When we looked at this we looked at dispersion estimates and number of 2nd
and 3rd fibers modeled. We found concatenation to slightly outperform
averaging (though computational time will be substantially longer). One
isn¹t going very wrong either way though.
Peace,
Matt.
On 2/24/15, 8:57 AM, "M Barendse" <[log in to unmask]> wrote:
>Thank you for your reply.
>One more follow up question to all DTI experts:
>It has been suggested that averaging the two volumes is better for SNR
>than concatenating them:
>https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1105&L=FSL&D=0&P=30679
>However, the opposite has also been suggested:
>https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;bde7018a.1407
>How do I determine what the best approach is for my analyses?
>Kind regards,
>Marjolein
>
>On Fri, 20 Feb 2015 16:15:09 +0000, Vasilis M. Karlaftis
><[log in to unmask]> wrote:
>
>>Dear Janna Marie,
>>
>>I am not an expert but these processing steps seem fine to me; I am also
>>using them for my analysis. I think that in Stam's link
>><http://fsl.fmrib.ox.ac.uk/fslcourse/lectures/practicals/fdt1/index.html#
>>eddy>
>>it
>>says exactly what you have to do for 2 blips.
>>Just be sure that your index and acqparams files have the correct values,
>>e.g. all A volumes have 1 and P volumes have 2 in index; and your
>>acqparams
>>file has 0 -1 0 "dwell time" in first row and 0 1 0 "dwell time" in
>>second
>>row.
>>
>>I would also suggest to include the --sse option when running dtifit, if
>>you aim to evaluate the DT model at some point.
>>
>>Best regards,
>>Vasilis
>>
>>On 20 February 2015 at 12:37, Janna Marie Hoogendam
>><[log in to unmask]
>>> wrote:
>>
>>> Dear Stam,
>>>
>>> Thanks for your reaction. We have read the documentation, but did not
>>>find
>>> information about combining two maps, one acquired in A-P, the other
>>>in
>>> P-A direction.
>>> We have done the following:
>>> - combined the bvecs and the bval files (for example: paste -d" "
>>> 2001210/*WIP*_A*.bval 2001210/*WIP*_P*.bval > 2001210/DTI_A_P.bval)
>>>for the
>>> two directions
>>> - used fslmerge to combine the nii files
>>> - ran topup
>>> - performed eddy using these command:
>>> fsl_sub eddy --imain=DTI_A_P_1sliceless --mask=DTI_b0_brain_mask
>>> --index=index.txt --acqp=acqparams.txt --bvecs=DTI_A_P.bvec
>>> --bvals=DTI_A_P.bval --fwhm=0 ‹ topup=topup_DTI_b0s
>>> --flm=quadratic --out=eddycor_DTI
>>> - performed dtifit
>>> fsl_sub dtifit -k eddycor_DTI -m DTI_b0_brain_mask -r DTI_A_P.bvec -b
>>> DTI_A_P.bval -o fitted_DTI
>>>
>>> Is this the correct way to combine the AP and PA maps?
>>>
>>> Thanks a lot,
>>>
>>> Marjolein and Janna Marie
>>>
>>>
>>>
>>> > Op 20 feb. 2015, om 13:21 heeft Stamatios Sotiropoulos <
>>> [log in to unmask]> het volgende geschreven:
>>> >
>>> > Dear Janna
>>> >
>>> > You first need to run topup and then eddy, to correct for
>>> susceptibility-induced and then eddy current-induced distortions.
>>>There is
>>> extensive documentation on how to do that on the FSL webpage, but also
>>> practicals from the FSL course that walk you through both steps. See
>>>for
>>> instance:
>>> >
>>> >
>>>http://fsl.fmrib.ox.ac.uk/fslcourse/lectures/practicals/fdt1/index.html
>>> >
>>> > Hope this helps
>>> > Stam
>>> >
>>> >
>>> > On 20 Feb 2015, at 12:08, Janna Marie Hoogendam
>>><[log in to unmask]>
>>> wrote:
>>> >
>>> >> Dear all,
>>> >>
>>> >> Maybe this question escaped your attention, but we (Marjolein and
>>> myself) would be very pleased to hear your reaction to the issue raised
>>> below.
>>> >> Thanks in advance,
>>> >> best wishes,
>>> >>
>>> >> Janna Marie
>>> >>
>>> >>> Op 11 feb. 2015, om 10:33 heeft M Barendse
>>><[log in to unmask]>
>>> het volgende geschreven:
>>> >>>
>>> >>> Dear FSL users,
>>> >>> We have acquired DWI data with a Philips 3T scanner. I would like
>>>to
>>> run eddy and DTIFIT on these data. Our scans are all in Œfold-over
>>> direction = ³AP²¹, and we have 1 b0 and 30 dwi¹s with Œfat shift
>>>direction
>>> = ³A² as well as 1 b0 and 30 dwi¹s with Œfat shift direction = ³P²¹.
>>>This
>>> should be equivalent to phase encoding directions A -> P and P -> A or
>>>vice
>>> versa.
>>> >>> However I¹m unsure whether it is possible to combine all my data in
>>> one file before running eddy and DTIFIT. For now I have just used
>>>paste to
>>> combine the bvecs and bval files (for example: paste -d" "
>>> 2001210/*WIP*_A*.bval 2001210/*WIP*_P*.bval > 2001210/DTI_A_P.bval) and
>>> fslmerge to combine the nii files. I want to run eddy and DTIFIT with
>>>the
>>> following commands:
>>> >>>
>>> >>> fsl_sub eddy --imain=DTI_A_P_1sliceless --mask=DTI_b0_brain_mask
>>> --index=index.txt --acqp=acqparams.txt --bvecs=DTI_A_P.bvec
>>> --bvals=DTI_A_P.bval --fwhm=0 --topup=topup_DTI_b0s --flm=quadratic
>>> --out=eddycor_DTI
>>> >>>
>>> >>> fsl_sub dtifit -k eddycor_DTI -m DTI_b0_brain_mask -r DTI_A_P.bvec
>>>-b
>>> DTI_A_P.bval -o fitted_DTI
>>> >>>
>>> >>> Would this be the right way of processing? Or should I process the
>>>set
>>> of images with Œfat shift direction = ³A²¹ separately from those with
>>>Œfat
>>> shift direction = ³P²? Or is it possible to average both sets of scans
>>> (they have the same bvecs)?
>>> >>> Thank you in advance for any help or suggestions.
>>> >>> Kind regards,
>>> >>> Marjolein
>>>
>>
|