Hi Donna,
No need to apologise, probably plenty of Patriot fans on the list who
are happy to have something to distract them today.
So - the context is FMRI voxelwise cross-subject analysis and the
right way to align them. Obvious options being linear (affine),
nonlinear and even more sophisticated options such as doing it on the
cortical surface with FreeSurfer. Of the first two options, affine
registration is still widely-used in FEAT; this is robust and is what
we've done for a long time. Some researchers have compared this with
other nonlinear registration options, and in some cases found
nonlinear to be less robust, particularly when the nonlinear
registration is applied straight to the FMRI data, rather than via a
nice quality structural image. However, we have now pretty much
finished developing FNIRT generic nonlinear registration (including
relative bias field modelling), which gives very very nice results
when applied to (e.g.) structural images or FA data. We have put this
into FEAT (in conjunction with an initial FLIRT-based linear
registration) and I guess will be recommending using this as the
default. This will be released in FSL 4.1 at some point hopefully not
too long away.
So, in summary, we would probably expect the default pipeline to move
to using nonlinear registration, but there are a lot of existing
studies where linear has been used perfectly well. Of course, none of
this even mentions other, more complicated issues, such as correcting
for the effects of structural changes across subjects (e.g. see Oakes'
NeuroImage paper) - there's still lots of research to be done in these
areas!
Cheers, Steve.
On 4 Feb 2008, at 15:48, Donna Dierker wrote:
> Steve and FSL users,
>
> I apologize for the fuss and confusion. I certainly should have
> thought to ask users to reply to me rather than the list. (I had
> forgotten that the list is configured to reply to the list, rather
> than to the poster. And I'm not recommending the list behavior be
> changed; I prefer it as it is.)
>
> The reviewer's comment was not about FSL flirt in particular;
> rather, he/she insinuated linear/affine volume-based registration
> (VBR) is obsolete (the context being ordinary fMRI analysis). (If
> the reviewer is right, the paper under review is less relevant.)
>
> While I certainly agree that nonlinear VBR has advantages in many
> contexts, I don't think affine/linear VBR is irrelevant. In fact, I
> believe affine VBR has distinct advantages over nonlinear methods in
> some contexts. This reviewer is clearly a big nonlinear VBR fan,
> who needs to be reminded how widely used affine VBR is. Steve's
> point is well-taken: I don't need a reply count to make my point.
>
> Apologies again,
>
> Donna
>
> On 02/04/2008 09:24 AM, Martin M Monti wrote:
>> Steve, sorry to suggest differently, but given that it's highly
>> likely that most people on this list use FLIRT, wouldn't it be more
>> productive if she actually posted what the uncertainty is about?
>> Quite simply, if it is a reasonable criticism/request then almost
>> everybody will be interested in what the reviewer has to say, the
>> more considering the fact that we may get similar feedback in
>> future reviews.
>>
>> Of course, if Donna feel it's a confidential matter -- perfectly
>> understandable -- or a not-too-reasonable comment, as it sometimes
>> happens, then never mind...
>>
>> just a suggestion though,
>>
>> all the best
>>
>> martin
>>
>> Steve Smith wrote:
>>> Hi - could I suggest that people email Donna Dierker directly,
>>> rather than generate too many emails in the inboxes of people on
>>> the list!
>>>
>>> Donna - not sure what the reviewers are asking about; FLIRT is
>>> certainly one of the most widely-used of the components in FSL. If
>>> you search in NeuroImage there's over 80 articles mentioning
>>> FLIRT, but I'm not sure if that's the kind of statistics that
>>> you're after.....
>>>
>>> Cheers, Steve.
>>>
>>>
>>>
>>> On 4 Feb 2008, at 15:00, David Gutman wrote:
>>>
>>>> I regularly "FLIRT" as well.
>>>>
>>>> DAG
>>>>
>>>>
>>>>
>>>> On Mon, Feb 4, 2008 at 9:42 AM, Donna Dierker <[log in to unmask]
>>>> > wrote:
>>>> I'm hoping to be inundated with replies, which will help me
>>>> address a
>>>> reviewer comment.
>>>>
>>>> Donna Dierker
>>>> Van Essen Lab
>>>>
>>>>
>>>>
>>>> --
>>>> David A Gutman, M.D. Ph.D.
>>>> Department of Psychiatry & Behavioral Sciences
>>>> Emory University School of Medicine
>>>
>>>
>>> ---------------------------------------------------------------------------
>>> 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
---------------------------------------------------------------------------
|