As a short follow-up on this, I just wanted to mention that I also fixed
spm_spm_bayes for SPM2. Users who want to do PPMs under SPM2 can get the
revised code from
ftp://ftp.fil.ion.ucl.ac.uk/spm/spm2_updates
Best wishes,
Klaas
At 17:17 18/05/2005, you wrote:
>Because SPM5b is just a beta version of SPM5, there are still a few bugs that
>we are working on ironing out. Remember, that those of you who are trying
>SPM5b are our beta testers, and we rely on you feeding bug reports back to
>us.
>
>There have been a couple of significant fixes made in the last few days day,
>so we advise anyone who is trying SPM5b to install the updates from:
>
> ftp://ftp.fil.ion.ucl.ac.uk/spm/spm5b_updates
>
>These will update your version to Revision 167. The changes we made are
>summarised by Updates_README.txt .
>
>The most significant changes are:
>
>SPM2 compatible images with .mat files were not being properly handled.
>spm_get_space was ignoring contents of .mat files for single volume files,
>whereas nifti() was using the contents of the .mat files to over-ride the
>sform and qform matrices. This has been fixed.
>
>spm_spm_Bayes.m: PPMs used to crash under MATLAB 6.5.0 R13 due to a bug with
>sparse matrices. The code now reverts to full matrices when this MATLAB
>version is detected. I successfully tested the revised code for 2nd level
>analyses, but using full matrices could cause memory problems with large data
>sets at the 1st level (the user is being warned in the command window about
>this).
>
>Other changes are mostly small ammendments to the documentation.
>
>Best regards,
>-John
|