Stimulus response detection in FMRI using digital filters, temporal clustering analysis and a three-dimensional selection criterion

A data-driven fMRI (functional magnetic resonance imaging) analysis method is proposed to detect responses when no clear a priori information is available to predict response onset or duration in a single-event or non-repetitive multi-event experiment. It is developed from pre-existing temporal clustering analysis methods with the applications of digital filters and temporal averaging. A novel three-dimensional selection criterion is employed to make use of the contribution from the signal of neighbouring voxels both in the same and adjoining slices on transverse section. The method has been applied to the whole-brain fMRI data from three different kinds of sensory experiments and the results demonstrate that the method is able to identify the time bins, during which the stimuli were applied and the voxels contributed to these identifications were in the brain regions, which were typically found activated in these kinds of sensory experiments.