Dear Marcos,
these are just mean and standard deviations of intensities of the images
at different levels of processing. What the best measure is depends on
what you want to do with them.
Cheers,
/Eugenio
On Tue, 2013-06-11 at 11:46 -0300, Marcos Martins da Silva wrote:
> Hi, Eugenio,
> TY again for your help. I did the computing without errors after your
> suggestions. I also computed the intensities based on nu.mgz, norm.mgz
> and T1.mgz. I would like to know the best measure I should consider,
> including because I intend to compare these values with intensity data
> from basal ganglia and cortical areas as I can find in aseg.stats. It
> follows my results:
> norm.mgz based
> meannorm= 78.725
> variancenorm = 20.258 (stdev = 4.500888801)
>
> nu.mgz based
> meannu = 87.362
> variancenu = 26.343 (stdev =5.132543229)
>
> T1.mgz based
> meant1 = 84.023
> variancet1 = 23.277 (stdev =4.824624338)
>
>
> It follows the MatLab/Octave code I used
> Norm=MRIread('norm_postres.mgz');
> datanorm=Norm.vol(:);
> Ps=nMRIread('posterior_right_subiculum.mgz');
> postsub=double(Ps.vol(:));
> meannorm=sum(datanorm.*postsub)/sum(postsub);
> variancenorm=sum((datanorm-meannorm).^2.*postsub)/sum(postsub);
> Nu=MRIread('nu_postres.mgz');
> datanu=Nu.vol(:);
> meannu=sum(datanu.*postsub)/sum(postsub);
> variancenu=sum((datanu-meannu).^2.*postsub)/sum(postsub);
> T1=MRIread('T1_postres.mgz');
> datat1=T1.vol(:);
> meant1=sum(datat1.*postsub)/sum(postsub);
> variancet1=sum((datat1-meant1).^2.*postsub)/sum(postsub);
>
> The *postres.mgz files are nu, norm and T1 resampled as you suggested.
> And nMRIread is the workarounded version of MRIread I said in previous
> message.
> Cheers,
> Marcos
> Em Seg, 2013-06-10 às 16:58 -0400, Juan Eugenio Iglesias escreveu:
> > Hi Marcos,
> > 1. your fix of MRIread.m is great. I forgot of this bug; we should
> > totally take care of it.
> > 2. a problem is that nu.mgz (or norm.mgz) and the posteriors are in
> > difference voxel space. So, you need to resample norm.mgz to the space
> > of the subfields. To do so, you can use mri_convert with the option -rl
> > ("reslice like"):
> > mri_convert norm.mgz norm_resampled.mgz -rl posterior_subiculum.mgz
> > 3. Now you can do:
> > A=MRIread('norm.mgz');
> > data=A.vol(:);
> > B=MRIread('posterior_subiculum.mgz');
> > post=double(B.vol(:));
> > mean=sum(data.*post)/sum(post);
> > variance=sum((data-mean).^2.*post)/sum(post);
> > Cheers,
> > /Eugenio
> >
> > On Mon, 2013-06-10 at 17:17 -0300, Marcos Martins da Silva wrote:
> > > Hi, Eugenio
> > > TY for your fast help. I understood you were suggesting to compute
> > > that on MatLab and I tried this:
> > >
> > > NU=MRIread('nu.mgz') ### that runs ok
> > > Ps=MRIread('posterior_left_subiculum.mgz') ### it fails with the
> > > following message
> > > WARNING: error reading MR params
> > > Attempted to access mr_parms(1); index out of bounds because
> > > numel(mr_parms)=0.
> > >
> > > Error in MRIread (line 100)
> > > tr = mr_parms(1);
> > >
> > > I solved creating a little customized nMRIread.m with the following
> > > changes:
> > >
> > > if numel(mr_parms) > 0
> > > tr = mr_parms(1);
> > > flip_angle = mr_parms(2);
> > > te = mr_parms(3);
> > > ti = mr_parms(4);
> > > else
> > > mr_parms(1) = 0;
> > > mr_parms(2) = 0;
> > > mr_parms(3) = 0;
> > > mr_parms(4) = 0;
> > > tr = mr_parms(1);
> > > flip_angle = mr_parms(2);
> > > te = mr_parms(3);
> > > ti = mr_parms(4);
> > > end
> > >
> > > With those changes I assigned 0 to mr-parms elements so it runs
> > > without errors
> > > But whe I tried the following line I copy and pasted from your
> > > message:
> > > mean=sum(Ps.*NU)/sum(Ps); ### I got the following error
> > > Undefined function 'times' for input arguments of type 'struct'.
> > >
> > > Any help?
> > > Cheers, Marcos
> > >
> > > PS: I promised I will post the final solution to list. But I guess it
> > > is more didactic if these little problems are solved before.
> > > TY, again.
> > > Em Seg, 2013-06-10 às 14:19 -0400, Juan Eugenio Iglesias escreveu:
> > > > Hi Marcos,
> > > > the right way of doing this is using the soft posteriors to compute the
> > > > mean and variance, rather than thresholding at p=0.5. For instance, if
> > > > you wanted to compute the mean and variance of the intensitites of the
> > > > subiculum, you would do something like this:
> > > > mean=sum(Ps.*NU)/sum(Ps);
> > > > variance=sum(Ps.*(NU-mean).^2)/sum(Ps);
> > > > (where Ps is the posterior of the subiculum)
> > > > Cheers,
> > > > /Eugenio
> > > >
> > > >
> > > >
> > > > On Mon, 2013-06-10 at 15:12 -0300, Marcos Martins da Silva wrote:
> > > > > Hi, Freesurfer Experts
> > > > >
> > > > > After usual processing with recon-all -all we get the aseg.stats file
> > > > > with several data including intensity values like:
> > > > > normMean normStdDev
> > > > > normMin normMax normRange
> > > > > Left-Hippocampus 77.8939 7.5748 46.0000
> > > > > 105.0000 59.0000
> > > > >
> > > > > How could I get similar results for each hippocampal subfield,
> > > > > assuming I also generated all posterior*.mgz files corresponding to
> > > > > each subfield?
> > > > > I guess I should first binarize each of the subfields file with a
> > > > > threshold=127 to map all the pertinent voxels, and then use these
> > > > > files as masks over nu.mgz and calculate the intensities values. But I
> > > > > do not know the best way to accomplish this after the binarize step.
> > > > >
> > > > > Thank you in advance for any help.
> > > > >
> > > > > Marcos
> > > > > _______________________________________________
> > > > > Freesurfer mailing list
> > > > > Freesurfer@nmr.mgh.harvard.edu
> > > > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
> > > >
> > >
> >
>
--
Juan Eugenio Iglesias, PhD
http://www.jeiglesias.com
igles...@nmr.mgh.harvard.edu
Athinoula A. Martinos Center for Biomedical Imaging
149 Thirteenth Street, Suite 2301
Charlestown, Massachusetts 2129
U.S.A.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.
