Dear all,
I tried to run snpm3b with spm2 and Matlab 6.5.1 for a VBM analysis - comparison of 2 groups with 37 subjects in total.
I used "Multiple group one scan one subject" plugin and was able to go through setup and computation steps.
The problem emerges when I try to view results of cluster level analysis: the problem is Matlab could not load the SnPM_ST.img (in my case about 800 MB), and gave this error:
SnPM: snpm_pp
=======================================================================
Working on spatial extent...
Computing suprathreshold voxels...done
Loading & conditioning SupraThreshold statistics...??? Error using ==> load
Unable to read MAT file C:\MATLAB6p5p1\work\SnPM_ST.mat
File may be corrupt.
Error in ==> c:\spm2\toolbox\snpm3\snpm_pp.m
On line 659 ==> load(fullfile(CWD,'SnPM_ST'))
??? Error while evaluating uicontrol Callback.
This is not limited to just one analysis, it happened several time with different setups.
Have you got any idea what might result in such problem?
Just another question:
when setting parameters for the clusterwise analysis, a prompt "Cluster-def. thereshold (p<=0.01, t>=2.44) with preselected value 2.44 for my specific analysis appears. How it comes to such values? Is there any information that may help in picking the threshold for cluster-level VBM analysis? I thought it might be OK to use the threshold that corresponds to voxelwise p<=0.05 (that I can get from spm, when doing snpm without variance smoothing), since the correction for multiple comparison is done at the cluster level. Or am I wrong?
Thank you in advance
Tomas Kasparek
Dep. of Psychiatry, Masaryk University
Brno
the Czech Republic
|