help > importing ROI step not working?
Showing 1-3 of 3 posts
Display:
Results per page:
May 4, 2022  03:05 PM | Alex G
importing ROI step not working?
Hello,

I've completed preprocessing of all of my subjects (~180), but in the GUI, when I click "done" it moves to importing participants, conditions, and ROIs. Everything is quick until the ROIs, which at first was taking a very long time for all subjects. I started doing them in batches of 10 which worked (~10 per day) but now it won't even finish 1 subject in a whole day (regardless of whether I send it to background processing or not). The data are attached to an external hard drive, which I know can delay things, but I'm not sure why it would be taking this long. Why would this be? I've been at this for about 3 months now. How can I fix this?

I can try running the job on a cluster, but I'm not sure what in line command codes I can use to not rerun preprocessing and everything else (I just want the equivalent of selecting "done" on the Setup bar without overwriting all the progress I've already made with importing the first several subjects).

Thanks in advance
May 4, 2022  08:05 PM | Alfonso Nieto-Castanon - Boston University
RE: importing ROI step not working?
Hi Alexandra

The Matlab command:

   conn_batch('Setup.done', true, 'Setup.overwrite', false);

is equivalent to clicking 'Done' on the Setup tab and selecting the option 'do not overwrite (skip already computed subjects/ROIs)'. If you want that to run in your cluster instead of in your local computer, simply use:

   conn_batch('Setup.done', true, 'Setup.overwrite', false, 'parallel.N', 10);


where you may change the '10' number to indicate in how many different nodes you want this process to be divided into.

That said, I think there may be other issues at play here that may be making your analyses so slow. Unless the amount of functional data that you have for each individual subject is huge (as in hundreds of runs per subject) I see no reason to justify the ROI import step taking a whole day for a single subject (in normal circumstances it should not take longer than a few minutes per functional run). How many runs you have per subject, what is their length (number of scans) and spatial resolution, and how many ROIs you are including in your CONN project?

Best
Alfonso

Originally posted by Alexandra Muratore:
Hello,

I've completed preprocessing of all of my subjects (~180), but in the GUI, when I click "done" it moves to importing participants, conditions, and ROIs. Everything is quick until the ROIs, which at first was taking a very long time for all subjects. I started doing them in batches of 10 which worked (~10 per day) but now it won't even finish 1 subject in a whole day (regardless of whether I send it to background processing or not). The data are attached to an external hard drive, which I know can delay things, but I'm not sure why it would be taking this long. Why would this be? I've been at this for about 3 months now. How can I fix this?

I can try running the job on a cluster, but I'm not sure what in line command codes I can use to not rerun preprocessing and everything else (I just want the equivalent of selecting "done" on the Setup bar without overwriting all the progress I've already made with importing the first several subjects).

Thanks in advance
May 5, 2022  03:05 PM | Alex G
RE: importing ROI step not working?
Hi Alfonso,

Thanks for your response. The subjects come from a few different studies, but have max 2 resting state scans each (~6 minutes each scan, 12 minutes total). Spatial resolution ranges from 2-3 mm. I parcellated the whole brain into ~135 ROIs. Would this slow it down more? How long should I expect each participant to take? 

Thanks again