Hi Matthew and Jalmar,

Yes, you are right that the preprocessing steps (including coregistration/normalization) in CONN are only applied to your functional/structural volumes (they will not touch your ROI files). If all of your ROIs are in subject-space (already co-registered to your original functional data before any of the coregistration/normalization steps are applied to your functional/structural data during preprocessing) there are a couple of approaches you could use:

1) the simplest approach is to define, in the Setup.functional tab, the source of ROI-level functional data as your functional data prior to normalization, while leaving your fully-preprocessed data as the source of voxel-level functional data. This will make sure that your ROI-level data is extracted from the correctly-coregistered functional data (both ROIs and functional data in subject-space), while still using the fully-preprocessed&normalized functional data for voxel-level analyses. When using the GUI you can do this simply by selecting the 'Other: manually define (other conventions)' option and then enter in the "Find string(s)" field "^sw" (without the quotes) and leave empty the "Replace with string(s)" field (i.e. the ROI-level functional volumes is in the same folder as the voxel-level data and the filenames do not contain the the leading 'sw' characters -which are associated with normalization and smoothing-). You can do the same using batch scripts using the batch.Setup.roiextract and batch.Setup.roiextract_rule fields (or the batch.Setup.roiextract_functionals field if you prefer to explicitly point to your prior-to-normalization files, instead of using a filename-rule). 

2) alternatively, as you suggest, you may apply the same normalization transformation to your ROI files to have them in MNI space by using either the files named 'y_*.nii' (if using SPM12) or the files named '*_seg_sn.mat' (if using SPM8). These files are created during the normalization step and they will be located in in your functional/structural data folders. You can use in SPM the option "normalize (Write)" to do this (e.g. in SPM12 you would enter for each subject the y_*.nii files in the 'deformation field' fields, and your ROI files in the 'Images to write' field to do this). Note that for this to work correctly the original preprocessing steps must not contain any of the "Center to (0,0,0) coordinates" steps, since those transformations will not be reflected in these y_*.nii or *_seg_sn.mat files. 

Hope this helps
Alfonso

Originally posted by Matthew Defenderfer: