Abstract

INTRODUCTION

Neurologists nowadays regularly encounter statistical parametric mapping in journal articles that report the results of a functional neuroimaging study. The number of such studies has risen dramatically in the last decade, and an understanding of how typical functional MRI (fMRI) experiments are analysed will help the clinician critically reading the literature. Functional imaging studies are typically undertaken to compare brain responses either from a single population in two different conditions (for example, healthy volunteers in the presence vs. the absence of a flickering visual stimulus), or from two populations (for example, neurological patients performing a task vs. control subjects performing the same task).

THE BASIC IDEA

Brain responses are recorded using an imaging technique such as fMRI, which allows the repeat-ed measurement of brain activity at hundreds of thousands of points, or voxels, throughout the brain (Fig. 1). Each voxel represents physiological responses from a small anatomical portion of