Hello everyone,
I encounter a problem with FSL's "film_gls" (v5.0.9) when performing first level GLM analysis in surface mode. Data were projected to fsaverage5 space (10242 vertices).
FILM command:
film_gls --in=test.lh.func.gii --in2=lh.midthickness.surf.gii --mode=surface --rn=STATS --con=design.con --pd=design.mat
The following error occurs:
paradigm.getDesignMatrix().Nrows()=413
paradigm.getDesignMatrix().Ncols()=6
sizeTS=413
numTS=10242
Calculating residuals...
Completed
Estimating residual autocorrelation...
Calculating raw AutoCorrs... Completed
Tukey M = 20
Tukey estimates... Completed
Completed
Prewhitening and Computing PEs...
Percentage done:
An exception has been thrown
Runtime error:- detected by Newmat: process fails to converge
MatrixType = Rect # Rows = 6; # Cols = 6
Trace: SVD; pinv.
When I add an option for autocorrelation estimation "--ac", the following error occurs:
paradigm.getDesignMatrix().Nrows()=413
paradigm.getDesignMatrix().Ncols()=1
sizeTS=413
numTS=10242
Calculating residuals...
An exception has been thrown
Logic error:- detected by Newmat: incompatible dimensions
MatrixType = Rect # Rows = 1; # Cols = 6
MatrixType = Rect # Rows = 1; # Cols = 1
Trace: MatrixMult.
The command works perfectly fine with "--noest" flag and in volumetric data. The design matrix has 413 rows (timepoints) and 6 columns (predictors). It was created with FSL FEAT/Glm Setup.
It appears to me that FILM does not even read the surface file -- I receive the same error ("incompatible dimensions") even when no input2 ("--in2") is specified.
Many thanks in advance for any help.
Best regards, Blazej
|