help > Realign, unwarp, normalize functional runs separately in CONN
Showing 1-4 of 4 posts
Sep 15, 2021 08:09 PM | Keith Dodd
Realign, unwarp, normalize functional runs separately in CONN
Hello,
I have resting state data split into 2 runs/sessions (with 1 structural). However, in between the two runs, the subjects have exited and then reentered the scanner. I noticed that the standard pipeline analyzes the separate sessions together (i.e. all slices from all runs are realigned and unwarped to the first slice of the first session, and then later this is segmented and normalized). I am worried that this may not be entirely accurate since the patient exits the scanner and re-enters in between as there might be different spatial distortions depending on head positioning changes?
If this is a potential issue (how much might I need to worry about this?), then I want the functional realignment and unwarping, as well normalization, to be done per session separately, instead of combined. However, I have not been able to figure out how to tell CONN to do this. Basically, I want to compare the runs and have CONN treat it as 2 runs to compare for each subject with a single structural scan (so I do not believe I want to tell CONN that these are separate subjects...), but to realign, unwarp and normalise each run separately to help keep the warping of these runs to be the more accurate.
How might I do this within the preprocessing GUI? Additionally, is there any method in CONN to check how well the two runs were realigned and unwarped to the first slice of the first run? To see how much of a factor the leaving and re-entering the scanner had?
Thank you!
I have resting state data split into 2 runs/sessions (with 1 structural). However, in between the two runs, the subjects have exited and then reentered the scanner. I noticed that the standard pipeline analyzes the separate sessions together (i.e. all slices from all runs are realigned and unwarped to the first slice of the first session, and then later this is segmented and normalized). I am worried that this may not be entirely accurate since the patient exits the scanner and re-enters in between as there might be different spatial distortions depending on head positioning changes?
If this is a potential issue (how much might I need to worry about this?), then I want the functional realignment and unwarping, as well normalization, to be done per session separately, instead of combined. However, I have not been able to figure out how to tell CONN to do this. Basically, I want to compare the runs and have CONN treat it as 2 runs to compare for each subject with a single structural scan (so I do not believe I want to tell CONN that these are separate subjects...), but to realign, unwarp and normalise each run separately to help keep the warping of these runs to be the more accurate.
How might I do this within the preprocessing GUI? Additionally, is there any method in CONN to check how well the two runs were realigned and unwarped to the first slice of the first run? To see how much of a factor the leaving and re-entering the scanner had?
Thank you!
Sep 16, 2021 12:09 AM | Alfonso Nieto-Castanon - Boston University
RE: Realign, unwarp, normalize functional runs separately in CONN
Hi Keith,
If you want to first evaluate how well the two runs would be realigned and unwarped to the first slice of the first run, you could do that by first running the standard realign&unwarp step, and then either creating the QA plot labeled "QA realignment: display center slice across multiple sessions", or similarly by using spm_check_registration and selecting the first slice of the two functional runs separately for each subject.
That will often be perfectly fine, even if the different runs are really acquired in different visits/sessions, but if that is not the case in your dataset and you would like to preprocess the different runs separately, you may do that in CONN simply by running preprocessing twice, each time selecting an individual run (uncheck the option that reads 'process all sessions' and then select the session that you would like to preprocess).
Hope this helps
Alfonso
Originally posted by Keith Dodd:
If you want to first evaluate how well the two runs would be realigned and unwarped to the first slice of the first run, you could do that by first running the standard realign&unwarp step, and then either creating the QA plot labeled "QA realignment: display center slice across multiple sessions", or similarly by using spm_check_registration and selecting the first slice of the two functional runs separately for each subject.
That will often be perfectly fine, even if the different runs are really acquired in different visits/sessions, but if that is not the case in your dataset and you would like to preprocess the different runs separately, you may do that in CONN simply by running preprocessing twice, each time selecting an individual run (uncheck the option that reads 'process all sessions' and then select the session that you would like to preprocess).
Hope this helps
Alfonso
Originally posted by Keith Dodd:
Hello,
I have resting state data split into 2 runs/sessions (with 1 structural). However, in between the two runs, the subjects have exited and then reentered the scanner. I noticed that the standard pipeline analyzes the separate sessions together (i.e. all slices from all runs are realigned and unwarped to the first slice of the first session, and then later this is segmented and normalized). I am worried that this may not be entirely accurate since the patient exits the scanner and re-enters in between as there might be different spatial distortions depending on head positioning changes?
If this is a potential issue (how much might I need to worry about this?), then I want the functional realignment and unwarping, as well normalization, to be done per session separately, instead of combined. However, I have not been able to figure out how to tell CONN to do this. Basically, I want to compare the runs and have CONN treat it as 2 runs to compare for each subject with a single structural scan (so I do not believe I want to tell CONN that these are separate subjects...), but to realign, unwarp and normalise each run separately to help keep the warping of these runs to be the more accurate.
How might I do this within the preprocessing GUI? Additionally, is there any method in CONN to check how well the two runs were realigned and unwarped to the first slice of the first run? To see how much of a factor the leaving and re-entering the scanner had?
Thank you!
I have resting state data split into 2 runs/sessions (with 1 structural). However, in between the two runs, the subjects have exited and then reentered the scanner. I noticed that the standard pipeline analyzes the separate sessions together (i.e. all slices from all runs are realigned and unwarped to the first slice of the first session, and then later this is segmented and normalized). I am worried that this may not be entirely accurate since the patient exits the scanner and re-enters in between as there might be different spatial distortions depending on head positioning changes?
If this is a potential issue (how much might I need to worry about this?), then I want the functional realignment and unwarping, as well normalization, to be done per session separately, instead of combined. However, I have not been able to figure out how to tell CONN to do this. Basically, I want to compare the runs and have CONN treat it as 2 runs to compare for each subject with a single structural scan (so I do not believe I want to tell CONN that these are separate subjects...), but to realign, unwarp and normalise each run separately to help keep the warping of these runs to be the more accurate.
How might I do this within the preprocessing GUI? Additionally, is there any method in CONN to check how well the two runs were realigned and unwarped to the first slice of the first run? To see how much of a factor the leaving and re-entering the scanner had?
Thank you!
Sep 16, 2021 08:09 PM | Keith Dodd
RE: Realign, unwarp, normalize functional runs separately in CONN
Thank you, this is really helpful.
If I run the preprocessing twice, one for each run, will that effect how I will be able to compare the data later after the preprocessing? Based on what I am seeing, I suspect this will not be an issue, but I would like to double check before I pull the trigger on that since the processing has been taking a while.
Perhaps there is a method in CONN that will allow me to increase the processing speed (use more cores?), but I am unsure about that.
If I run the preprocessing twice, one for each run, will that effect how I will be able to compare the data later after the preprocessing? Based on what I am seeing, I suspect this will not be an issue, but I would like to double check before I pull the trigger on that since the processing has been taking a while.
Perhaps there is a method in CONN that will allow me to increase the processing speed (use more cores?), but I am unsure about that.
Sep 23, 2021 03:09 AM | Keith Dodd
RE: Realign, unwarp, normalize functional runs separately in CONN
OK, So I did this, but I cannot seem to figure out how to see the
QA plots for both sessions? When I run the QA plots now for all
subjects, it will only show me the functional information for
Session 1, but no information for Session 2 (despite the fact that
I have definitely run this). How can I create QA plots for both
sessions so I can check how the realignment and then warping went
for both sessions and not just session 1?
Thanks!
Keith
Originally posted by Alfonso Nieto-Castanon:
Thanks!
Keith
Originally posted by Alfonso Nieto-Castanon:
Hi
Keith,
If you want to first evaluate how well the two runs would be realigned and unwarped to the first slice of the first run, you could do that by first running the standard realign&unwarp step, and then either creating the QA plot labeled "QA realignment: display center slice across multiple sessions", or similarly by using spm_check_registration and selecting the first slice of the two functional runs separately for each subject.
That will often be perfectly fine, even if the different runs are really acquired in different visits/sessions, but if that is not the case in your dataset and you would like to preprocess the different runs separately, you may do that in CONN simply by running preprocessing twice, each time selecting an individual run (uncheck the option that reads 'process all sessions' and then select the session that you would like to preprocess).
Hope this helps
Alfonso
Originally posted by Keith Dodd:
If you want to first evaluate how well the two runs would be realigned and unwarped to the first slice of the first run, you could do that by first running the standard realign&unwarp step, and then either creating the QA plot labeled "QA realignment: display center slice across multiple sessions", or similarly by using spm_check_registration and selecting the first slice of the two functional runs separately for each subject.
That will often be perfectly fine, even if the different runs are really acquired in different visits/sessions, but if that is not the case in your dataset and you would like to preprocess the different runs separately, you may do that in CONN simply by running preprocessing twice, each time selecting an individual run (uncheck the option that reads 'process all sessions' and then select the session that you would like to preprocess).
Hope this helps
Alfonso
Originally posted by Keith Dodd:
Hello,
I have resting state data split into 2 runs/sessions (with 1 structural). However, in between the two runs, the subjects have exited and then reentered the scanner. I noticed that the standard pipeline analyzes the separate sessions together (i.e. all slices from all runs are realigned and unwarped to the first slice of the first session, and then later this is segmented and normalized). I am worried that this may not be entirely accurate since the patient exits the scanner and re-enters in between as there might be different spatial distortions depending on head positioning changes?
If this is a potential issue (how much might I need to worry about this?), then I want the functional realignment and unwarping, as well normalization, to be done per session separately, instead of combined. However, I have not been able to figure out how to tell CONN to do this. Basically, I want to compare the runs and have CONN treat it as 2 runs to compare for each subject with a single structural scan (so I do not believe I want to tell CONN that these are separate subjects...), but to realign, unwarp and normalise each run separately to help keep the warping of these runs to be the more accurate.
How might I do this within the preprocessing GUI? Additionally, is there any method in CONN to check how well the two runs were realigned and unwarped to the first slice of the first run? To see how much of a factor the leaving and re-entering the scanner had?
Thank you!
I have resting state data split into 2 runs/sessions (with 1 structural). However, in between the two runs, the subjects have exited and then reentered the scanner. I noticed that the standard pipeline analyzes the separate sessions together (i.e. all slices from all runs are realigned and unwarped to the first slice of the first session, and then later this is segmented and normalized). I am worried that this may not be entirely accurate since the patient exits the scanner and re-enters in between as there might be different spatial distortions depending on head positioning changes?
If this is a potential issue (how much might I need to worry about this?), then I want the functional realignment and unwarping, as well normalization, to be done per session separately, instead of combined. However, I have not been able to figure out how to tell CONN to do this. Basically, I want to compare the runs and have CONN treat it as 2 runs to compare for each subject with a single structural scan (so I do not believe I want to tell CONN that these are separate subjects...), but to realign, unwarp and normalise each run separately to help keep the warping of these runs to be the more accurate.
How might I do this within the preprocessing GUI? Additionally, is there any method in CONN to check how well the two runs were realigned and unwarped to the first slice of the first run? To see how much of a factor the leaving and re-entering the scanner had?
Thank you!