 |
 |
Hi, We don't have a command to do what you are suggesting. The more standard way is to just regress out the spikes which then replaces the values with the average over all the timepoints, not just the neighbouring timepoints, which can be done with fsl_regfilt. If you want to just do the average of the neighbours then you'll need to write your own script. All the best, Mark > On 23 May 2017, at 02:16, Rita Elena Loiotile <[log in to unmask]> wrote: > > Hi FSL experts, > > Is there an easy way to replace spikes with interpolated values from neighboring timepoints? I'm wondering if FSL has an easy hidden command, before I write my own script for this. > > I'm using fsl_motion_outliers --dvars --fdrms to find outlier TRs. Normally, I would add these to my first-level GLM with a regressor for each outlier (1 at the TR and 0 elsewhere). However, I'm currently doing a resting state analyses. So, I'd be regressing out the motion effects and using the residual of the GLM for my subsequent analysis. Because there are no task regressors to pick up any of the variance, this would result in each outlier regressor completely modeling the TR it represents (where it is set to 1). Therefore, the residual-- i.e., my new "clean" functional-- would rewrite each spike as the implicit baseline. This seems to lose information. > > I think a better approach would be to set the outlier TR to an average of the neighboring non-spike TRs. However, one would just have to control for cases in which outliers are successive TRs. > > Please let me know if there is a good command to use or if there are better approaches. > > Thank you kindly, > Rita > >