Functional Topography: Multidimensional Scaling and Functional Connectivity in the Brain

Abstract

In neuroimaging, functional mapping usually implies mapping function into an anatomical space, for example, using statistical parametric mapping to identify activation foci, or the characterization of distributed changes with spatial modes (eigenimages or principal components) (Friston et al., 1993a). This article is about a complementary approach, namely, mapping anatomy into a functional space. We describe a simple variant of multidimensional scaling (principal coordinates analysis; Gower, 1966) that uses functional connectivity as its metric. The scaling transformation maps anatomy into a functional space. The topography, or proximity relationships, in this space embody the functional connectivity among brain regions. The higher the functional connectivity, the closer the regions. Functional connectivity is defined here as the correlation between remote neurophysiological events. The technique represents a descriptive characterization of anatomically distributed changes in the brain that reveals the structure of corticocortical interactions in terms of functional correlations. To illustrate the approach we have analyzed data from normal subjects and schizophrenic patients obtained with PET during the performance of word generation tasks. In particular, we focus on prefrontotemporal integration in normal subjects and show that, in schizophrenia, the left temporal regions and prefrontal cortex evidence abnormal functional connectivity.