Testing different noise models output from fmriprep - unclear results with ICA-AROMA

[ChrisGorgolewski]

@Michael_Hallquist

Briefly, the main difference is whether voxelwise BOLD data are regressed on noise components alone (this is what you’re doing with 3dTproject, I believe) versus regressed on all timecourses from the ICA decomposition (the non-aggressive fsl_regfilt approach). In the fsl_regfilt case, the partial regression coefficients reflect the association of a noise component with the data in the presence of both noise and ‘good’ components. The signal to be removed is computed by multiplying the partial coefficients for the noise components by the timecourses for those components, summing the results, then subtracting this from the BOLD timecourse.

Very nice explanation of what is going on. This approach divides the variance shared between noise and signal into the the noise and signal coefficients. In other words some, but not all and not none of the noise/signal shared variance will be removed from the data. How do you think this approach would compare to orthogonalizing the noise components against signal components prior to regression (in other words making sure none of the noise/signal shared variance is removed from the data)?