Hi Alfonso,

I would like to be able to specify unsmoothed data for the WM/CSF PCA extraction (to prevent smoothing in cortical signal in WM/CSF) and smoothed data for everything else. From what I can understand from your code, the source for the ROIs is specified in the source field of ROI_SubjectXXX_SessionXXX.mat and it is the same source for all ROIs, I am unsure on how to change the source only for the WM/CSF ROIs. Any suggestions?

Kind regards,
Vincent.

Hi Vincent,

The ROI extraction procedure already attempts to use the unsmoothed data (if it exists) for all of the ROIs (for the exact same reason you mention). The unsmoothed BOLD data is assumed to be in the same location as the smoothed volumes but without an initial 's' in the filename (this is the typical convention in SPM, which is appropriate when you use the default conn spatial preprocessing steps). If you use a different convention for unsmoothed/smoothed volume filenames please use the attached conn_process.m file instead (this patch works for version 13o) and manually modify the field CONN_x.Setup.unsmoothedvolumes field from the command line before running your 'setup' step, e.g.:

global CONN_x;
CONN_x.Setup.unsmoothedvolumes=@(x)regexprep(x,'^(.*[\\/])?s([^\\/]+)$','$1$2');

(this example is for SPM convention -it removes a leading 's' from the filename-; modify as you please for other conventions)

Regarding the application of this rule differently for different ROIs, this is currently not possible in the toolbox but I will add a new field in the Setup->ROIs tab to enable/disable this behavior separately for each ROI (and perhaps a more user-friendly way to deal with different filename conventions) in the next release of the toolbox (hopefully in a couple of days).

Hope this helps and let me know your thoughts.
Best
Alfonso

 

Originally posted by Vincent Beliveau:

Hi Alfonso,

Thanks for your answer. I think it would be great to be able to specify exactly which unsmoothed volumes should be used for the ROIs extraction and which smoothed volumes should be used for the correlation analysis. Combined with the option to toggle on/off specific rules, I think most situations should be covered ;)

I'm looking forward for the new release!

Kind regards,
Vincent.

Hi Vincent,

I have added this functionality to the latest release of the toolbox (conn v.13.p). I tried to strike some balance between flexibility and simplicity particularly for the GUI, so some of the new options are currently available only through batch process while others are directly available through both batch processes as well as the gui (this may change in the future).

In the gui Setup->Functional tab you can now specify whether ROI BOLD timeseries should be extracted from: a) the selected functional files; b) the same filenames but removing a leading 's' from the file name (this is SPM's convention for 'unsmoothed' volumes); or c) define a different filename rule (e.g. if you want to specify 'unsmoothed' volumes but your data comes from a non-SPM pipeline). You can also do this using batch processes (see help conn_batch; batch.Setup.roiextract and roiextract_rule fields). In addition if you want to manually specify a completely different set of files to be used for ROI extraction you may also do so now using batch processes (see help conn_batch;  batch.Setup.roiextract_functionals field). Last if you want to specify that some ROIs should instead extract their BOLD timeseries from the original functional volumes instead (typically smoothed) you may also do that now using batch processes (batch.Setup.rois.unsmoothedvolumes field). 

Hope this helps and let me know if you run into any issues and/or any suggestions/thoughts.

Best
Alfonso


Originally posted by Vincent Beliveau: