Hi Alfonso,

I was hoping you might be able to give some clarification on the data contained within the ROI_Subject_Condition files within the results/preprocessing directory of a given conn project. As a bit of context, I preprocessed data using CONN, and a colleague then used the preprocessed functional data to generate a project on SPM. The data from this SPM project was then used to extract ROI timeseries data using ROI's specified by the CONN default atlas. I then tried to reproduce these data by extracting from the ROI_Subject_Condition files directly, but was surprised to find the data did not match. I have worked with the data from these files in the past, so I am familiar with how they are formatted (I am extracting from the relevant cell within the 'data' cell array for each ROI), but I am curious about what this data actually is and why they would be different from those extracted from SPM?

I've also been meaning to ask a couple questions regarding the formatting of these files, so I'll add them here as well. I have multiple sessions and multiple conditions within my project, and I've noticed that rather than only containing the relevant volumes for the given condition, each ROI_Subject_Condition contains data for all sessions present for a given subject. I've been working under the assumption that these data are ordered by session, and thus have only been extracting the sections relevant to the condition I am looking at (e.g if I am looking at session two, I extract only the data starting at 1 + the number of volumes in session 1, and ending at that value + the total volumes contained in session 2). is this correct?

I've also noticed that the data doesn't vary between different condition files (eg the data field of ROI_Subject001_Condition002 and ROI_Subject001_Condition003 are identical). And finally, I've noticed that there is never a condition001 file, and as a result the list of conditions in condition000 are all offset by one (Condition002 is condition 1, Condition003 is conditon 2, etc). I was wondering if there was a reason for these redundancies, or if it is simply a quirk of how the data needs to be coded to function properly? Mostly I just want to make sure my understanding of these files is correct, and I'm hoping you might be able to clarify these points so I can have a better understanding of these data going forward.

Thanks and best,

Alex Belden

Hi Alex,

Yes, your interpretation about the data being sorted by sessions is correct. To double-check you can also use the contents of the conditionweights field which will identify which scans/timepoints are associated with this condition.

Regarding those redundancies in CONN's files, where the field data contains the raw data even for samples/timepoints where the condition is not defined, is mainly to accommodate cases where the source data varies per condition -e.g. when using condition-specific band-pass filters- or analyses that work on the original timeseries -e.g. gPPI-). 

And regarding your first question, I am unsure what the data extracted by CONN may not match the data extracted manually. I imagine that, if you are using the data in CONN's results/preprocessing folder, then the differences are possibly due to CONN's ROI timeseries having been already denoised? If that is not the case please let me know more details on how you extracted the timeseries manually to see if that helps suggest other potential differences.

Hope that helps
Alfonso

Originally posted by Alexander Belden: