On 23 August 2012 09:59, Pavel Hok <[log in to unmask]> wrote:

Dear FSL Experts,

I encountered a crash in FEAT higher-level analysis after following the instructions from FSL FAQ No 49. "How do I run higher-level FEAT when some inputs are first-level and some are higher-level?".

Most of our subjects had two runs per condition, but some of them did not complete the second run. To overcome that I ran Fixed Effects Higher-level FEAT analysis on each subject separately to get their subject mean. Then I wanted to get simple group average, so according to the FAQ I used cope#.nii.gz images as input for the next higher-level stats. From the subjects with two runs, I used the #.gfeat/cope1.feat/stats/cope1.nii.gz, while for the subjects with only one run the cope1.nii.gz from the original first-level feat directory.

After running FLAME1 with simple group mean design, I noticed that something wrong was with the plots: all the subjects with averaged runs had the identical values for the best voxel, while those with only one run differed. When I ran the FLAME1+2, the script crashed with following error printed out repeatedly:

An exception has been thrown
Logic error:- detected by Newmat: index error: requested index = 11

MatrixType = Rect # Rows = 10; # Cols = 1
Trace: Mcmc_Mh::setup; Gsmanager::run.

During one of the conditions, all the subjects had both runs completed. In this case both FLAME1 and FLAME1+2 crashed with the following error:

An exception has been thrown
Runtime error:- detected by Newmat: matrix is singular

MatrixType = Crout # Rows = 1; # Cols = 1
Trace: Crout(lubksb); GeneralSolv; Gsmanager::flame_stage1_onvoxel; Gsmanager::flame_stage1; Gsmanager::run.

Any help would be appreciated!

Thanks!

Pavel Hok