Randomized parcellation based inference

Abstract

29 Accepted 5 November 2013 30 Available online xxxx 3213 34 Keywords: 35 Group analysis 36 Parcellation 37 Reproducibility 38 Multiple comparisons 39 Permutations 54 53 55 Introduction 56 Analysis of brain images acquired on a group of subjects makes it 57 possible to draw inferences on regionally-specific anatomical properties ? Correspondingauthorsat:CEA,DSV,I2BM,Neurospinba?t145,91191Gif-Sur-Yvette, France. E-mail addresses: benoit.da_mota@inria.fr (B. Da Mota), bertrand.thirion@inria.fr (B. Thirion). 1 These authors contributed equally to this work. 2 URL: http://www.imagen-europe.com 1053-8119/$ ? see front matter ? 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.neuroimage.2013.11.012 ?2Q46 ?2Q57 ?article info abstract Neuroimaging group analyses are used to relate inter-subject signal differences observed in brain imaging with 40 behavioral or genetic variables and to assess risks factors of brain diseases. The lack of stability and of sensitivity 41 of current voxel-based analysis schemes may however lead to non-reproducible results. We introduce a new 42 approach to overcome the limitations of standard methods, in which active voxels are detected according to 43 a consensus on several random parcellations of the brain images, while a permutation test controls the false 44 positive risk. Both on synthetic and real data, this approach shows higher sensitivity, better accuracy and higher 45 reproducibilitythanstate-of-the-artmethods.Inaneuroimaging?geneticapplication,wefindthatitsucceedsin 46 detecting a significant association between a genetic variant next to the COMT gene and the BOLD signal in the 47 left thalamus for a functional Magnetic Resonance Imaging contrast associated with incorrect responses of the 48 subjects from a Stop Signal Task protocol.