Combining Functional MRI Data on Multiple Subjects

Abstract

Worsley et al. (2002) propose a practical approach to multiple-subjectfunctional MRI data analyses which uses the EM algorithm to estimate thebetween-subject variance component at each voxel. The main result ofthis article is a demonstration that the much more efficientNewton-Raphson algorithm can be reliably used for these calculations.This result follows from an extension of a simple algorithm proposed byMandel and Paule (1970) for the one-way unbalanced ANOVA model, twovariants of which have been shown to be equivalent to modified ML andREML, in which the ``modification{''} is that the within-subjectvariances as treated as known.