Hi HIma and Steve,
Thanks for your replies. I'm not using it to model behavioural data,
at this stage, I was using it to investigate between-group
differences in FA. I used the conventional randomise and whilst
there were a number of significant voxels on the uncorrected images
(p<0.05), none survived the corrected comparison. I've also tried
different cluster 'c' values (in randomise), to no avail, even though
on the uncorrected image there are a couple of clusters of voxels.
So I thought I'd try TFCE.
Thanks
Suzanna
On 19 Feb 2008, at 04:39, Steve Smith wrote:
> Hi - Hima's right, the TFCE option is a lot slower - but then the
> best things in life are worth waiting for.....
>
> If your model already contains non-zero-mean EVS (e.g. explicitly
> models the groups' means) then you _don't_ need the -D option as
> you want to leave the mean in the data.
>
> I would recommend reducing the -n option to say 500 to get a feel
> for whether the results are interesting and whether it's worth
> waiting for the full 5000 permutations analysis.
>
> We're working on speeding up the TFCE option - hopefully in the
> next release it will be faster.
>
> Cheers, Steve.
>
>
>
>
> On 19 Feb 2008, at 03:39, Himachandra Chebrolu wrote:
>
>> Suzanna-
>>
>> I have used the same command to do my analysis. It has not
>> produced any results as it's taking very long than without the -T
>> optin in there. What you have on the command line looks fine to
>> me. However I would suggest using the -D option to demean the
>> data, use -V for verbal info, -v for variance smmothing. These
>> options are not a must but my personal preferences and others
>> doing the same stuff can shed more light on other options.
>>
>> You do have to specify "-n 5000" to do 5000 permutations as the
>> default is set to 5000. Howver if you want to increase/decrease
>> the number of permutations use the -n option.Hope this helps.
>>
>> Cheers,
>> Hima
>>
>> P.S: My personal expereince has been that with the -T option each
>> contrast is taking 3 days to finish 5000 permutations. Without the
>> -T option I had each contrast done in 10 hours. I have 100
>> subjects in my analysis. Let me please know if you have similar
>> time consumption.
>>
>> Suzanna Laycock <[log in to unmask]> wrote:
>> Hi,
>>
>> Would someone mind stating the exact command line for running the
>> TFCE analysis.
>>
>> I'm using the following command:
>>
>> $ randomise -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -
>> d design.mat -t design.con -n 5000 -T
>>
>> Do you need to state the number of permutations, and should that be
>> set at 5000?
>>
>> Thanks
>> Suzanna
>>
>>
>> Looking for last minute shopping deals? Find them fast with Yahoo!
>> Search.
>
>
> ----------------------------------------------------------------------
> -----
> 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
> ----------------------------------------------------------------------
> -----
|