Thanks for the responses.
I had run across this older post:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;4a3b9626.1107
Is it still the case that you can't specify a lesion mask via the GUI or
is that an option now?
-Joel
On 05/09/2013 08:53 AM, Mark Jenkinson wrote:
> Hi,
>
> They cause huge problems if you don't use a lesion mask, but if you do use a lesion mask then it should be fine.
>
> All the best,
> Mark
>
> On 9 May 2013, at 14:39, Joel Bruss <[log in to unmask]> wrote:
>
>> Sorry to threadjack but I have a follow-up:
>>
>> Is there a consensus for subjects with lesions? Even if the image quality is good for the T1, do large areas of lesioned tissue tend to cause problems with FNIRT?
>>
>> -Joel
>>
>> On 05/09/2013 01:25 AM, Mark Jenkinson wrote:
>>> Dear Marco,
>>>
>>> FNIRT is always our recommendation unless the image quality is too poor to allow FNIRT to get a good registration. If, for any reason, you cannot get a good result with FNIRT, as assessed by manual inspection, then do not use a bad result, but try and improve it. If it is not possible to get a FNIRT result which is better than the FLIRT one then it would be OK to go with the FLIRT one, but I would be surprised if this were the case with good images. We always use FNIRT in our lab.
>>>
>>> All the best,
>>> Mark
>>>
>>>
>>> On 9 May 2013, at 04:46, Marco <[log in to unmask]>
>>> wrote:
>>>
>>>> Dear FSL experts,
>>>>
>>>> Is there a clear consensus on whether FNIRT should be preferred to FLIRT when performing normalization to MNI152 for group analyses of fMRI experiments?
>>>>
>>>> In discussing with various people, I noticed that some have a strong preference for FNIRT (and indeed the FNIRTed images do look very beautifully registered to the MNI space). However, I have also heard others expressing a preference for FLIRT, because they feel less comfortable in having the data more aggressively deformed than with a linear registration.
>>>>
>>>> What are your recommendations? Are there papers on the FNIRT vs FLIRT comparison that you would recommend?
>>>>
>>>> Thanks,
>>>>
>>>> Marco
|