HI - I think that looks sensible - if you want to cross-check you should be able to output an f-test and compare that too. Cheers. On 5 Nov 2010, at 18:58, Filippo Migliorati wrote: > Hallo FSL experts, > I'm sorry for the newbie question. I would like to know how to compute the R-squared to determine the proportion of variance that a glm model explains. > I have fitted to some resting-state data a model made of 19 EVs (18 of them are of no-interest) using fsl_glm, demeaning both the data and the design matrix (I haven't used the options to normalize them to unit variance). > Since R-squared can be calculated as 1-(residual_var/total_var), I have used fslmaths as follows: > > fslmaths filtered_functional_data -mul ROI_mask -Tstd -sqr TotVar -----> to compute the total variance of the signal in a given ROI > fslmaths fsl_glm_results_sigsq -div TotVar -sub 1 -abs R-squared > > Is that correct? > Thanks, > Filippo Migliorati --------------------------------------------------------------------------- Stephen M. Smith, Professor of Biomedical Engineering Associate Director, Oxford University FMRIB Centre FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK +44 (0) 1865 222726 (fax 222717) [log in to unmask] http://www.fmrib.ox.ac.uk/~steve ---------------------------------------------------------------------------