Statistical parametric mapping

Abstract

In summary, statistical parametric mapping is a flexible and powerful way of testing data acquired using functional imaging techniques to establish patterns of brain activation. Although appearing complicated, it is in fact based on well-established statistical techniques familiar throughout clinical science. For spatially distributed data that form a continuous time series, such as functional neuroimaging data, there are a number of important modifications to allow parametric statistics to be used, but their details are generally not important for a critical appreciation of published data. The most important issue faced by the user of these techniques is the multiple comparisons problem, a consequence of the ability of modern MRI scanners to acquire vast amounts of data at high spatial resolution. However, the power of current analysis techniques can encompass this, while remaining flexible enough to accommodate the next generation of imaging techniques. © 2004 Blackwell Publishing Ltd.