cluster report
debbie
Posted on 10/14/07 13:16:13
Number of posts: 1
hi there pls experts,
in my ER-MRI seed PLS
group analysis I found 1 significant LV. this LV has both positive and
negative voxel saliences (blue and yellow voxels). the max
bootstrap ratio in the positive direction is 12. in the negative
direction, it is 2.
my problem is that PLS wouldn't let me set a bootstrap ratio threshold larger
than 2 (the negative ratio, giving an error 'invalid threshold' message).
how can i set a larger threshold, in order to be able to isolate clusters of salient voxels in the positive direction?
if
i cannot set a larger threshold, is there another way to receive a
cluster report that would use a higher (say, 5 instead of 2)
thanks!
debbie
Untitled Post
I'm Online
jshen
Posted on 10/14/07 16:29:23
Number of posts: 291
> in my ER-MRI seed PLS
> group analysis I found 1 significant LV. this LV has both positive and
> negative voxel saliences (blue and yellow voxels). the max
> bootstrap ratio in the positive direction is 12. in the negative
> direction, it is 2.
> my problem is that PLS wouldn't let me set a bootstrap ratio threshold larger
> than 2 (the negative ratio, giving an error 'invalid threshold' message).
> how can i set a larger threshold, in order to be able to isolate
> clusters of salient voxels in the positive direction?
BS ratio reflects the reliability of the voxel activations. We only want to set 1 threshold to exclude those (false) activations in which the voxel values are close to 0, and keep both positive BS ratio and negative BS ratio. By default, we set 95 percentile of the BS ratio data as that threshold value, and include both positive and negative BS ratio beyond that threshold ( only exclude |BSRatio| < threshold ). Therefore, I suggest that you keep the active voxel with BSRatio = -2. Otherwise, the threshold you entered (e.g. 5) is not valid.
> if i cannot set a larger threshold, is there another way to receive a
> cluster report that would use a higher (say, 5 instead of 2)
No.