Hello,
I have used FEAT recently to do pre-stats only and then in second run to do stats+post-stats. I have then used afni to look at the result. What I have found, and what is somewhat confusing, is that the filtered_func_data.img in the stats+post-stats feat directory is different from filtered_func_data.img in the pre-stats feat directory. In fact, it is the same as the raw data. (I have run the stats+post-stats analysis by selecting the feat pre-stats directory instead of the raw dataset). The confusing part is that I have expected the two to be the same. Also, when I look at the tstats.img as an overlay in afni, there are colored pixels outside of the brain.
Then, I used afni to look at some other data I have analyzed previously with feat4. It was basically the same analysis (slightly different model), same data, only that time I did the full analysis - only one fsl run. If I look at that data in afni the same way as above, filtered_func_data.img is obvioulsly the one that was brain-stripped and smoothed, and the overlay does not show any colored pixels outside of the brain.
Now, all this makes me wander, what did feat use as input in the first case described above?
One more thing - might be related, might not be. When I did the pre-stats, the *.feat directories were created in the same directory where the raw images lived. Then I moved those *.feat directories into a separate folder, wanting to keep the raw data separate from the analysis data. I then proceeded to run the stats+post-stats, but feat was quiting with an error saying that the original data has 0 time points. I had to move the analysis folders back with the raw data in order for this to work. Why is that? Could it be that stats analysis used raw data and not pre-stats analyzed data?
Thank you as always...
zrinka
|