Monday Analysis Seminar: Spatial Thresholding Procedures for fMRI Analysis

2017-11-13 15:00 - 2017-11-13 16:00
Faculty of Science Building #3 Room 210
Maya Zhelyazkova (Sofia University, Bulgaria)
Many fMRI experiments have a common objective of identifying active voxels. This is commonly done in single subject experiments for example by performing individual voxelwise tests of the null hypothesis that the observed time course is not significantly related to an assigned reference function. A voxel activation map is then constructed by applying a thresholding rule to the resulting t-statistics. Typically the task related activation is expected to occur in clusters rather than in isolated single voxels. A variety of spatial thresholding techniques have been proposed to reflect this belief including smoothing the raw t-statistics, cluster size inference, and spatial mixture modeling. We examined the performance of these procedures first in terms of Receiver Operating Characteristics (ROC) curves, second we considered the accuracy of these spatial thresholding procedures in estimation of the size of the activation region in terms of bias and variance. Also, we propose and study a Bayesian spatial mixture model for the spatial thresholding problem. We apply this method to a real fMRI study and compare its performance in simulations with other methods proposed for fMRI analysis.

Monday Analysis Seminar