On 03/11/2009 13:05, "John" <[log in to unmask]> wrote:
> Hi,
>
> I'm looking confirm my fieldmap acquisition process and data preprocessing
> steps.
>
> My plan is to acquire two distinct GRE field maps with differing echo times
> using our 8Ch head coil. I have a couple questions specific to my GE CV's
> and a couple more generic questions.
>
> My GE CV's are
>
> rhrcctrl = 15 (for magnitude, phase, real, imaginary, )
Hi,
You don't need all of those, since you can get magnitude & phase from the
real & imaginary parts. Having said that, I usually save magnitude, real &
imaginary (rhrccctrl = 13).
>
> autolock = 1
>
> Do I need the raw p-file, if so, for what?
No, you shouldn't really need the raw data, so leave autolock to its default
value
>
> nograd = 1
>
> Should this be set to match the acquisition I'm correcting with the
> fieldmap? Ie, if my DTI acqusition has gradwarp, I should be acquiring my
> fieldmap with gradwarp also? Or is the opposite true?
I always leave gradwarp on for fieldmap acquisitions, since it's on for the
EPI that I want to distortion correct.
>
> saveinter = 1
>
> If i'm processing the data in FUGUE do i need the intermediate files, ie
> independent coil element (real, imaginary, mag and phase) images. I didn't
> think I but i read in another post (014548) that these could be used.
>
> Also, I read in another post (014548) that it may be preferable to switch to
> the headcoil to acquire fieldmaps, this relates to my saveinter question
> above, but is this only if i'm processing the fieldmaps using Bernstein MRM,
> 1994 method? Can I process my 8ch data through FUGUE?
If the pulse sequence you use supports it, use ASSET with an asset factor of
1. That way the real, imaginary & phase images are calculated correctly (ie
not as the sqrt(SoS). In that case you don't need to set saveinter = 1.
If the sequence doesn't support ASSET with a factor of 1, you'll need to use
saveinter = 1 to get the individual coil images. You can then use
Bernstein's method to calculate the average phase difference image of all
the coil elements for the two echoes. One disadvantage is you end up with at
least 36 images per slice of the field map (if you save real & imaginary or
magnitude & phase).
>
> I think what has led to confusion is I have began to process some
> preliminary data. When I apply fslinfo to my data I get:
>
> data_type INT16
> dim1 256
> dim2 256
> dim3 4
> dim4 1
> datatype 4
> pixdim1 0.9375000000
> pixdim2 0.9375000000
> pixdim3 4.0000000000
> pixdim4 1.0000000000
> cal_max 0.0000
> cal_min 0.0000
> file_type NIFTI-1+
>
> That is, my datatype isn't 32 (as for complex images)
It doesn't have to be, you can supply fugue with a float field map, which is
just the scaled unwrapped phase difference image from the field map
acquisition
>
> I am using dcm2nii GUI to convert my .dcm images to .nii.gz, that should be
> ok, right?
>
> From my acquisition with the above parameters I have 4 images per slice (as
> per my rhrcctrl vaules). The 1st and 3rd images range from 0- about 3140
> (this is pi *1000), and it looks quite like a normal MR image, my 2nd image
> looks like a badly tuned TV and my 4th image is blank, though i did acquire
> the images on a phantom.
I've never output magnitude, real, imaginary and phase, so I don't know what
order they'll be. Obviously the magnitude will just look like a standard MRI
as you describe. The real and imaginary will have smoothly varying dark &
light bands, while the phase will also appear banded, but with sudden jumps
from light to dark where the phase wraps occur.
Dave
>
> Does this sound OK for processing through FUGUE?
>
> Apologies for posting on fieldmaps, I know this has been pretty well
> convered but I think I can't see the wood for the trees.
>
> Many thanks
> John
>
|