Doubtless it is because the .img file does not have any geometry with it. If it's there, it's in the form of a .mat file. If it's not there, then no program (mri_convert, tkmedit, or anything else) is going have a clue as to which way is up. To some extent, you can undo the damage from using ANALYZE format with mri_convert using the -iid, -ijd, and -ikd arguments. Each flag must be followed by 3 values, which the direction cosines for the col (i), row (j) and slice (k). You don't have to know these excatly, just the major direction. Play with them until the volume is properly oriented in tkmedit. Note: you won't be able to tell whether it is left-right reversed (but you can't tell that now). It's better to run mri_convert on the source dicom. Dicom has the geometry info in it (usually), and mri_convert can extract it.

doug




antao@itsa.ucsf.edu wrote: 
I have a image xAD1378-T1.img with spm format. after I converted it to
COR with mri_convert xAD1378-T1.img AD1378/mri/orig and run FLIRT as Xiao 
suggests, I found that the anterior and posterior parts of tranformed
image was inversed. Xiao thinks it is due to mri_convert, Does anyone
know how to fix it? thx,

Antao

---------- Forwarded message ----------
From: Xiao Han <xhan@nmr.mgh.harvard.edu>
Date: Mon, 22 Aug 2005 16:03:46 -0400 (EDT)
Subject: Re: [Freesurfer] coregistration

It has nothing to do with fsl_register.
It's because of the conversion from .img to COR format.
That's why I suggested you to convert all your data to COR or mgz first 
before processing.

The reason is because Analyze store data in an anterior-to-posterior 
fashion, but when mri_convert read the data, it assumed 
posterior-to-anterior (due to a missing .mat file).

I dont know how to fix that. You can send an email to freesurfer list.

-Xiao


On Mon, 22 Aug 2005, antao@itsa.ucsf.edu wrote:

  
After I run:
fsl_register_new -i AD1378/mri/orig -r average_305.mnc -o AD1378/tmp/orig
-dof 9 -cost corratio -ltamat AD1378/mri/transforms/talairach.lta

I check the tranformation with: tkmedit -f average_305.mnc -aux
AD1378/tmp/orig. I find that the anterior and posterior parts of  the
image AD1378/tmp/orig are inversed. Do you know why?

When I run fsl_register_new with images .img before converting to COR,
the result was good.

Antao

n Mon, 22 Aug 2005 14:16:34 -0400 (EDT) Xiao Han wrote:

    
On Mon, 22 Aug 2005, antao@itsa.ucsf.edu wrote:

      
This are the steps I use fsl_register_new:
1) mksubjdir AD1378
2) mri_convert xAD1378-T1.img AD1378/mri/orig
3) fsl_register_new -i AD1378/mri/orig -r average_305.mnc -o
AD1378/mri/orig -dof 9 -cost corratio -ltamat
AD1378/mri/transforms/talairach.lta

I have a question here. Should I replace orig when I run
        
fsl_register_new
      
or I should choose different filename?
        
No, you shouldn't replace the original orig. Use a different name for the
-o option.

      
After I run fsl_register_new, I should replace talairach.xfm in
        
recon-all
      
script to talairach.lta and run recon-all. Do I need add -notalairach in
stage1?
        
Right, you can replace talariach.xfm with talairach.lta. Also specify
-notalairach when you run recon-all.

-Xiao

      


    





_______________________________________________
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