Dear SPM (and Will)
I am helping someone with a Bayesian first level esimation and we have
run into an error when calculating the mask.
Failed 'Model estimation'
Index exceeds matrix dimensions.
In file "/master_raid/users/hlee/spm/spm8/spm_vb_neighbors.m" (v2887),
function "spm_vb_neighbors" at line 45.
In file "/master_raid/users/hlee/spm/spm8/spm_spm_vb.m" (v3055),
function "spm_spm_vb" at line 642.
In file "/master_raid/users/hlee/spm/spm8/config/spm_run_fmri_est.m"
(v2928), function "spm_run_fmri_est" at line 297.
The following modules did not run:
Failed: Model estimation
The problem occurs because in this dataset thresholding the mask
results in their being only 1 voxel in the last slice, that survives
the threshold. Looking at spm_spm_vb.m at line 636, if there is 1
voxel then CrS = sum(Cm); is 1 so the if condition gets executed.
However, only 1 voxel gets passed to spm_vb_neighbors which then fails
at line 45 because DIM is only 1 dimensional = 1 voxel rather than 2
dimensional. So the spm_vb_neighbors code needs to probably exclude
the case where there is only 1 voxel.
Any help would be appreciated.