help
help > RE: Using CONN's ART output in other software
Aug 28, 2015 02:08 AM | Alfonso Nieto-Castanon - Boston University
RE: Using CONN's ART output in other software
Hi Darren,
Yes, your approach looks perfectly fine and the thresholds values seem conservative enough, I see no issues at all. Also regarding the order of realignment/slice-timing you are right that particularly for event-related designs slice-timing information can have a larger impact so it is typically justified to use the alternative order (I was just mentioning in case you were not aware of the discussion regarding this issue). Last, just to mention that as well, one very good alternative for those cases where both slice-timing information and motion might have both non-negligible effects would be Roche 2011 four-dimensional registration approach (although unfortunately that is currently only available in python as far as I know)
Best
Alfonso
Originally posted by Darren Yeo:
Yes, your approach looks perfectly fine and the thresholds values seem conservative enough, I see no issues at all. Also regarding the order of realignment/slice-timing you are right that particularly for event-related designs slice-timing information can have a larger impact so it is typically justified to use the alternative order (I was just mentioning in case you were not aware of the discussion regarding this issue). Last, just to mention that as well, one very good alternative for those cases where both slice-timing information and motion might have both non-negligible effects would be Roche 2011 four-dimensional registration approach (although unfortunately that is currently only available in python as far as I know)
Best
Alfonso
Originally posted by Darren Yeo:
Hi Alfonso,
Thank you so much for your detailed clarification! I'm aware of the difference in the order of preprocessing steps and I agree with the potential implications that you've raised. In BrainVoyager, however, the default sequence involves slice-timing correction before realignment. From what I've read, it seems appropriate to perform slice-timing correction first (probably more for an event-related design than a blocked design) if the TR is greater than 2s and if the scanning sequence is interleaved.
To what extent of in-scanner motion is considered low, in your opinion? For my current data, I used thresholds of 3SD for the deviation from the global mean signal and 1 mm for the composite subject motion (supposedly a combination of 1 mm for translational displacement and 0.2 rad for rotational displacement). My subsequent threshold for subject exclusion is 20% of volumes are outliers. Is that conservative enough?
Thanks!
Best,
Darren
Thank you so much for your detailed clarification! I'm aware of the difference in the order of preprocessing steps and I agree with the potential implications that you've raised. In BrainVoyager, however, the default sequence involves slice-timing correction before realignment. From what I've read, it seems appropriate to perform slice-timing correction first (probably more for an event-related design than a blocked design) if the TR is greater than 2s and if the scanning sequence is interleaved.
To what extent of in-scanner motion is considered low, in your opinion? For my current data, I used thresholds of 3SD for the deviation from the global mean signal and 1 mm for the composite subject motion (supposedly a combination of 1 mm for translational displacement and 0.2 rad for rotational displacement). My subsequent threshold for subject exclusion is 20% of volumes are outliers. Is that conservative enough?
Thanks!
Best,
Darren
Threaded View
| Title | Author | Date |
|---|---|---|
| Darren Yeo | Aug 25, 2015 | |
| Alfonso Nieto-Castanon | Aug 26, 2015 | |
| Darren Yeo | Aug 27, 2015 | |
| Alfonso Nieto-Castanon | Aug 28, 2015 | |