|
Abstract:
|
Functional magnetic resonance imaging (fMRI) data is often analyzed via the voxel-wise general linear model (GLM) approach. A relatively small amount of mis-modeling, however, can result in severe power loss, and inflate the false positive rate. We propose a simple procedure that can be used to identify possible voxels or regions of the brain where model misfit may be present. The key idea is that if there is model misfit, residuals will be systematically larger in mismodeled segments of the time series. By looking at the weighted sum of consecutive residuals using a moving window, our method can pick out regions of a residual time series in which the residuals are consistently larger than expected by chance, while ignoring spurious large residuals. The effectiveness of the procedure is examined using a simulation study and by applying to an fMRI data set.
|
