In that case, I think you will want to change the design at each voxel based on which subjects are present. I have not tried to do this with pvr, but there is no reason it should not work. To do this, create a binary volume from your data that has the 0s in it, something like
mri_binarize --i y.mgh --abs --min .0000000000001 --o framemask.mgh
The run mri_glmfit with --frame-mask framemask.mgh
doug
On 11/08/2013 01:11 PM, celine@nmr.mgh.harvard.edu wrote:
Thanks Doug, Actually it is still the --pvr analysis, so I am regressing vertex by vertex two different modalities in one group of subjects, but for one of them, we apply a threshold to remove regions that have bad quality data. Let me know if I am not clear! Celine
you will want to use the -no-prune flag.If thisis a group analysis, you should be very, very careful that you know what you are doing. In general, setting values to 0 prior to group analysis will create biased results. What you really need to do is have adifferent model for each voxel that includes/excludes certain subjects at certain voxels. If this is what you want, let me know and I'll give you instructions on how to do it.
doug
On 11/08/2013 12:46 PM, celine@nmr.mgh.harvard.edu wrote:
Hi, When using glm_fit with an input volume that has many voxels with a 0 value (due to a previous thresholding for quality reasons), and the voxel removes might be distributed randomly across our population, is there a way not to take into account those voxels in the glm_fit model? I saw the --pune flag, but I am not sure if it is the one to use. Thanks Celine _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Yes thanks so much, it worked very fine Celine
In that case, I think you will want to change the design at each voxel based on which subjects are present. I have not tried to do this with pvr, but there is no reason it should not work. To do this, create a binary volume from your data that has the 0s in it, something like
mri_binarize --i y.mgh --abs --min .0000000000001 --o framemask.mgh
The run mri_glmfit with --frame-mask framemask.mgh
doug
On 11/08/2013 01:11 PM, celine@nmr.mgh.harvard.edu wrote:
Thanks Doug, Actually it is still the --pvr analysis, so I am regressing vertex by vertex two different modalities in one group of subjects, but for one of them, we apply a threshold to remove regions that have bad quality data. Let me know if I am not clear! Celine
you will want to use the -no-prune flag.If thisis a group analysis, you should be very, very careful that you know what you are doing. In general, setting values to 0 prior to group analysis will create biased results. What you really need to do is have adifferent model for each voxel that includes/excludes certain subjects at certain voxels. If this is what you want, let me know and I'll give you instructions on how to do it.
doug
On 11/08/2013 12:46 PM, celine@nmr.mgh.harvard.edu wrote:
Hi, When using glm_fit with an input volume that has many voxels with a 0 value (due to a previous thresholding for quality reasons), and the voxel removes might be distributed randomly across our population, is there a way not to take into account those voxels in the glm_fit model? I saw the --pune flag, but I am not sure if it is the one to use. Thanks Celine _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Hi Doug and freesurfer team
We just realized that the output framemask created by mri_binarize does not have the same dimension than the input file y.mgh, which is a concatenated file created through mris_preproc.
so the dimension of our y.mgh is: 163842 x 1 x 1 x 33 and by running this command (see below): mri_binarize --i y.mgh --abs --min .0000000000001 --o framemask.mgh
the dimension of output file framemask.mgh is: 163842 x 1 x 1
Would there be a way to get a 4D binary mask of zero-voxels that I could use subsequently with the --frame-mask flag of mri_glmfit?
Thanks for your help Celine
Yes thanks so much, it worked very fine Celine
In that case, I think you will want to change the design at each voxel based on which subjects are present. I have not tried to do this with pvr, but there is no reason it should not work. To do this, create a binary volume from your data that has the 0s in it, something like
mri_binarize --i y.mgh --abs --min .0000000000001 --o framemask.mgh
The run mri_glmfit with --frame-mask framemask.mgh
doug
On 11/08/2013 01:11 PM, celine@nmr.mgh.harvard.edu wrote:
Thanks Doug, Actually it is still the --pvr analysis, so I am regressing vertex by vertex two different modalities in one group of subjects, but for one of them, we apply a threshold to remove regions that have bad quality data. Let me know if I am not clear! Celine
you will want to use the -no-prune flag.If thisis a group analysis, you should be very, very careful that you know what you are doing. In general, setting values to 0 prior to group analysis will create biased results. What you really need to do is have adifferent model for each voxel that includes/excludes certain subjects at certain voxels. If this is what you want, let me know and I'll give you instructions on how to do it.
doug
On 11/08/2013 12:46 PM, celine@nmr.mgh.harvard.edu wrote:
Hi, When using glm_fit with an input volume that has many voxels with a 0 value (due to a previous thresholding for quality reasons), and the voxel removes might be distributed randomly across our population, is there a way not to take into account those voxels in the glm_fit model? I saw the --pune flag, but I am not sure if it is the one to use. Thanks Celine _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
celine＠nmr.mgh.harvard.edu -
Douglas N Greve