Dear FSL group,
I have run fsl_motion_outliers in order to perform "scrubbing" as suggested by Power et al. (2012, 2013). However, I have a couple of questions I hope you could answer.
1. After running fsl_motion_outliers -i filtered_func_data.nii.gz -o DVARS --dvars --nomoco --thresh=3, I get a txt file containing 18 columns. Within each column the values are all zeros except for a value of one at the timepoint that is considered to be the outlier. According to the fsl_motion_outliers page this text file should be included in the GML as an additional confound EV so that outlier timepoints are ignored during the statistical inferences. How does FEAT know which timepoints to ignore based on a txt file containing ones and zeros?
2. Power et al. (2012, 2013) suggest 2 metrics for evaluating the extent of micromovements (FD, DVARS). I was wondering which one is considered most robust/accurate? On the fsl_motion_outliers page it is stated that the DVARS is slightly superior as it is based on intensity differences within the realigned timeseries and not the motion correction parameters (as is the case with FD). Hence, its less susceptible to the effects of inaccurate motion correction. Does this render DVARS a slightly more robust/accurate metric?
3. At what point should scrubbing best be performed when using DVARS as a metric? I ran fsl_motion_outliers on the preprocessed filtered_func_data. Is this correct?
Any assistance would be greatly appreciated.
All the best,
Mauricio Delgado
|