 |
 |
Hi Anderson, Thanks! I have put the new stats file snapshot into dropbox (and moved all files into FSL/OLD). Since I used the logp condition it means that p< 0.05 to p< 0.001 should correspond to a p range of 0.95 to .999. Converting this into -logp means that my range will be .0000435 (.999) to 0223 (.95) in terms of values for the overlay, as seen in the stats screen shot and uploaded files for left hemisphere when using the following command: palm -i bh.thickness.10mm.mgz -d design.mat -t design.con -o bh.thickness -n 500 -approx tail -corrcon -s bh.white -T -tfce2D -logp -m bh.mask.mgz -nouncorrected I get a large amount of speckled results in areas that DO make sense. I think for the surface data I was going to rerun it with 15mm as opposed to 10mm as this may help reduce the amount of small clusters that are next to each other. Do you see any other changes you would make to the command above prior to my rerunning. For instance are there any other approximations or acceleration methods that you think may be beneficial? Would 500 simulations be sufficient using this method for publications or do you recommend using a larger number? Best, Ajay