Thanks for your reply. I'm a little perplex, having read the manuals and
looking at the output. You mention I should BET the nodif_brain_mask? I
thought the nodif_brain_mask are obtained from running BET. Do you mean I
have to run BET on the nodif_brain_mask again (i.e twice)?
I tried doing Eddycorrect on the original 4D full DTI data. Then I extract
the 1st B0 image and ran BET on it to get a nodif_brain_mask. I ran DTIFIT
using this mask, the eddycorrect 4D DTI data ( with skull on) and the
bvecs and bvals. However, the results FA maps has 0 intensity voxels. i'm
not sure why that would be the case. The original data is from a 3T
siemens scanner with 5B0 and 30 dir, B1000. I'm not sure how the original
dicom data can be assessed.
I understand from the manual that the nodif_brain_mask used in DTIFIT and
bedpost are gotten from running BET on the eddycorrect DTI data or from
the 1st B0. I'd like to ask if it would be ok to BET the full DTI data
using the -F command, as it seems the only way I can get the FA map output
with no 0 voxels. (I read from the manual this is usually use for FMRI
data and the brain_mask from running -F would result in a slightly dilated
brain_mask).
The other question I'd like to ask is if it would make much difference
doing Eddy correction on the original DTI data compared to doing eddy
correction on the 4D nodif_brain output which has no skull. I notice the
FA values would be slightly lower if the data is processed in the latter
way before feeding into DTIFIT, and wonder which FA map output would be
more accurate to use?
Many Thanks for your kind advice.
Siewmin
Hi,
>
> You would not normally BET the full DTI data in this way. You should
> normally BET the nodif_brain_mask (see FDT and TBSS manuals) and then
> dtifit uses this mask.
>
> Steve.
>
>
> On 10 Mar 2009, at 10:12, Siew-Min Gan wrote:
>
>> Hi Steve,
>> yeah, I think its unusual too! I think I may have come up with
>> some "solutions" but not sure if and why it may work.
>>
>> The dicom DTI data (30 dir,5 B0) is coverted to ANALYZE and then to
>> NIFTI
>> I then merge the 35 nifti files into a 4D vol file.
>>
>> These are the procedures I did where the output FA map has 0
>> intensity:
>>
>> i. Eddycorrect the 4D vol DTI file (output data.nii.gz)
>> ii. BET the data.nii.gz file ( using standard brain extraction)
>> (output
>> nodif_brain_mask)
>> iii. DTIFIT in the directory containing (data.nii.gz with skull on),
>> nodif_brain_mask, bvecs and bvals)
>>
>> I made some changes to the method above which seemed to derive FA maps
>> with no 0 intensity voxels. However, the FA values differ in each
>> voxels
>> from the two methods I tried.
>> Please let me know which way you think may work better.
>>
>> If I repeat the above steps in order, but when running bet, I choose
>> "Apply to 4D FMRI data" with -F -f 0.2 -g 0 instead of "running bet
>> using
>> standard brain extraction" with -f 0.5 -g 0, the output FA maps from
>> DTIFIT do not have 0 intensity voxels.
>> Alternatively, if I do BET as step i with -F -f 0.2 -g 0, and then
>> eddycorrect the 4D deskulled dti brain output from BET, the output
>> FA maps
>> also do not have 0 intensity voxels.
>>
>> The FA values are different in the FA maps derived from the 3
>> different ways.
>> i) The 1st FA map from eddycorrect_thenbet_DTIFIT procedure has very
>> little white rim, but has 0 intensity voxels in brain white matter
>>
>> ii) The FA map from eddycorrect_thenbet(-F)_DTIFIT has no 0 intensity
>> voxels, but a denser white rim voxels as well as brain white matter
>> voxels
>> with intensity values > 1. These brain matter voxels with intensity >1
>> coincides with the 0 intensity voxels in the 1st FA map voxels above.
>>
>> iii) The FA maps from bet(-F)_theneddycorrect_DTIFIT has no 0
>> intensity
>> voxels, it has a dense white rim voxels but much less brain white
>> matter
>> voxels with intensity values >1.
>> White matter Voxels with 0 intensity on the 1st FA map or intensity
>> >1 on
>> the 2nd FA map above, would have a FA values of between 0 and 1 in
>> the 3rd
>> FA map. White matter Voxels with intensity>1 on the first two FA
>> maps (i.e
>> 1.03) would have an FA values of 0.33 on this FA map. For most
>> voxels of
>> 0<intensity<1, the FA values in the 1st and 2nd FA map are identical
>> but
>> it is lower in the 3rd FA map.
>>
>> May I ask if the registration in eddycorrect on deskulled brain (i.e
>> output from BET) works better than on brain_with skull on? I'm also
>> curious why would the FA values in the FA maps output be different
>> in the
>> 3 above methods, in terms of voxels with 0 intensity corresponding to
>> those >1, and the absolute values between the 2nd and 3rd FA maps.
>>
>> This would help me decide whether to use the 2nd or 3rd method to
>> process
>> the data.
>>
>> Many thanks, and apologies for the long email!
>>
>> Siewmin
>>
>>
>>
>>
>>
>> You mean you're getting zeroes in the FA output from dtifit _before_
>>> any further processing (e.g. from BET or tbss_1_preproc) ? I think
>>> this would be unusual and you should carefully check the data that
>>> was
>>> input to dtifit.
>>>
>>>
>>>
>>>
>>> On 8 Mar 2009, at 05:42, Siewmin Gan wrote:
>>>
>>>> Hi all,
>>>> I've noticed some pixels with 0 intensities in the white
>>>> matter of my FA map output
>>>> from DTIFIT, and bloches around the white matter regions (with 0
>>>> intensities) in the all_FA
>>>> map from tbss_3_postreg -S output. These occur near the pixels/
>>>> regions around the
>>>> temporal stem white matter.
>>>> I've checked my bvals, and bvecs (30 directions) which looks ok,and
>>>> would like to ask if it's
>>>> "normal" to have some pixels with 0 intensities in FA maps?
>>>> However, the bloches of 0 intensity in the all_FA map in tbss are
>>>> almost 10-15 pixels big.
>>>> Because they occur around the white matter, some bits of the FA
>>>> skeleton passing by the
>>>> temporal stem is broken up. I'm not sure why this problem exist.
>>>>
>>>> Much appreciated for some kind advice.
>>>>
>>>> Thanks.
>>>>
>>>> Siewmin
>>>>
>>>
>>>
>>> ---------------------------------------------------------------------------
>>> 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
>>> ---------------------------------------------------------------------------
>>>
>>
>
>
> ---------------------------------------------------------------------------
> 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
> ---------------------------------------------------------------------------
>
|