Hi Richard,
I've had a look at your images and there is one significant problem.
Your whole functional image has only 48mm coverage in one direction
and hence is itself a partial FOV, not a whole brain image as we
recommend.
Consequently it is difficult to get great registrations. In future,
try to get full
brain coverage.
However, even so I managed to get what looked like a good registration
to me using:
flirt -in whole_func -ref 3dvolEQ_brain_007 -omat wh23d.mat -out wh23d
-dof 6 -searchrx -180 180 -searchry -180 180 -searchrz -180 180
(this is the same as 6dof, full search in FEAT)
As there is also distortion in the EPI images I couldn't tell how well
the
functional (3 slice) and "whole" brain EPI (16 slice) were registered,
although they looked ok and probably should be fine as long as there
wasn't much motion in the timeseries.
One thing I wasn't clear on was which registrations you were doing?
You said you did "timeseries--(3 DOF)--> anatomical--(12 DOF)-->volume"
but which "volume" is this?
If it is the individual anatomical T1 scan, then it is within the
individual subject and
you should be ok with 6 dof (or 7 if the scanner calibration might not
be constant).
If the "volume" is a standard space image, then you need another
registration step.
What you want is:
3 slice EPI --(3 dof)--> 16 slice EPI --(6 dof)--> individual
anatomical --(12 dof)--> standard
This is easy to set up in feat, and based on the data I think it should
work, although
not as well as if you had a full brain coverage EPI for the second
image.
If you are still unhappy with the registrations, then I would advise
creating a
weighting volume that is zero over the distorted areas, as these are
probably
the main cause of error, apart from the small FOV, in these
registrations.
Hope this helps.
All the best,
Mark
On Monday, July 28, 2003, at 06:52 pm, Richard Albistegui-DuBois wrote:
> Yep, we were using full search. I'm not sure where the registration
> report page is, sorry.
>
> The files should be available at the following:
>
> Volume: http://www.brainmapping.org/fsl/3dvolEQ_brain_007.img (and
> .hdr)
> 16-slice EPI: http://www.brainmapping.org/fsl/whole_func.img (and .hdr)
> Sample 3-slice volume from timeseries:
> http://www.brainmapping.org/fsl/example_func.img (and .hdr)
>
> Thanks much for your help. Please feel free to email me directly at
> [log in to unmask]
>
> On Saturday, July 26, 2003, at 03:48 AM, Stephen Smith wrote:
>
>> Hi - hard to say - looks like you are probably trying the right things
>> -
>> if you can make the data available on a website we'll take a look.
>> When
>> you look at the registration report page, which of the various steps
>> is
>> the problematic one? For the 16-slices EPI to anatomical flirting, did
>> you
>> turn on full search?
>>
>> Thanks, Steve.
>>
>>
>>
>> On Fri, 25 Jul 2003, Richard Albistegui-DuBois wrote:
>>
>>> I am trying to use FLIRT to align a very limited set of slices with a
>>> 3d T1 volume. I have tried it following the instructions for
>>> limited-z
>>> datasetson the website, as well as with more standard techniques, and
>>> they both give me the same incorrect results.
>>> What I have:
>>> A functional timeseries of three oblique (mostly coronal, rotated
>>> slightly toward axial) 3mm/0 skip slices
>>> An anatomical EPI set of 16 slices, including the three in the above
>>> set
>>> A T1 3d volume taken on a different day
>>>
>>> When I do FLIRT with timeseries--(3 DOF)-->anatomical--(12
>>> DOF)-->volume, the end result has the correct rotation, but is not in
>>> the correct place. Aligning the anatomical alone to the volume with
>>> 12
>>> DOF is similar--it is rotated correctly, but placed incorrectly. It's
>>> about 1-2 cm too anterior.
>>>
>>> I've also tried this with the ventricles weighted, with no change in
>>> results.
>>>
>>> Can anyone offer a suggestion?
>>>
>>> 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
>>>
>>
>> Stephen M. Smith MA DPhil CEng MIEE
>> Associate Director, FMRIB and Analysis Research Coordinator
>>
>> Oxford University Centre for Functional MRI of the Brain
>> John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
>> +44 (0) 1865 222726 (fax 222717)
>>
>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>>
>>
> 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
