I fully agree: On our system readjusting the table greatly reduced the vibration atrefact. However, it did not eliminate it and in some patients it can still be quite nasty. However, I would also consider it as one of the first and most important steps.
2mm isotropic with 30 dirs in about 5min is the same what we use clinically. Multi-channel recordings can be nice for probtrack wen you put your seeds/targets into the cortex.
Cheers-
Andreas
________________________________
Von: FSL - FMRIB's Software Library im Auftrag von Jones, Richard
Gesendet: Di 16.09.2008 15:27
An: [log in to unmask]
Betreff: Re: [FSL] AW: [FSL] AW: [FSL] voxel size vs structure size
On the Trio the adjustment of the patient table appears to be crucial.
We had two nominally identical Trio scanners at two different sites, on
one the vibration was typical of that for diffusion sequences, on the
other the vibration was terrible. One could reduce the vibration
somewhat by changing the number/orientation of the diffusion gradients
but the problem was not solved until Siemens readjusted the table. It
may be that you have the same problem.
With the standard Siemens array head coil we acquire whole brain, 2mm
isotropic DTI data with 30 directions in approx 5. minutes. While the
SNR is less than one would like this is a reasonable time for clinical
studies. How much more channels will help will depend on what you are
interested in, if you are looking mainly at the FA etc. of white matter
then I doubt it will help very much since the bulk of the white matter
is located away from the surface of the brain.
Richard
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
Behalf Of Andreas Bartsch
Sent: Tuesday, September 16, 2008 6:22 AM
To: [log in to unmask]
Subject: [FSL] AW: [FSL] AW: [FSL] voxel size vs structure size
Hi,
stabilzing the table is good, and we also had some success with head
cushoining and fixation. Switching the phase or blip encoding may help
(as may prone instead of supine positioning if the patient tolerates it
and fits into the coil) but I generally don't like R/L for symmetry
reasons and because distortion correction is not perfect. Hmmh, the
gradient encoding may have an impact.
Matt: could you send me the FMRIB vecs you've used? Just mail it to me
directly, I want to compare it to the ones I've implemented (they may be
identical but I'd like to check and try if there's a difference).
Eliminating the artefact is not easy (even though I would hope Siemens
puts more efforts into it): The artefact probably results from an
interaction of the gradient strength and the entire scanner
architecture. So it won't be easy to get rid of it;(
Cheers-
Andreas
________________________________
Von: FSL - FMRIB's Software Library im Auftrag von Matt Glasser
Gesendet: Di 16.09.2008 04:04
An: [log in to unmask]
Betreff: Re: [FSL] AW: [FSL] voxel size vs structure size
Out of curiosity, who did you talk to in St Louis? At Emory we were
able to reduce the vibration artifact by packing the head tightly or
reducing the TR. Additionally, using R/L phase encode direction (with
distortion correction) and the fMRIB's gradient table may have helped.
Peace,
Matt.
________________________________
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
Behalf Of Scott Kolbe
Sent: Monday, September 15, 2008 8:22 PM
To: [log in to unmask]
Subject: Re: [FSL] AW: [FSL] voxel size vs structure size
I agree Andreas
the vibration on that thing is shocking. I was talking to a guy from St
Louis who said they were able to stablise their table somewhat to reduce
the artefact. I can't understand why Siemens haven't done something
about this. re: TSE and SSFP-DWI, are those now available as standard on
the latest OS upgrade? we just got the 32-channel head coil so i will
try higher res. cheers, scott
Andreas Bartsch wrote:
Hi Matt / Scott / Dianne-
on our TIM Trio I have gotten down to 1.5*1.5*2 mm voxels with DWI-EPI
at fair SNR with a single average using a 32 channel head coil. The
previous stripe artefact has been eliminated on the most recent numaris
version, the vibration artefact still bothers us. For peripheral nerve
studies using TSE- or SSFP-DWI can be advantageous.
Cheers-
Andreas
________________________________
Von: FSL - FMRIB's Software Library im Auftrag von Matt Glasser
Gesendet: Mo 15.09.2008 06:02
An: [log in to unmask]
Betreff: Re: [FSL] voxel size vs structure size
Scott,
I guess I wasn't really clear there. I have seen reducing the FOV to
reduce SNR (even when voxel size was kept constant) for a physics reason
that I did not really understand. What I meant to say was to only
acquire a few slices in the region you are interested in so that you can
acquire many averages in a reasonable amount of time. Regarding SNR, I
thought that 4-6 averages at 60 directions and 1.3mm isotropic would
produce good results (expect to scan for a long time though if you are
going for whole brain coverage, however). SNR is definitely not too low
if you acquire the averages (this on a TIM Trio, I don't off hand recall
what coil we used).
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
Behalf Of Scott Kolbe
Sent: Sunday, September 14, 2008 10:52 PM
To: [log in to unmask]
Subject: Re: [FSL] voxel size vs structure size
Hi Matt
you mentioned reducing the FOV,
have you tried this on your Trio? what was the SNR like at 1.3mm? we
tried this a couple of years ago and got some really ugly zebra stripes
in our images. cheers Scott
--
========================
Scott Kolbe
MS Imaging Group
Howard Florey Institute &
Centre for Neuroscience
University of Melbourne
VIC, Australia, 3010.
ph: +61 3 8344 1887
email: [log in to unmask]
website: http://www.neuroimaging.org.au/index.php?id=383
Kochunov, Peter wrote:
For a project like this one might think about using a TSE-DTI
sequence.
You won't have many of the limitations of the EPI-DTI sequence, although
it will be only practical if you can sufficient SNR through multiple
averages.
cheers
pk
________________________________
From: FSL - FMRIB's Software Library on behalf of Dianne
Patterson
Sent: Sun 9/14/2008 4:28 PM
To: [log in to unmask]
Subject: Re: [FSL] voxel size vs structure size
Thanks so much...that gives me a place to start!
-Dianne
On Sun, Sep 14, 2008 at 2:01 PM, Matt Glasser <[log in to unmask]>
<mailto:[log in to unmask]> wrote:
Dianne,
I think you ought to be able to get much better resolution
than
that. I have scanned a human at 1.3mm isotropic on a Siemens 3T
scanner. If you are specifically interested in small nerves, it would
probably be best to get as high resolution as possible and scan a
limited field of view with many averages. To measure the FA of a small
structure accurately, you would need at least one voxel without partial
voluming (or devise a method to correct for partial voluming). I have
not personally done this, but theoretically you could use higher
resolution anatomical imaging to estimate the fraction of the structure
that was present in the DTI voxel, and if you knew the mean FA of
whatever else was in the voxel, you could estimate the FA in the
structure of interest. There may be issues with such an approach that
others may point out, and I could see it as difficult to get past
reviewers, however, and it would rely on perfect alignment between the
anatomical image and DTI (i.e. any EPI susceptibility distortion would
have to be corrected). Increasing the number of directions will not
compensate for poor spatial resolution. Increasing the number
directions will help in reconstructing multiple fiber directions for
tractography, but it sounds like you are more interested in quantitative
DTI values.
Peace,
Matt.
________________________________
From: FSL - FMRIB's Software Library
[mailto:[log in to unmask]] On
Behalf Of Dianne Patterson
Sent: Friday, September 12, 2008 10:00 PM
To: [log in to unmask]
Subject: [FSL] voxel size vs structure size
Dear Group,
We have a 3 tesla GE scanner and we get 2.6 mm isotropic
voxels in
the brain.
I recently talked to a colleague who would love to image
nerves in
the neck that are about 3 mm thick.
Can anyone tell me:
1) Whether 2.6 mm voxels would be appropriate for
identifying and
measuring FA along such a small diameter structure.
2) What the relationship needs to be between voxel size
and
structure size
3) If increasing angular resolution can compensate for the
lack of
spatial resolution
I appreciate your kindness and support..thankyou,
Dianne
--
Dianne Patterson, Ph.D.
[log in to unmask]
University of Arizona
SHLS 328
621-5105
--
========================
Scott Kolbe
MS Imaging Group
Howard Florey Institute &
Centre for Neuroscience
University of Melbourne
VIC, Australia, 3010.
ph: +61 3 8344 1887
email: [log in to unmask]
website: http://www.neuroimaging.org.au/index.php?id=383
|