An optimized method for estimating intracranial volume from magnetic resonance images

Abstract

The accuracy and efficiency of protocols to measure intracranial volume (ICV) from volumetric magnetic resonance imaging (MRI) studies has not been formally analyzed. The ICV of 30 control participants was obtained by tracing every slice of a MRI data set on which the cranial cavity appeared, and compared with estimated ICVs calculated by progressively selecting one of every x slices (i.e., "1-in-x") as a sampling strategy. The reliability and precision of each sampling strategy was then determined. There was virtually no reduction in reliability at the 1-in-10 sampling strategy, with a reliability exceeding 0.999. ICV can be confidently traced using a 1-in-10 sampling strategy, which should result in significant time savings.