Steve,
You're absolutely right that this step is quite redundant as far as the end
result goes.
My experience with registering these modalities is that infrequently flirt
crashes when it is faces with registering grossly different voxel
spacing/array sizes.
Now, this could have been a relict from the earler, less stable versions of
the flirt. So, normally I just run an extra step that just resamples it into
the proper spacings.
pk
----- Original Message -----
From: "Steve Smith" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Tuesday, March 18, 2008 12:36 PM
Subject: Re: [FSL] White matter hyperintensity and FLAIR processing
> Hi - I hate to disagree with PeterK (he has a much faster car than I do
> and a much bigger spear-gun) but I don't think there's any reason to
> upsample the FLAIR before feeding it into FLIRT. You can just register it
> directly to the MPRAGE.
>
> Cheers, Steve.
>
>
>
> On 18 Mar 2008, at 17:26, Agnieszka Burzynska wrote:
>
>> Hi,
>> My first plan was to do:
>> 1)FLAIR-> resample to 1mm isovoxel-> to MPRAGE
>> 2) MPRAGE->MNI
>> 3) FLAIR->step 2 displacement field->to MNI space
>>
>> However, the resampled FLAIR looks, well, ...like FLAIR resampled from
>> 5mm
>> to 1mm :)
>>
>> Do you have any ideas and experience when can I do the resampling the
>> latest?
>> Is it better to use 1x1x1mm T1 image but from different scan or worse
>> resolution scan from the same session? Does it make any difference for
>> the
>> FLIRT functions (if it needs to rotate the brain more, etc.)?
>>
>> Sorry to trouble you, but I am new to registration and dont have the
>> "feeling" yet what is the best.
>>
>> Thanks a lot,
>> Aga
>>
>>
>>
>> On 18.03.2008 17:02 Uhr, "Minjie Wu" <[log in to unmask]> wrote:
>>
>>> Hi,
>>>
>>> I think the registration from FLAIR image into MNI space can be
>>> carried out in a hierarchical way
>>> in FSL:
>>> 1. FLAIR to MPRAGE, use rigid or affine registration , and produce
>>> affined-transformed flair
>>> 2. MPRAGE to MNI use nonlinear registration
>>> 3. displacement field from step 2 can used to further transform
>>> affined transformed flair into
>>> MNI.
>>>
>>> Also there is a method for automated WMH segmentation and localization.
>>> " A fully automated method for quantifying and localizing white matter
>>> hyperintensities on MR images."
>>> Psychiatry Research: Neuroimaging, Volume 148, Issue 2-3, Pages 133-142
>>>
>>> Good luck,
>>> Minjie Wu
>>>
>>> On Tue, Mar 18, 2008 at 11:45 AM, Peter Kochunov <[log in to unmask]
>>> > wrote:
>>>> I think you should have little troubles registering two datasets. One
>>>> step
>>>> you can do to insure stability is reslicing the FLAIR data into 1x1x1
>>>> resolution before regestering it to T1w. Also you might want to play
>>>> with
>>>> the cost function in flirt. The normalized correlation don't seem to
>>>> work as
>>>> well as say mutual info.
>>>>
>>>> pk
>>>> ----- Original Message -----
>>>> From: "Agnieszka Burzynska" <[log in to unmask]>
>>>> To: <[log in to unmask]>
>>>>
>>>>
>>>> Sent: Tuesday, March 18, 2008 10:39 AM
>>>> Subject: Re: [FSL] White matter hyperintensity and FLAIR processing
>>>>
>>>>
>>>>> Dear Peter,
>>>>> Thank you a lot for your email.
>>>>> I can do without the automation (will do manual identification of
>>>>> affected
>>>>> tissue), but I would like to have my FLAIR images in the MNI space.
>>>>> Do you
>>>>> know if anyone does such registration with 5mm slice thickness? I
>>>>> know you
>>>>> had 1mm isovoxel in your recent study and others with my parameters
>>>>> just
>>>>> do
>>>>> manual tracing on FLAIR image + multiply by slice gap to get the WMH
>>>>> volume.
>>>>>
>>>>> Do you know if I will get this registration well in FLIRT and my
>>>>> MPRAGE or
>>>>> shall I use a 1x1x1 T1-WI of these subjects (I may get them, but
>>>>> acquired
>>>>> in
>>>>> different MRI session- or will it cause even more problems with my
>>>>> FLAIR
>>>>> quality?). I am not too familiar with FLIRT yet.
>>>>>
>>>>> Thanks a lot,
>>>>> Aga
>>>>>
>>>>>
>>>>> On 18.03.2008 14:33 Uhr, "Peter Kochunov" <[log in to unmask]>
>>>>> wrote:
>>>>>
>>>>>> Aga,
>>>>>>
>>>>>> Unfortunatly, I don't think automated processing will be possible
>>>>>> with
>>>>>> this
>>>>>> dataset. The only workable approach to do automated HWM lesion
>>>>>> segmentation
>>>>>> is by the use of T2-maps produced by the multi-echo sequences.
>>>>>> pk
>>>>>> ----- Original Message -----
>>>>>> From: "Agnieszka Burzynska" <[log in to unmask]>
>>>>>> To: <[log in to unmask]>
>>>>>> Sent: Tuesday, March 18, 2008 7:09 AM
>>>>>> Subject: [FSL] White matter hyperintensity and FLAIR processing
>>>>>>
>>>>>>
>>>>>>> Dear all,
>>>>>>> I am searching for the most straightforward way to estimate the WMH
>>>>>>> load
>>>>>>> in
>>>>>>> my subjects.
>>>>>>> 1) Was there any follow-up on the message from 2002? (see below)
>>>>>>> 2) My FLAIR images have resolution 1.3x0.9x5mm and my MPRAGE
>>>>>>> 2x2x1mm.
>>>>>>> Does
>>>>>>> it mean I will have hard time in registering them to MNI?
>>>>>>>
>>>>>>> Thank you!
>>>>>>>
>>>>>>> Aga
>>>>>>>
>>>>>>>
>>>>>>> On Mon, 22 Jul 2002, Wei Wen wrote:
>>>>>>>
>>>>>>>> Dear Steve,
>>>>>>>>
>>>>>>>> It may be possible to extract (segment) white matter
>>>>>>>> hyperintensity
>>>>>>>> (FLAIR
>>>>>>>> sequence) using a combination of bet, fast etc. as discussed at
>>>>>>>> FSL
>>>>>>>> course
>>>>>>>> in Melbourne. Could I send you a couple of such images and see if
>>>>>>>> you
>>>>>>>> could
>>>>>>>> tweak the programs (or come up with some suggestions) for this
>>>>>>>> purpose?
>>>>>>>>
>>>>>>>> Many thanks,
>>>>>>>>
>>>>>>>> Wei Wen
>>>>>>>> NPI, Prince of Wales Hospital/UNSW
>>>>>>>> Australia
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>
>>>>
>>>
>>>
>>
>
>
> ---------------------------------------------------------------------------
> 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
> ---------------------------------------------------------------------------
>
|