Dear Romain,
It may be that the phase maps on Siemens are not unwrapped but simply acquired
with a small enough TE spacing that there is no significant amount of wrapping.
If they are actually unwrapped then you are right that the range can easily be
more than 2*pi. However, we normally find that we need to rescale a 0 to 4096
range to 0 to 2*pi radians. Check to see if this works appropriately for your
images.
All the best,
Mark
On 24 May 2012, at 23:55, romain valabregue wrote:
> Hi Mark
>
> Thanks for those precision. For the Bruker data it is indeed easy since I have a phase difference map that is directly express in Hz. So I just multiply it by 6.28 and it seems to work well (after correction the result is reasonable)
>
> I also have data for Siemens (Trio or Verio both in vb17), where I also get a phase map. Looking at different post (and at the doc) I need to rescale the phase so that the values are between -pi and pi.
>
> I do not understand why the value should be between -pi and pi : the phase map seems to be already unwrapped. I gues so since looking at the data I do not see any strong transition in the phase value.
> So if the phase map is indeed unwrapped you can have phase values greater than pi ?
>
> what scaling do I need to do for Siemens phase map ?
>
> thanks for your help
>
>
> Romain
>
>
> Le 23/05/2012 11:06, Mark Jenkinson a écrit :
>> Dear Romain,
>>
>> You must have a magnitude image and an image that encodes the
>> field differences. We opt for using the units rad/s, which is easy to
>> convert from Hz (just multiply by 6.28) or Tesla (multiply by 42 MHz/T
>> to get Hz and then multiply by 6.28 to get rad/s).
>>
>> The latter information is derived from phase differences.
>> Depending on your scanner (and I have *no* experience with Bruker
>> I'm afraid), you may get one image (the difference) or two. These
>> may be phase values directly (though often scaled to span a large
>> integer range, like 0 to 4096 for 0 to pi radians) or may be in
>> real and imaginary parts. The tool fslcomplex has been written
>> to allow you to convert between the different forms. However,
>> once you have a phase difference you still need to divide by the
>> time difference between the images (normally the echo time
>> difference of the fieldmapping sequence if it is gradient-echo
>> based, but there are other alternatives too). Only after you've
>> divided by this time difference will you get units of rad/s rather
>> than just radians.
>>
>> I hope this helps.
>> All the best,
>> Mark
>>
>> On 16 May 2012, at 01:39, romain valabregue wrote:
>>
>>> Dear all
>>>
>>> I am trying to unwarp dti data from a mouse brain (with a Bruker 11.7 T)
>>>
>>> I get a little bit confused with the documentation
>>> http://www.fmrib.ox.ac.uk/fsl/fugue/feat_fieldmap.html
>>>
>>> from what I understand I have a magnitude volume (that is easy) and a phase volume that is in Hz
>>> does anyone use Buker data and can confirm this ?
>>>
>>>
>>> So the only scaling I need is to multiply this phase volume by 6.28 and I do not need to enter the TE difference
>>> is that right ?
>>>
>>> I get value between -9000 and +4000 (or -3000 + 3000 if I restrict to the brain)
>>> does it seems reasonable ?
>>>
>>> thank you for your help
>>>
>>> Romain
>>>
>
|