Dear SPMers,
I am having trouble treating single-subject, two condition studies under
SPM. I have tried using the PET_motor scans available on the SPM FTP
site to test the software. Could someone tell me if I'm doing something
wrong:
1. I select the PET-SPECT module from SPM, the the Statistics
function.
2. The data I give for the study is:
Design type: Single subject, replication of conditions
# of conditions: 2
No Global normalisation (the images are already normalised)
No Grand Mean Scaling
Grey Matter Threshold: 0.8
# of contrasts: 1
Contrast 1: [-1 1] (I want to find activation areas against
the baseline) - Should I type "[-1 1]" or "-1 1"?
SPM produces a design matrix which seems correct, but then outputs the
following error on the console:
??? Error using ==> *
Inner matrix dimensions must agree.
Error in ==> c:\program files\matlab\toolbox\local\spmnt\spm_spm.m
On line 127 ==> C = K*C;
I introduced a breakpoint, and verified that matrix C (covariate
subpartition of design matrix), is empty (correct, as I have no
covariates), while K is a sparse 12x12 identity matrix (the Toeplitz
convolution matrix of SIGMA, which is 0), justifying the impossible
multiplication error reported by Matlab. Where is my error? should I
introduce confounding covariates? How many?
I should just add that I already tried this study using winSPM, without
covariates, obtaining the same design matrix, and correctly obtaining
results considered acceptable. I am now using SPM for NT, under Matlab
5.
Thanks for your time. I greatly apreciate any help,
--
Sergio Carvalho
---------------
[log in to unmask]
If at first you don't succeed, skydiving is not for you
|