Abstract
A fundamental goal in “brain mapping” with functional Magnetic Resonance Imaging (fMRI) is localising the parts of the brain activated by a task. The standard tool for making this inference has been Random Field Theory (RFT), a collection of results for Gaussian Processes of the null statistic image (implemented in the two most widely used packages, SPM & FSL). RFT provides inference on individual voxels (voxel-wise) and sets of contiguous suprathreshold voxels (cluster-wise) while controlling the familywise error rate, the chance of one or more false positives over the brain. I will discuss how RFT methods have been used for the past 25 years, show some small-scale evaluations that pointed to problems with RFT when the degrees-of-freedom are low. I will then show results from a recent study based on the wealth of (1000’s of) publicly available resting-state fMRI datasets; these massive evaluations show that, even with n=20 or 40 subjects, RFT suffers from slightly conservative voxel-wise inferences and catastrophically liberal cluster-wise inferences. I will discuss the reasons for these failures of RFT and practical solutions going forward.
Bio
Thomas Nichols is a Wellcome Trust Senior Research Fellow in Basic Biomedical Science, a Professor and the Head of Neuroimaging Statistics at the Institute for Digital Healthcare, holding a joint position between Warwick Manufacturing Group & the Department of Statistics. Before joining the University of Warwick he was the Director of Modelling & Genetics at the GlaxoSmithKline Clinical Imaging Centre at Hammersmith Hospital in London, where he worked on statistical methods for fMRI in the context of clinical trials, and integrating genetic data into brain image analyses. Before coming to the UK he was an Associate Professor of Biostatistics at the University of Michigan, and in 2001 received his Ph.D. in statistics from Carnegie Mellon University where he also trained in cognitive neuroscience. He has been active in the field of functional neuroimaging since 1992, when he worked at the University of Pittsburgh’s PET Center as a programmer and statistician. Dr. Nichols’ research focuses on modelling and inference of neuroimaging data, including PET, fMRI & M/EEG.