Dear Mark,
Thank you very much! FSL is really a very strong image post process tool. I love the tool more and more!
Thanks all your hard work!
在 2012-11-26，上午6:11，Mark Jenkinson <[log in to unmask]> 写道：
> Hi,
>
> To add to this, I believe that you have both a FLAIR and an MPRAGE image of your subject (based on one of the emails you sent a while back). In this situation you should do the following:
>
> - register the FLAIR to the MPRAGE using FLIRT with 6 DOF
> (use brain extraction if it works well, but it is not compulsory)
>
> - register the MPRAGE to the MNI 2mm image with FLIRT (12 DOF)
>
> - register the MPRAGE to the MNI 2mm image with FNIRT (initialise it with the FLIRT matrix from the step before)
>
> - combine the FLAIR2MPRAGE and the MPRAGE2MNI transformations in applywarp to go from FLAIR 2 MNI
> (using --premat for the FLAIR2MPRAGE matrix ; and you can use any resolution of the MNI in this call)
>
> - to go from MNI to FLAIR, you will need to run "invwarp" to get the MNI2MPRAGE warp, and then you also invert the matrix (using "convert_xfm" or "InvertXFM" GUI) and use the inverted warp and inverted matrix in an applywarp call (this time using --postmat for the inverted matrix)
>
> You might also like to look at the examples in: https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FNIRT/UserGuide#Example_uses
> as the procedure you are using here is the same as that used for registering functional images to the MNI152.
>
> You can then decide what space to create appropriate masks and then transform these masks (with applywarp) to the desired destination space.
>
> All the best,
> Mark
>
>
>
>
> On 25 Nov 2012, at 15:33, Jesper Andersson <[log in to unmask]> wrote:
>
>> Dear Ai,
>>
>>> I know it's a impossible mission for me to get a standard FLAIR image. As an unprofessional, I think this need making standard MRI scan for many normal people, and merge these data together. Finally get a standard image. Is this correct?
>>
>> That is correct. It is not an impossible task, but you are right that one needs some experience to do it.
>>
>>>
>>> I segmented white matter lesions(WMLs) in the 3D FLAIR image and I draw a brain anatomical region mask in the standard MNI152_1mm. I just want to add the mask of WMLs onto the brain region mask to calculate the volume of WMLs in the different brain anatomical region.
>>>
>>> FNIRT is not possible for FLAIR registration, but the results of FLIRT is good. Could I use FLIRT to do it?
>>
>> I think for this case FLIRT would give you sufficient accuracy.
>>
>> Jesper
>>
>>
>>
>>>
>>> Thanks again!
>>>
>>> 在 2012-11-25，下午10:17，Jesper Andersson <[log in to unmask]> 写道：
>>>
>>>> Dear Ai,
>>>>
>>>>> Dear FSL experts:
>>>>>
>>>>> Is 3D FLAIR image not suitable for FNIRT to MNI152 image? I tried many times, but found the distortion is very obvious. Just like the following(Sorry to attach the image, I do not know how to describe)
>>>>
>>>> Fnirt works by matching one image to another, and usually that other image is some "standard image". For fnirt to work it is important that the order of tissue type intensities are the same in the two images. In your FLAIR image the intensities are in the order CSF-white-grey whereas e.g. in the standard space T1 image it is CSF-grey-white.
>>>>
>>>> If you want to register your FLAIR data you need an image in standard space with the same order of intensities, and I am afraid that we don't supply one with the FSL distribution. So you would have to make one yourself, or get it from someone who has made one.
>>>>
>>>> Jesper
>>>>
>>>>
>>>>
>>>>>
>>>>> My 3D FLAIR data voxel size is 1*1*1mm^3. Fist register it to MNI152_T1_2mm using FLIRT. Then register it to MNI152_T1_2mm using FNIRT(default .cnf file). Finally resampling to MNI152_T1_1mm.
>>>>> <屏幕快照 2012-11-25 下午7.58.29.png>
>>>
>>
>
