Hi Jason, I've just checked this scenario with Tom;
So the statistic of interest does not, for its null, have a known
mean or symmetry. Hence the normal trick for testing a single group
with permutation testing (which is valid if the data is known to be
zero-mean and symmetric in the null case) doesn't apply, so I'm
afraid that permutation testing isn't valid in this case.
Sorry! Cheers, Steve.
On 12 Jul 2007, at 21:25, Jason Steffener wrote:
> I have taken three images per subject, with voxel values normally
> distributed. I then combined them by taking the magnitude across
> images. The result is an image with a square root of Chi-squared
> distribution.
>
> I have this for ten subjects and would like to know which voxels
> have values consistently (across subjects) large values.
>
> I would like to do a one group analysis but with these non-normally
> distributed images.
>
> Thank you,
> Jason.
> ----- Original Message ----
> From: Steve Smith <[log in to unmask]>
> To: [log in to unmask]
> Sent: Thursday, July 12, 2007 3:25:57 PM
> Subject: Re: [FSL] Randomise on a group of Non-normal data
>
> Hi,
>
> Sign flipping generates the null distribution for there being a non-
> zero mean effect, which doesn't match what you have I suspect. What
> do you mean by "significant" - what is your model?
>
> Cheers, Steve.
>
>
>
>
> On 12 Jul 2007, at 17:06, Jason Steffener wrote:
>
>> Hello all,
>> I have a single group of images, one per subject, where the voxel
>> values are distributed as square root of Chi-squared
>> distributions. I would like to perform a group level analysis on
>> this data to find those voxels that are consistently significant
>> across the group. I am wondering if the randomise program is
>> applicable. I know that for a single group of subjects the program
>> flips signs. So would this applicable for the distributions I have?
>> If not does anyone have any other suggestions.
>>
>> Thank you for any advice.
>>
>> Jason.
>
>
> ----------------------------------------------------------------------
> --
> ---
> Stephen M. Smith, Professor of Biomedical Engineering
> Associate Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ----------------------------------------------------------------------
> --
> ---
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|