help > intensity normalization functional data
Showing 1-4 of 4 posts
Display:
Results per page:
Sep 15, 2015  08:09 PM | Anna Zilverstand
intensity normalization functional data
Hi Alfonso,

we preprocessed our data twice, using once fsl and once afni scripts. While fsl has an intensity normalization step for the preprocessing of functional data, the afni pipeline does not. When I compare the data visually it looks as if CONN is applying an intensity thresholding? Is this only for visualization purposes, or also when computing the correlation matrixes (I am doing graph theory analyses)? If it would apply intensity thresholding in the computations, is it possible to set the threshold? 

Thank you so much!!!
Anna
Sep 18, 2015  03:09 AM | Alfonso Nieto-Castanon - Boston University
RE: intensity normalization functional data
Hi Anna,

CONN will apply intensity thresholding to the functional data only if you select the 'implicit masking' option in the Setup.Options 'Analysis mask (voxel-level)' field. If you do, CONN will use the default SPM procedure for implicit masking (which is based on a 80% threshold on the global signal). If you use instead the default 'explicit masking' option then it will use SPM's 'brainmask' masking instead. Any of these masking procedures only affect voxel-level analyses (and for those it defines the voxels that are going to be included in the analyses; i.e. voxels outside the masks are going to be disregarded in the rest of the voxel-level analyses). ROI-level analyses (and the derived graph-theory analyses) are unaffected by the choice of voxel-level masking. If you wish to use an alternative masking procedure for voxel-level analyses, you could either define explicitly your choice of mask file, or alternatively if using implicit masking perhaps edit the conn_process.m file in the line that reads "a=a&(b>0.80*mb);" (line 494 in release 15e) and modify there the 0.80 value to an alternative value if you wish?. I will see if I can easily add the option to more easily define/modify this implicit-masking threshold in the next release. 

Hope this helps
Alfonso
Originally posted by Anna Zilverstand:
Hi Alfonso,

we preprocessed our data twice, using once fsl and once afni scripts. While fsl has an intensity normalization step for the preprocessing of functional data, the afni pipeline does not. When I compare the data visually it looks as if CONN is applying an intensity thresholding? Is this only for visualization purposes, or also when computing the correlation matrixes (I am doing graph theory analyses)? If it would apply intensity thresholding in the computations, is it possible to set the threshold? 

Thank you so much!!!
Anna
Sep 21, 2015  02:09 PM | Anna Zilverstand
RE: intensity normalization functional data
Hi Alfonso, great, thanks for the explanation!!! All the best, Anna
Oct 30, 2020  04:10 PM | Sarah Kark - UCI
RE: intensity normalization functional data
Hi Alfonso,

So does 'explicit masking' option perform intensity normalization?

I'd like to intensity normalize my swau images for entry into wavelet despiking program.

Thank you!
Sarah