On Saturday, April 26, 2003, at 06:41 AM, Peter Bannister wrote:
Sorry to bring this back up...I have been trying this, as you
suggested, with no success. The output files from mcflirt are identical
(compared with diff) to the input files. Here are the command lines I
have been using:
mcflirt -in AudMCor -out AudMCor_mcflirt20 -fov 20
I've alse tried it with FOVs of 10 and 2, and with and without the -2d
option. The program does not crash, but the output is identical to the
input.
These are 3-slice, 3mm thick/0 mm skip analyze 4d images. I know they
have some motion (not a huge amount, but some) by visual inspection.
Am I doing something wrong?
Thanks again for your help, and your patience.
-Richard
> Dear Richard
>
> Apologies for the confusion arising from your use of mcflirt. The next
> release includes a number of updates which we hope will make it easier
> to process 2d/ small FOV data.
>
> Currently it is not possible to perform a restricted DOF correction
> limited to in-plane movements only - mcflirt's approach (at present) is
> to pad out thin or single-slice data and then treat it as 3d.
>
> In the interim I would suggest running mcflirt as normal but with the
> -fov
> 20 option. This will pad the input data to a volume whose fov in the
> slice
> direction is 2cm which may yield more satisfactory results.
>
> In any event, as this approach is somehwat heuristic, I would strongly
> recommend inspecting the corrected data and playing around with the fov
> parameter until you are satisfied with the results. We have managed to
> "tune" the programme to handle motion-corrupted single-slice data
> generated at FMRIB in a satisfactory manner by experimenting with fov
> sizes.
>
> Hope this gives you something to work with
>
> regards,
> Peter
>
Richard Albistegui-DuBois
UCLA Brain Mapping
660 Charles Young Drive South, Room 237
(310) 206-4456 phone
(310) 794-7406 fax
[log in to unmask]
AIM: dubistegui
|