brundavanam.divya@spes.uniud.it

DEAMS - UNITS

(Supevisor Nicola Torelli)

Application of spatial statistics for inference on imaging data

My current research explores the performance of trend surface modeling as an alternative approach for inference in order to achieve better Type I error control compared to Random Field Theory (RFT) - based methods. We model spatial correlation using a set of biologically informed basis functions, taking advantage of a reduced penalty for multiple comparisons since the number of basis functions required (typically a few hundred) is much lower than the number of voxels modeled in a mass univariate analysis.