Hi experts,
(Thanks Wolf, that script is a big help, though the paper is too hard for me...)
Since I don’t have very deep background of statistics, here I just need the confirmation of my conjunction procedure, and share what I’ve observed.
I did a higher-level group analysis containing several contrasts. Among them, I want to find the common activations in copes 7-10.
Below is my procedure:
fslmaths gfeat/cope7.feat/stats/zstat1 -min gfeat/cope8.feat/stats/zstat1 -min gfeat/cope9.feat/stats/zstat1 zstat_min_7to9
easythresh zstat_min_7to9 gfeat/cope7.feat/stats/zstat1 2.3 0.01 gfeat/cope7.feat/example_func conj_7to10
Then there appears a file called thresh_conj_7to10 (is this the conjunction result representing common activations?).
Am I correct?
Four small questions:
(1)I can’t execute easythresh_conj so I tried easy thresh. Is it correct?
(2)Do I have to specify –s here? I think it is better to specify but only for 1st level analysis, right?
(3)Do I have to specify <brain_mask>?
(4)The example_func is specified in cope7, but I think it is the same if I specify other cope (because in our analysis, subject's images had been transformed to standard space before group analysis). Right?
Meanwhile, I can’t help to compare the image from the Pre-threshold masking method in FEAT. In this analysis, in Data of FEAT gui, I first chose cope7 in group.gfeat, and in Pre-threshold masking, I select thresh_zstat of COPE2. Then, just repeated to perform 3 FEATs to get the overlap of the 4 copes images, called overlap_thresh_7to10.
Now, in fslview, I open the following 6 images to compare:
(1)gfeat/cope7.feat/stats/thresh_zstat1
(2)gfeat/cope8.feat/stats/thresh_zstat1
(3)gfeat/cope9.feat/stats/thresh_zstat1
(4)gfeat/cope10.feat/stats/thresh_zstat1
(5)thresh_conj_7to10
(6)overlap_thresh_7to10
I found some voxels are significant in each image in (1)-(4), but not in (5). On the other hand, voxels that are significant in (5) are not always significant in all the rest 4 images. In contrast, all the voxels activated in (6) are significant in (1)-(4), and if a voxel fails to survive anyone in (1)-(4), it will not appear in (6).
I believe (5) is the conjunction result I want but, since it represents “common activations.” Am I right?
Just feel it’s weird because it’s not always activated in each image in (1)-(4). Maybe someone can explain this for a person like me who has no deep background of statistics….
Thanks in advance☺
Mark
|