Hi Ankit,
your tfce_corrp maps are actually maps representing *1-pval*, so if you want to look at p<0.01, you need to have your tfce_corrp >0.99.
So in fslview, you need to put the min at 0.99 and max say at 1 (though the max threshold does not cut your values, just saturates the colours of the look-up table).
Each value in the map corresponds to 1-p, so if you have a value of 0.466, then the corresponding p-value for this voxel is 0.534, which is then clearly not significant...
Cheers,
Gwenaëlle
--- En date de : Jeu 7.1.10, Ankit Master <[log in to unmask]> a écrit :
> De: Ankit Master <[log in to unmask]>
> Objet: [FSL] how threshold tbss_tfce_corrp_tstat image
> À: [log in to unmask]
> Date: Jeudi 7 Janvier 2010, 19h03
> Hello,
>
> I have referred to some of the previous threads on the
> similar topic but I
> wasnt really clear about certain things so I am just trying
> clear some of my
> doubts.
>
> I have an image which is corrected for family wise error
> name of the image -- tbss_tfce_corrp_tstat
>
> 1) I have issues with thresholding this image in fslview. I
> am trying to
> look for all the voxels with p<0.01. Clearly I need to
> threshold my image
> with a value of 0.99, so how do i set my min and max
> values
>
> min 0.99 max 1.0
> or
> min 0 to max 0.99 ?
>
> 2) Say I go ahead with the second option; min 0 and max
> 0.99, after
> thresholding my image under this range in fslview, I see a
> lot of voxels in
> my tbss_tfce_corrp_tstat image and when I try to look at
> the intensity of
> each of these voxels, I get to see a myraid range of
> intensity values (e..g.
> 0.466, 0.78, 0.0012, etc) . Do these intensity values
> correspond to the p
> value ? if not what are these intensity values ? how do i
> interpret them ?
>
> Kindly advice. Thank you in advance.
>
> Ankit Master
> Biomedical Engg
> Mayo Clinic.
>
|