficient, 0.99 and 0.97, respectively).
Conclusions: Computer-assisted planimetric techniques provide a reliable measurement of ventricular hematoma volume, but are susceptible to higher absolute error when assessing larger hematomas.