Detecting activations in event-related fMRI using analysis of variance

Abstract

The most common design of a functional MRI (fMRI) experiment is a block design. The use of rapid imaging, however, and carefully designed paradigms makes the separation of cognitive events possible. Such experiments make use of event-related paradigms, in which a task involving several cognitive processes is repeated. In analyzing data from such experiments, existing methods often prove inadequate, because the prediction of the exact shape or timing of the time course is difficult. Here we present an analysis of variance (ANOVA) method for analyzing fMRI data that does not require any assumptions about the shape of the activation time course. Consequently, this method can simultaneously detect brain areas showing a variety of stimulus- locked time courses in the same experiment. The utility of this technique is demonstrated by the analysis of data from two event-related paradigms in which regions of activation are detected that correspond to a variety of distinct neural processes, yielding significantly different temporal signal changes.