Just one word of warning though..... the field map correction method comes with mixed blessings when you're dealing with diffusion MRI data.
See Figure 1 in the following article: http://onlinelibrary.wiley.com/doi/10.1002/nbm.1543/full to see how you can create 'new anatomy' with this technique.
Blip-up/ blip-down would be preferable if you can do it.
Derek
______________________________
Derek K Jones
CUBRIC
School of Psychology
Cardiff University
Park Place
Cardiff
CF10 3AT
Tel: +44 (0)29 2087 9412
Fax: +44 (0)29 2087 0339
Web: http://www.cf.ac.uk/psych/cubric/index.html
>>> Matt Glasser <[log in to unmask]> 15/10/10 8:20 PM >>>
I would also recommend distortion correcting your data (e.g. by acquiring a
field map) if you are going to go for a higher bvalue. The brainstem tends
to have some significant EPI distortion, which will change the trajectory of
your pathways.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Stamatios Sotiropoulos
Sent: Friday, October 15, 2010 11:20 AM
To: [log in to unmask]
Subject: Re: [FSL] dti scanning parameters for tracking to medulla
Dear Robert,
deciding scanning parameters is not straightforward, as many parameters
interact with each other. There is a nice figure that shows (based on
simulations) what the performance of bedpostx is on different types of data
(see Fig. 8 in the paper http://www.ncbi.nlm.nih.gov/pubmed/17070705). As
you can see there, with b=1000 and 60 directions you can reliably resolve
crossings of 50-60 degrees or more (for a reasonable SNR value of 15).
Increasing the number of directions is beneficial, but maybe not that
beneficial as increasing the b value.
On the other hand, reducing voxel size or increasing the b-value will reduce
the SNR. I would avoid going to very small voxel sizes (maybe 2x2x2 would
be a good compromise). Also, 120 directions would take more than half an
hour to acquire. If you are not happy with your current data, I would
recommend trying first 60-80 directions and b=1500-2000. In general, some
pilot scans would help to determine what is better for a specific
application.
Hope this is helpful,
Stam
the protocol you described (b=1000, 60 directions)
On 14 Oct 2010, at 21:58, Robert Schulz wrote:
> Dear FSLers,
>
> thanks a lot for all the previous very helpful comments on my questions.
Basically everything seems to work properly now, but I am wondering whether
to trust my tracking results:
>
> I would like to track primary and secondary motor areas down to the the
medulla obl., and up again for robust tracks and further normalisation of
voxel connectivity profiles. During testing, I work on 2,3 iso, 60
directional b1000 dti data. Exlusion & waypoint criteria for CS tracts
sometimes lead to only 300 tracts in healthy subjects from SMA 300
seedvoxel, about 96000 from M1 300 seed voxels at the WM-GM interface. The
data were pulse-dated.
>
> 1) Would you trust those data?
> 2) Which scanning parameters would you suggest to get reliable
cortical-medullar voxel connectivity profiles? I think of b1500, 120
directions, 1.7 or even better iso-resolution. Pulse-gated I guess. Too
much?
>
> Thanks again for your comments.
>
> Best
>
> Robert
>
|