Hi Ai,
> I check the .mat file, and the result is following.
>
> Edwardai$ cat dti_FA_flirt.mat
> 1.034498772 0.1789141133 -0.1023610637 -57.65311518
> -0.2155201898 1.069033547 -0.1687029893 18.25888102
> 0.08074783853 0.1553506599 1.024006642 -3.616879374
> 0 0 0 1
>
> Is this .mat file good or bad?
One cannot really judge that from the .mat file itself. You need to apply it to your image (using -applyxfm) to see how well it has been registered to your -ref).
Jesper
>
> 在 2012-11-21,下午4:12,Jesper Andersson <[log in to unmask]> 写道:
>
>> Dear Ai,
>>
>>> I met some problems about FA registration, the following is my protocol.
>>>
>>> 1) My DTI image voxel size is 2*2*2mm.
>>> 2)First I got FA map using FDT in FSL.
>>> 3) Then registering FA map to standard FMRIB58_1mm using FLIRT in FSL and got .mat file.
>>> 4) Last using following command line in terminal to FNIRT.
>>>
>>> fnirt --in=dti_FA_flirt.nii.gz --aff=dti_FA_flirt.mat --config=/usr/local/fsl/etc/flirtsch/FA_2_FMRIB58_1mm.cnf --cout=coef_dti_FA_to_FMRIB58 --iout=dti_FA_fnirt --jout=jac_dti_FA_to_FMRIB58 --jacrange=0.01,100
>>>
>>> But the results is terrible(the following image), where did I go wrong?
>>
>> I think your command looks fine. Often when fnirt fails spectacularly, as in your case, it is because it has been given a poor initial start by the --aff transform. I suggest you first check how good dti_FA_flirt.mat is. If that looks good you can upload your data to us (through our uploading service) and we'll have a look at them for you.
>>
>> Jesper
>>
>>
>>> <屏幕快照 2012-11-21 下午2.19.50.png>
>
|