Dear Stefan,
> Dear Jesper,
>
> Thank you for your quick and detailed response.
> I have used topup via command line:
> topup --imain=AP_PA.nii --datain=accparams.txt --out=topup --iout=topup_unwarped.nii
>
> The AP_PA.nii file contains two 3D b0 volumes with opposite phase encoding direction (as mentioned below). Apart of the phase encoding direction, the same scanning parameters were used for both volumes.
>
> So the accparams.txt file has two lines (below)
> The distortions in the frontal area are not as well corrected as I have hoped for (see attached image).
No, you should be able to do better than that.
>
> I ran topup also with the same files but fed in fsl's b02b0.cfg file, which I changed only to avoid subsampling. And the correction seems to be improved. So I believe with a more elaborated search for good parameters, I may be able to get a better correction.
Do you have any space above or below the brain? If so I would recommend either removing the top or bottom slice, or you can take a copy of the top slice and stick it back on top. That way you can use b02b0.cf as is. The sub-sampling step is not just to save time, it is also important to get the algorithm started in the right direction.
>
> Does topup register the b0 volumes (rigidly) before warping? Otherwise I may have to consider to align my b0 volumes beforehand.
Yes, there is a rigid registration as part of the estimation in topup. Trying to register a blip-up to a blip-down outside of topup is probably going to be difficult.
Jesper
>
> Thanks agian for you help
> Stefan
>
>
>
>
>
> On 2016-05-24 12:51, Jesper Andersson wrote:
>> Dear Stefan,
>>> I try to correct my DWI for EPI distortions with topup. As know from the FSL topup documentation, I need an acquisition file to feed the scanning parameters to topup (--datain).
>>> I have two b0 volume with opposite phase encoding direction A>>P and P>>A, so my acquisition file would have two rows with
>>> 0 -1 0 X
>>> 0 1 0 X
>>> However I have problems to define the X value.
>>> First I have computed it according to the documentation with: X = echo spacing(in ms) * (EPI factor/GRAPPA - 1) * 10^-3, which in my case would be: X = 0.72 * (128/2 - 1) = 0.04536
>>> Unfortunately, the result is not satisfying.
>> I’ll reply to the readout time question below. Here I want to say that
>> if the results are not satisfying, the readout time is not the issue.
>> In what way are the results not satisfactory? What did your command
>> line look like?
>>> -> Do I compute the value correctly?
>> Yes
>>> -> Is the result maybe a limit of topup, or can I compute the factor in a different/more accurate way?
>> No, the above is correct.
>>> -> Why do I not have to consider partial Fourier acquisition, as this influences the acq. time as well?
>> The relevant thing is the off-resonance phase-evolution over the
>> course of the acquisition, and that doesn’t change with partial
>> Fourier. In the partial Fourier case there is an implicit assumption
>> that the phase-evolution in the “unacquired” part is the same as in
>> the acquired part.
>> Jesper
>>> Details of my scan from protocol:
>>> Echo spacing = 0.72ms
>>> EPI factor = 128
>>> Bandwidth = 1698 Hz/Pxl
>>> Further acquisition details from the dicom header in the screenshot, attached to this e-mail.
>>> Hope you can help me.
>>> Thank you very much in advance
>>> Stefan
>>> <Capture.PNG>
> <AP.PNG>
|