Meeting Banner
Abstract #3651

A Permutation Test Statistical Analysis of Learning Induced DTI Changes

Ido Tavor1, Shlomi Lifshits2, Shir Hofstetter1, Yaniv Assaf1

1Department of Neurobiology, Tel Aviv University, Tel Aviv, Israel; 2Department of Statistics and Operations Research, Tel Aviv University, Tel Aviv, Israel

Diffusion MRI is widely used in recent years to investigate the relation between brain structure and cognitive abilities, by scanning subjects before and after a learning task. Methodologies applied in such studies require spatial normalization procedure and a statistical comparison in the group level. We propose an approach to execute a permutation test for performing statistical comparison between 2 DTI scans in the single subject level. The main principle is to examine in which voxels the change in DTI parameters is bigger than the inherent noise. We demonstrate that structural brain changes are detectable using such a method.

Abstract PDF Presentation Video

Keywords

abilities able acquisition addition anatomical another anterior apart applied architecture bigger bilaterally biological bootstrap brain clusters cognitive cohorts computer cortex dataset decrease described detect determining diffusion diffusivity distribution enables equal equals examine execute expressed fact feasible fiber game greater hemispheres hypothesis immediately impossible included induced induces inherent insular investigate isotropic kind learning maps memory methodologies methodology middle native navigation need neurobiology neuron noise normalization null operations orientation others part participating permutation permutations plasticity post posterior practice previous principle procedure proportion propose protocol racing randomly real rearranged recent reduction reductions rejected relation represented require requires reshuffled resolution reveal sampled scanned scanning separation shuffled space spares spatial statistic statistical statistics structural structure studies subject subjects task tensor training uncertainty visualizing white widely years zero