Dear Martina
Thanks for posting this great reference. Am I correct that figure 3 shows, that physiological modelling is needed even if FAST18 is used? It would have been great to see if Volterra expanded motion regressors (24 parameters) can reduce/remove the peak at low frequencies still present in the “No Model” data after HP filtering and RETROICOR modelling, or if we need to do both RETROICOR and FAST when TR is sufficiently low. I think the reason for FAST not being able to remove the cardiac peak in figure 3 TR=0,35s is that the cardiac frequency is not constant through time, and that there is considerable spatial variation both in phase and magnitude. At lower TR it makes the noise look like it is white (because it is aliased into a range of frequencies). I also found it a bit puzzling that the area under the yellow curve sometimes look smaller than or equal to those of the the red and green curve, but I guess you might have used normalised residuals, as those outputted from the SPM model estimation, correct?
Best
Torben
> Den 22. feb. 2023 kl. 11.11 skrev Martina Callaghan <[log in to unmask]>:
>
> Dear Brandon
>
> Hopefully you would find this article, and references therin, useful: https://onlinelibrary.wiley.com/doi/full/10.1002/hbm.24218
>
> Best wishes
>
> Martina
|