Dear Sophie,

> Coregister (this moves the anatomic into the space of the functional)
> ·       Need to register anatomic (a.img) to functionals (af*.img)
> ·       Select coregister button
> ·       Number of subjects=1
> ·       Option coregister only
> ·       Modality of first target image=T1 MRI
> ·       Modality of first object image=T1 MRI
> ·       Select target image ****use 1st image of the run that was
> performed the
> closest in time to the anatomic
> ·       Select Object image, a.img for that subject
> ·       Select other images for subject 1 just press: DONE

There some possible reasons for this misregistration, you described.
But first of all, you should check the procedure, you described in you
e-mail. You said, that you perform a coregistration, that should moves the
anatomic scan into the space of the functionals. This means, that you had to
select: 'modality of first target image= EPI' instead of T1.
You should further use the 'CheckReg' option, to proof the goodness of
coregistration by selecting the a.img and the first af.img. By clicking into
the images, you can move through them and check in both images for example
the edges of ventricles etc.
If the coregistration failed, you can also try the alternative
coregistration by changing the 'corgistration defaults' to 'mutual
informations'.
I think, these procedure could help to solve your problem.

If this didn't help, you can used an other way of normalization, which is
usual used in context of voxel based morphometry. I know the problem, that
the gel pads of some headphones can introduce some problems in the
normalization procedure of T1 images, similar to that, you described.
The can solve this problem, by segmenting the corgistered T1 image.
Normalize the '*_seg1' image to the 'gray.img' of the 'apriori' directory
and use that normalization parameters in the write normalize procedure, as
you described.
This increase the normalization of the gray matter, with a small lost of
accuracy with in the white matter or other subcortical structures
(ventricles etc.).

Good luck,

Karsten

------------------------------------------------
Dipl. Phys. Karsten Specht

(Medizin Center Bonn, Spessartstrasse 9, 53119 Bonn, Germany)

currently at:
---------------------

Department of Biological and Medical Psychology
Aarstadveien 21
5009 Bergen
Norway
http://www.uib.no/ibmp/bjorg/english/
[log in to unmask]

--------------------------------------------------


> -----Opprinnelig melding-----
> Fra: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]]På
> vegne av Sophie Lafaille
> Sendt: Friday, October 19, 2001 5:20 PM
> Til: [log in to unmask]
> Emne: normalization skew
>
>
> I think I'm having a normalization problem. I have 2 group with 3
> experiments
> in each (wnw, stut, oral). 1 group has 3 subjects and the other
> has 2 (we've
> just started our experiments). In the jpegs(0.99 threshold) that
> I've attached,
> it shows that the normalization is off in the z axis (not
> stretched enough) and
> shifted to the left axially, and overstretched into the eyeballs!
> Since each
> experiment and group shows the same skew, it must be in my normalization
> procedure?
> Here is my procedure (taken from my manual that I'm building)
> slice timing
> Reorient anatomic and functional
> ·       use display to bring up a.img
> ·       set pitch and roll to 1.57
> ·       find Anterior Commisure
> ·       record AC numbers in mm. Eg (0.0 –8.8 –1.2)
> ·       enter into Right, forward & up, REVERSE numbers (0.0 8.8 1.2)
> ·       put  0 0 0 in mm box.
> ·       Check and make sure that crosshairs are on AC
> ·       Reorient image a.img
> ·       Use display to bring up functional
> ·       Use unix program to get numbers (this usually shifts
> x-axis by about
>         0.7 of a mm).
> ·       enter into Right, forward & up the numbers the program
> gives (0.7 8.8
> 1.2)
> ·       put  0 0 0 in mm box.
> ·       Check and make sure that crosshairs are on AC
> ·       Reorient image af*.img
>
> Use check reg to make sure transformations are correct
>
>
> Coregister (this moves the anatomic into the space of the functional)
> ·       Need to register anatomic (a.img) to functionals (af*.img)
> ·       Select coregister button
> ·       Number of subjects=1
> ·       Option coregister only
> ·       Modality of first target image=T1 MRI
> ·       Modality of first object image=T1 MRI
> ·       Select target image ****use 1st image of the run that was
> performed the
> closest in time to the anatomic
> ·       Select Object image, a.img for that subject
> ·       Select other images for subject 1 just press: DONE
>
>
>
>
>         Realign
> ·       number of subjects 1
> ·       number of sessions 3
> ·       select af*.img for all 3 experiments make sure to select
> the first run
> closest to a.img as the first run (all others in order of expt).
> ·       method- coregister ONLY
>
>
> use checkreg again to check all af*.img and a.img for correct location
>
> go to defaults and change normalization, parameter estimation to R is R
>         Normalize- Anatomic
> ·       press normalize button
> ·       determine + write normalize
> ·       1 subject, select a.img
> ·       select again a.img
> ·       use T1 as template
> ·       SINC interpolation
> ·       Use display to look at na.img to make sure crosshairs still on AC
>
>         Normalize- functionals
> ·       press normalize button and select write normalize only
> ·       select all af*.img files for ONE experiment for same subject
> ·       SINC interpolation
> ·       Do each run for that subject separately
>
> Does anyone have any ideas as to why my overall brain activation
> doesn't look
> right? Are there certain default parameters I should be changing?
> voxel size, reiterations etc...
>
>
> thanks sophie
>