Dr. Wilke,
One of your suggestions is to make sure all images have values in all
voxels. At what point in the preprocessing stream should one check for this,
how do you check and how do you correct if missing values are found?
Thank-you for any suggestions/references,
Julie
Julie E. McEntee, MA, CCRP
Senior Research Program Coordinator
Department of Psychiatry- Neuroimaging
Johns Hopkins University School of Medicine
600 N. Wolfe St./Phipps 300
Baltimore, MD 21287
Phone: 410-502-0468
Fax: 410-614-3676
On 12/10/09 5:06 AM, "Marko Wilke" <[log in to unmask]> wrote:
> Hello,
>
>> for a couple of my datasets the normalisation procedure results in bad
>> fits (see attached image for an example).
>
> Without knowing more about your exact procedure, your spm version, your
> sequence and some other points, it is hard to say exactly what went
> wrong. However, there may be some generic points that might help:
>
> - try using the anatomical image for normalization
> - try using unified segmentation for normalization
> - try using the mean EPI for normalization
> - try increasing the bias correction
> - try coregistering the image-to-be-moved to the template image instead
> of manually shifting it around
> - make sure all images have values in all voxels (the missing parts look
> like not all images had valid voxel values there, and one image is
> enough to screw a whole series)
>
> Other than that, no brilliant ideas.
>
> Hope this helps,
> Marko
|