On Wed, 24 Oct 2007 11:13:02 +0100, David Lythgoe
<[log in to unmask]> wrote:

>On 23/10/07 22:12, "Carlos Faraco" <[log in to unmask]> wrote:
>
>> I am trying to unwarp some images and have some questions concerning the
>> instructions in the documentation. To begin with I am acquiring separate
>> phase and magnitude images in DICOM format from the scanner (GE 3T) and am
>> converting them to nii format. Using fslinfo for the phase images I receive
>> data_type INT16 and datatype 4. So then this should mean I should converting
>> to floating point format (not sure though what datatype 4 means, though)
>> using the fslmaths command stated in 2)b of the documentation. The value for
>> the -div operation is stated as 2048, but the manual states this needs to be
>> adjusted for each scanner. How can this value be determined?
>
>Phase images from GE scanners are in radians multiplied by 1000, so should
>be in the range around -3142 -> 3142. If the numbers are all positive, then
>it's possible they've been acquired using a multi-channel receive coil, and
>you should discuss it with whoever's responsible your data acquisition. It
>us possible to use such data, but it requires a few changes on the
>acquisition side, and the processing is slightly more complicated.

So what you are saying, I think, is that I do not need to mult by 3.14159,
just div by 1000 (if it is the case that these values have originally been
multiplied by 1000)? 

How can I determine what these values are?
>
>>
>> When these files are then converted to radians how do I determine their
>> ranges?
>>
>> In acuqiring the two phase images, does one images just have to have a
>> slightly longer TE than the other, does it have to be an asymmetric echo (as
>> described in some posts), or both? What exactly is an asymmetric echo? One
>> post, #002856, describes the asym time for a gradient echo as "
>> the factor required to convert the phase difference in the scans to a
>> rad/s field value." I assume this factor is one of the values used in the
>> above fslmaths argument, but which one and how is this implemented in the
>> scan? This is really confusing to me.
>
>You can use either gradient echoes with different Tes, or a spin-echo/and an
>asymmetric spin echo. An asymmetric spin echo is just a spin-echo with the
>180 degree refocusing pulse shifted earlier or later.
>

Can a spin-echo sequence with a longer echo train be used (e.g., Fast
spin-echo or Turbo)?

To make sure, the spin-echoes should have the same TE, just one symmetric
and one asymmetric?

Additionally, since these images will be T1 weighted, how does this affect
the distortion correction of a T2 weighted? Is this similar to
warping/normalizing fMRI scans to the structurals?

>>
>> Lastly, I have read that it is recommended to take spin-echo scans for these
>> phase maps, but this takes an extremely long amount of time. The
>> documentation states gradient echo images are OK (which are what I have been
>> acquiring), but do these require any extra processing?
>
>They don't really require more processing, but I prefer them because they
>don't suffer from signal dropout as much as gradient echoes, especially in
>the regions where the correction is needed the most! You just need to know
>the echo shift, which is twice the shift of the 180 degree pulse, and is
>equivalent to delta TE for gradient echoes.

Could you clarify the above sentence? Are you refering to how the 180 occurs
at half of the TE on a regular spin-echo.

Thank you!

>
>Also, apart from EPI, I don't recall any of the GE pulse sequences
>supporting asymmetric spin-echoes without modification.
>
>
>Hope that helps,
>
>
>Dave
>
>>
>> I apologize for all the questions, but I have been trying to figure this out
>> for a while and am currently stuck!
>>
>> Thanks!
>>
>> -Carlos
>=========================================================================