Testing for difference between two groups of functional neuroimaging experiments

A research study that investigates a statistical inference method for sets of Talairach coordinates.
Finn Årup Nielsen, Andrew C. N. Chen, Lars Kai Hansen

Abstract

We describe a meta-analytic method that tests for the difference between two groups of functional neuroimaging experiments. We use kernel density estimation in three-dimensional brain space to convert points representing focal brain activations into a voxel-based representation. We find the maximum in the subtraction between two probability densities and compare its value against a resampling distribution obtained by permuting the labels of the two groups. As such it appears as a general method for comparing the local intensity of two non-stationary spatial point processes. The method is applied on data from thermal pain studies where "hot pain" and "cold pain" form the two groups.

Background

Are two sets of Talairach coordinates the same? The Hotelling's T2-test can be used to test this and we have previously applied this test (see Mining Posterior Cingulate). However, if the sets not well described by Gaussian destributions the Hotelling's T2-test is not a good choice.

Here, we instead uses kernel density estimates and resampling-based statistical tests to test whether two different sets of 3D coordinates are the same/different:

Our study = kernel density estimates + resampling + Talairach coordinates

Data comes from the Brede database, and functions from the Brede Neuroinformatics Toolbox are used to analyze the data.

References and Downloads

Researchers

Finn Årup Nielsen is a Post Doc at the Neurobiology Research Unit, Copenhagen University Hospital on a grant for postdoctoral studies from the Villum Kann Rasmussen Foundation.
  Andrew C. N. Chen is research professor at Center of Sensory-Motor Interaction at Aalborg University.
Lars Kai Hansen is Professor at Informatics and Mathematical Modelling, Technical University of Denmark