SAMPLING INTENSITY WITH FIXED PRECISION WHEN ESTIMATING VOLUME OF HUMAN BRAIN COMPARTMENTS

Abstract

Cavalieri sampling and point counting are frequently applied in combination with magnetic resonance (MR) imaging to estimate the volume of human brain compartments. Current practice involves arbitrarily choosing the number of sections and sampling intensity within each section, and subsequently applying error prediction formulae to estimate the precision. The aim of this study is to derive a reference table for researchers who are interested in estimating the volume of brain regions, namely grey matter, white matter, and their union, to a given precision. In particular, this table, which is based on subsampling of a large brain data set obtained from coronal MR images, offers a recommendation for the minimum number of sections and mean number of points per section that are required to achieve a pre-defined coefficient of error of the volume estimator. Further analysis onMR brain data from a second human brain shows that the sampling intensity recommended is appropriate.