Stimulus Response Detection in fMRI Using Temporal Clustering Analysis,Digital Filters and a Two-Dimensional Neighbourhood Test

Abstract

A model-free fMRI (functional magnetic resonance imaging) analysis method is proposed to detect gradually evolving physiological or pharmacological responses to a single-event or non-repetitive multi-event experiment without any a prior information on the paradigm. The method is developed from the pre-existing temporal clustering analysis method with the applications of digital filters, temporal averaging and a two-dimensional constraint test and makes use of the data from the whole brain instead of a selected single slice. We have applied the technique to the fMRI data from two different kinds of experiments and the results demonstrate that our method is able to identify brain regions and time bins during which the BOLD (blood oxygenation level dependent) MRI time series display responses to sensory stimuli.