Hi,

On 20 Nov 2007, at 19:24, Darren G wrote:

> Hi
>
> I have 2 questions
> 1) I would appreciate confirmation that I understand how FSL  
> organizes its
> pre-stats analysis. Let's say I do a single subject analysis and  
> perform
> motion correction, slice timing correction, bet brain extraction,
> registration to an anatomical standard brain and then the stats.
>
> Data are in directory WM. The root directory for the analysis is  
> anal. So in
> directory WM I end up with anal.feat or anal.ica or whatever the  
> analysis is
>
> from what I can tell
> 4Ddata -> mcflirt -> prefiltered_func_data_mcf
> prefiltered_func_data_mcf -> slicetimer -> prefiltered_func_data_sl

nearly - that should read:   prefiltered_func_data_st

> prefiltered_func_data_sl  -> bet -> prefiltered_func_data_bet
> prefiltered_func_data_bet -> (threshold, mask, smooth) ->  
> filtered_func_data
>
> the filtered_func_data appears to be located in the anal.ica  
> directory.
> then it is bandpass filtered, etc. and all the prefiltered volumes  
> are then
> deleted

not quite - any volumes to be deleted are deleted before any of the  
other pre-stats processing.

> for registration the middle volume is extracted from the dataset and  
> named
> example_func
> example_func -> flirt -> example_func2highres
> example_func2highres -> flirt (+transforms of highres to standard) ->
> example_func2standard.

again - example_func is extracted before any other pre-stats processing.

> The transforms from this are then applied to filtered_func_data and  
> at some
> point a registered/normalized version of filtered_func_data is  
> placed in the
> reg_standard directory so we end up with reg_standard/ 
> filtered_func_data.nii

In FEAT, this isn't done - all the timeseries analysis is done in  
native resolution, and only when a group-level FEAT is run, the  
summary stats images are upsampled to standard space.

In MELODIC, if you are doing multi-subject ICA, then yes the  
timeseries data filtered_func_data does get upsampled once it's needed  
for the multi-subject ICA analysis.

> This leads me to my second question
> 2) Once I have done all this to a dataset if I want to perform  
> additional
> analyses can I just take the reg_standard/filtered_func_data.nii 4D  
> dataset
> and use that as the input to feat, melodic, etc.? I assume that if I  
> did so
> I would then select NONE for mcflirt, slicetimer, bet, and  
> registration?

If you are using the FEAT and MELODIC GUIs this might be messy. In  
FEAT you never need the standard space data anyway, so this wouldn't  
fit how FEAT works. In MELODIC, yes if you wanted to try different  
multi-subject ICA analyses then you could reuse the reg_standard/ 
filtered_func_data files, but I think you would need to call melodic  
from the command-line in that case, as the GUI at this point always  
assumes that you are starting with native space raw data.

Cheers.



>
>
> thanks,
> darren


---------------------------------------------------------------------------
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
---------------------------------------------------------------------------