 |
 |
Hi Marco, Please, see below: > I'm running a seed-based resting state analysis on fMRI data. Now, I'd > like to make some cluster corrections on my data, but I have some issues > about FSL cluster command and the interpretation of results. first of all, > with the smoothest command, I worked out the smoothness and I know the FWHM > of the resels. My questions are: > > 1) I have used a z-fisher transformation on my data to stabilize the > variance of my t-maps of correlation between the seed and the rest of the > brain. should I use -i,--in,-z or --zstat for my filename of input volume? These are all the same. Use any. > and should I use -t,--thresh or--zthresh for thresholding the input > volume? I don't really understand the differences between these options. > These are also all the same. Use any. 2) I don't understand how this cluster correction is implemented in FSL. I > report a statement from > http://www.fil.ion.ucl.ac.uk/spm/doc/books/hbf2/pdfs/Ch14.pdf > > "If we have a two dimensional image with 100 resels, then the probability > of getting one or more blobs where Z is greater than 3.8, is 0.049. We can > use this for thresholding. Let x be the Z score threshold that gives an > E[EC] of 0.05. If we threshold our image at x, we can conclude that any > blobs that remain have a probability of less than or equal to 0.05 that > they have occurred by chance." > > Then, why FSL -t,--thresh or--zthresh allows to specify any threshold with > any -p,? what is the purpose of -t,--thresh or--zthresh, if what I really > need is only -p? aren't they simply related by a function? > The excerpt refers to voxel-level inference using the random field theory, but here it's about cluster-level inference. That requires a cluster-forming threshold (-t); once the clusters have been formed by thresholding, their p-values are calculated using --dlh and --volume. The --pthresh is the value to threshold the clusters after their p-values have been calculated. All the best, Anderson