Does Parametric fMRI Analysis with SPM Yield Valid Results? – An Empirical Study of 1484 Rest Datasets

Josephson Camilla

doi:10.3389/conf.fnhum.2012.208.00054

Abstract

Does Parametric fMRI Analysis with SPM Yield Valid Results? – An Empirical Study of 1484 Rest Datasets

Highlights

It has been debated for a long time if the assumptions that are required for standard parametric approaches really are appropriate for functional magnetic resonance imaging data
A large number of rest datasets have been analyzed to show that temporal correlations in resting state functional magnetic resonance imaging (fMRI) timeseries may show a more complicated structure, than previously assumed in conventional statistical models
Our work suggests a need to improve, or extend, the models of temporal correlations or stationary dependencies in single subject fMRI timeseries

Summary

Introduction

It has been debated for a long time if the assumptions that are required for standard parametric approaches really are appropriate for functional magnetic resonance imaging (fMRI) data. It has been debated how the problem of multiple testing should be solved. The recent advances in computer science, e.g. graphics processing units (GPUs), make it possible to perform conventional fMRI analysis in a few seconds (Eklund et al, 2011a, 2012) This permits using thousands of studies in the evaluation of analysis and inference procedures in fMRI data analysis, which was not previously possible.

Objectives

Methods

Results

Discussion

Conclusion