Comparison of whole-brain to region-based fMRI analyses

Conventional fMRI analyses primarily utilize a general linear model composed of a single reference vector for the whole-brain. A single reference vector assumes the hemodynamic response is homogenous between different individuals and regions of interest. This study sought to determine whether a regional-based analysis that accounts for individual and regional differences would yield more sensitive results in terms spatial extent. Fifteen subjects without neurological dysfunction were scanned in a 3T MRI using an experimental block design of visual fixation compared to randomized vergence eye movements. Two temporal independent component analyses were conducted on individual subject data sets. The first used the data set composed of all the voxels within the brain while the second used the data from a masked region to isolate the voxels within the frontal eye fields, a critical cortical region needed to evoke vergence eye movements. A hierarchical temporal independent component analysis was used to compute group-level results. Spatial maps were compared. The regional based analysis resulted in fewer clusters which had greater spatial extent compared to the whole-brain analysis which had more dispersion within the functional activity. This technique has the potential to study individual regions while accounting for the variability of the natural heterogeneous nature of the hemodynamic response found in fMRI data.