FSL doesn't care where the b0 images are, you just need to make sure that
bvals and bvecs correctly reflect that information. The main purpose having
more b0s is to have additional data averages at that point. Although SNR of
a b0 image is higher than a DWI, you acquire many more DWIs than b0s, so
having about 1 b0 for every 8 or so DWIs is a good idea. Interleaved b0s
can be useful if your DWIs have low SNR and you need to use them to remove
bulk subject motion however.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Andy Alexander
Sent: Thursday, March 18, 2010 2:07 PM
To: [log in to unmask]
Subject: Re: [FSL] multichannel dti
Vish,
On the GE scanner you can select the phase encoding direction, make
sure this is set properly (A/P). In your example you could swap x and
y gradients and get a decent looking color map, but just make sure the
signs are ok if you want to do any kind of tractography.
You could muck with tensor.dat if you want b=0 images sprinkled
through your data although I believe FSL wants all the b=0 images up
front. I don't see any good reason to intersperse them for most
applications however.
Andy
On Mar 18, 2010, at 1:04 PM, Vishwadeep Ahluwalia wrote:
> Hi all,
> I acquired DTI images at a GE-3T using a standard coil as well as an
> 8-channel coil (using ASSET). I then converted the respective dicoms
> into
> nifti using dcm2nii thus producing the respective the nifti files as
> well as
> the bvals and bvecs. dtifit produced the fa and V1 files that were
> different
> for each coil(see attached). The standard coil vectors seem to be in
> the
> correct orientation but the multichannel coil vectors seem
> interchanged
> between x and y ( red and green). The bvecs were the same for the
> two. Upon
> looking at the dicom headers i found that the standard coil data was
> acquired with 'In-plane phase encoding direction' as COL (y
> direction i
> guess) and 'In-plane phase encoding direction' as ROW ( x direction)
> which
> might explain the flip. So should i correct for this by simply
> exchanging
> the x and y vectors in bvecs?
>
> I also acquired 4 b0 volumes. but i found that all 4 are acquired at
> the
> start of the scan. Doesnt this defeat the purpose of acquiring
> multiple b0?
> I thought these have to be interleaved between different gradient
> directions. How can i do this on a GE Discovery -750 3T magnet?
> Also, i would appreciate any advice on using multichannel coil for
> dti. As
> in, is it better than standard coil for SNR, distortion etc. Thanks!
> -Vish
>
> <multicoil1.jpg><standardcoil1.jpg>
Andrew Alexander, Ph.D.
Waisman Laboratory for Brain Imaging
Departments of Medical Physics and Psychiatry
(608)265-8233
Waisman Center
1500 Highland Ave
Madison, WI 53705
[log in to unmask]
|