On Nov 20, 2007 12:01 PM, David Lythgoe <[log in to unmask]> wrote:

On 20/11/07 17:06, "Carlos Faraco" <[log in to unmask]> wrote:

> Alright, a few more things. I apologize if some of my questions may be
> rudimentary, but I just recently started to become familiar with fieldmaps
> and also do not have a physics/programming background.
>
> So the suggestion has been to use the real and imaginary images and
> calculate the phase from them using atan2, or whatever other method there
> may be. How is this different from using the phase image produced by the
> scanner?
>
> How is atan2 used? What would y and x be? I am looking at the syntax but do
> not understand it too well.

y is the imaginary component of the signal, and x is the real part. Part of
the reason I acquire real & imaginary is that the complex maths calculations
I did required them rather than magnitude & phase, so I acquire them rather
than calculate them. I'll send you an abstract.

>
> Dave, the reason you set saveinter = 1, is not necessarilly for the purpose
> of saving the individual coil images, but just to get the combined coil
> image in the appropriate units, correct?

I didn't know there was a way of getting the correct phase, so I needed all
the individual coil images in order to calculate the the phase difference
between the SE & ASE myself.

>
> If I use the real and imaginary images, do I still have to perform the
> processes described in the article?

I assume if you use the method you mentioned of turning asset on, then
setting the asset factor to 1 you won't have to, just calculate the phase
difference.

>
> John, what is your rationale for using the fast GRE and is there anything
> you do with it differently than with standard GRE? I originally thought the
> fast GRE acquired multiple lines of k-space per echo, however Dave informed
> me this is incorrect. Thanks for you suggestion on the masking, that makes
> sense. Do you just create this mask using FSL?

I think this is a matter of terminology. GE fast sequences acquire a line of
k-space at a time, whereas EPI acquires all lines of k-space in one shot (or
more). With GE sequences the fieldmap will be in (almost) undistorted space
similar to the structural (SPGR on GE). If you use EPI to create the
fieldmap, it will be distorted too.
As for the masking, applying bet to the magnitude image does a fair job.

>
> Lastly, any more suggestions as to things I should be doing are greatly
> appreciated. Again, I am fairly new to fieldmaps and am sort of figuring
> this out on my own.
>
> Thanks!