John
Thanks for your reply. I think I've found the answer. Somehow, the
header information was changed for a portion of the files, where the
number of slices was reduced (from 36 to 20, which happens to be close
to the mean location of the origin in these files). When these files
were mixed with files with the correct Z dimensions in the header, then
ALL files (those with and those without the correct Z dimension) were
masked to the same limited number of slices.
Is this an expected feature of spm99?
Thanks
Vince
> From [log in to unmask] Mon Aug 14 11:27:56 2000
> Date: Mon, 14 Aug 2000 16:26:50 +0100 (BST)
> From: John Ashburner <[log in to unmask]>
> Subject: Re: normalization data deletion
> To: [log in to unmask], [log in to unmask]
> MIME-Version: 1.0
> Content-MD5: Z1PDpXtekXQIVFCPRtFIcw==
>
> Are the structural and statistical files been registered together in
> any way? You should make sure this is done first, using <Check Reg>
> to see if it has worked. If the images were initially registered, then
> processing done using a different package will probably not result in
> any .mat files being retained for the final statistical images.
>
> Best regards,
> -John
>
> | Using spm99, I have been deriving spatial normalization parameters
> | using a high-resolution T1 structural and then applying these
> | parameters to statistical files (derived from non-spm software)
> | using the "Write Normalised Only" option. I find that in some
> | cases, spm deletes all if the data that fell below the origin (-Z)
> | of the original, un-normalized files when writing the normalized files.
> |
> | I have yet to find the one variable that predicts this behavior.
> | After some investigation, I've found that it always happens when I
> | mix multiple files of different precision (8bit, 16bit and/or floating
> | point) in one Write Normalised session, but I also find it
> | occasionally with single files put by themselves in a separate directory.
> | It happens with both sync and nearest neighbor interpolation. It happens
> | with both integer and non-integer target pixel sizes. I'm not sure
> | what else to try.
> |
> | Has anyone else seen this? Any suggestions would be greatly appreciated.
>
>
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|