The brains are our typical "LPS" axial. The neck goes to the bottom of the
brain stem. I made a new mask, which masked out more of the neck/brain stem
and it ran after that.
I've attached a screenshot of a scan, with the successful mask. Is this the
level I should be aiming for, or any other suggestions?
Thanks,
Chris
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Brian Patenaude
Sent: Wednesday, September 26, 2007 4:57 PM
To: [log in to unmask]
Subject: Re: [FSL] first_flirt
Does the brain within the image have an odd orientation? I'm just trying to
figure out why flirt would fail. Is there any strong bias field effects?
The ones that failed, did they have a significant amount of neck? If so, for
the mask used, does it extend far below the cerebellum (preferably not)?
Cheers,
Brian
> No, just 256x256x166 1x1x1mm axial.
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf Of Brian Patenaude
> Sent: Wednesday, September 26, 2007 4:35 PM
> To: [log in to unmask]
> Subject: Re: [FSL] first_flirt
>
> Do they have any strange orientation to begin with?
>
> Cheers,
>
> Brian
>
>> Yes the failure(s) have been at the first level. Sometimes the brain
>> ends upside-down and flipped, sometimes it appears zoomed in and
>> skewed.
>>
>>
>> -----Original Message-----
>> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>> Behalf Of Brian Patenaude
>> Sent: Wednesday, September 26, 2007 1:41 PM
>> To: [log in to unmask]
>> Subject: Re: [FSL] first_flirt
>>
>> Hi,
>>
>>
>>> Is the outcome of run_first_all supposed to be in standard mni
>>> space, or native subject space?
>>
>> All the segmentations should be in native space. The FIRST takes the
>> transformation matrix inputed, inverts it, then applies it to the
>> model prior to fitting.
>>
>>
>>> Also, any suggestions as what to do when first_flirt fails to
>>> register even after masking out the neck?
>>
>> Does the failure occur at the first stage (use -d option)?
>>
>> Cheers,
>>
>> Brian
|