Hi Alfonso,
I just got a reply from the journal where I resubmitted the results you advised me on.
It looks like one of the reviewer still has issues with the 3 mm motion threshold when it comes to subject exclusion. We have used it before but he/she still has doubts (please see below).
What do you suggest? Is there a better answer I can give other than saying that this approach has been used in previous studies?
If I follow his advice, I might lose the results...any alternative approach?
Here's his comment:
"Firstly, the issue of head motion. It is good that the authors checked and found no group differences in head motion. However, I find the motivation for a threshold of 3 mm a bit strange. The main issue to me is that head movement near or equal to the size of voxels means that at those points the voxels measured something else. Numerical correction for movement does not 'correct' for that voxel 'seeing' something else at that moment of movement. In this regard, movement of 50% of voxel size could already be considered as problematic. Thus, the issue is more conceptual than numerical: functional connectivity measures could be driven by an artifactual mixture of sources from different locations. That other studies may or may not have used such a lenient threshold is irrelevant.
I suggest that the authors investigate how many volumes or datasets contain head motion at or above threshold of 1 or 1.5 mm. Preferably, the authors also verify that the pattern of results do not change with a stricter threshold."
Thank you in advance
Francesco
I just got a reply from the journal where I resubmitted the results you advised me on.
It looks like one of the reviewer still has issues with the 3 mm motion threshold when it comes to subject exclusion. We have used it before but he/she still has doubts (please see below).
What do you suggest? Is there a better answer I can give other than saying that this approach has been used in previous studies?
If I follow his advice, I might lose the results...any alternative approach?
Here's his comment:
"Firstly, the issue of head motion. It is good that the authors checked and found no group differences in head motion. However, I find the motivation for a threshold of 3 mm a bit strange. The main issue to me is that head movement near or equal to the size of voxels means that at those points the voxels measured something else. Numerical correction for movement does not 'correct' for that voxel 'seeing' something else at that moment of movement. In this regard, movement of 50% of voxel size could already be considered as problematic. Thus, the issue is more conceptual than numerical: functional connectivity measures could be driven by an artifactual mixture of sources from different locations. That other studies may or may not have used such a lenient threshold is irrelevant.
I suggest that the authors investigate how many volumes or datasets contain head motion at or above threshold of 1 or 1.5 mm. Preferably, the authors also verify that the pattern of results do not change with a stricter threshold."
Thank you in advance
Francesco