Hello,
We use a standardized pipeline in spm that uses a combination of ART and ArtRepair to address subject motion and noise. We detect outlier volumes in ART using a 1mm threshold for rsFC data, and 4mm for task data, and then use ArtRepair to interpolate the motion outlier volumes. For task data, we also use art_despike to despike the data at 4% of the global mean. Our task data is either long block (eg 40 sec) or fast event-related. We leave the rsFC data alone and use despiking in conn. Then, we load the art_regression_outliers_and_movement.mat file into conn as a 1st level confound.
Do you see any issue with this approach in terms of denoising?
Many thanks in advance,
Patrick
We use a standardized pipeline in spm that uses a combination of ART and ArtRepair to address subject motion and noise. We detect outlier volumes in ART using a 1mm threshold for rsFC data, and 4mm for task data, and then use ArtRepair to interpolate the motion outlier volumes. For task data, we also use art_despike to despike the data at 4% of the global mean. Our task data is either long block (eg 40 sec) or fast event-related. We leave the rsFC data alone and use despiking in conn. Then, we load the art_regression_outliers_and_movement.mat file into conn as a 1st level confound.
Do you see any issue with this approach in terms of denoising?
Many thanks in advance,
Patrick