Dear Gareth (et al.),
I have been planning to write documentation for the existing code but have
not yet gotten around to it (isn't that always the case?). I will be away
for much of november but hope to work on the documentation during that time.
you are correct that we are working with a fairly standard tensor sequence,
using a high b-value (~800) for the 6 diffusion-weighted scans along with one
low-b scan. the one thing that we do that's a bit different is that we take
two interleaved sequences with a distance factor of 1, moving the second
series in the z dimension by the slice thickness, such that we have two
series that are interleaved - this seems to increase SNR substantially. I
would guess that the display and tensor calculation code will be general,
depending of course on whether the gradient directions match those that we
used (since the directions are hard-coded into the tensor calculation).
the routines do the following:
spm_tbx_diffusion.m - high-level interface to the spm toolbox menu
combine_diff_series.m - combines two interleaved series into a single series
spm_diffusion_convert.m - converts from siemens .ima images (in the
particular order created by our tensor sequence) to analyze format
spm_diffusion_calc.m - takes 7 images and a predetermined b-value and
calculates the tensor, FA, and other related information
load_tensor_data.m - reads in tensor data for viewing using rip (a package
not included here - I'll have to speak to the author of that package about
whether it can be distributed)
spm_diffusion_orthviews.m - obsolete, I will remove it from the archive
spm_diffusion_orthviews2.m - takes a grayscale image (either anatomy or FA)
and three color images (created by spm_diffusion_calc.m) and displays
orthogonal views of the color map and the grayscale image
the things that I see as the biggest needs at this point are:
1 - tensor calculation with user-specified gradient directions
2 - eddy current corrections
3 - a generalized conversion program for both siemens and GE data
I hope that you and others will find this code useful, and will contribute
your own modifications and additions. Please note that this is alpha-level
code and it has not been tested in a wide range of circumstances.
cheers,
russ
"Dr. Gareth Barker" wrote:
> > From: Russ Poldrack <[log in to unmask]>
>
> > Hi folks - I have created a project page on sourceforge.net for the
> > development of SPM toolboxes. For those of you who aren't familiar with
> > sourceforge.net, it's a site that provides hosting for open source
> > development projects. The primary feature is a web-accessible
> > concurrent versions system (CVS) server, which allows multiple
> > developers to work on the same code simultaneously.
> >
> > The home page for the project is http://spm-toolbox.sourceforge.net -
> > this page will lead you to the CVS repository, along with pointers to
> > CVS and sourceforge documentation.
> >
> > There are currently two projects under development: an ROI analysis
> > toolbox and a diffusion tensor analysis toolbox. If you are interested
> > in using or working on these toolboxes or developing additional
> > toolboxes, please register as a sourceforge developer and download the
> > code.
>
> I've had a quick look at the diffusion code, but as often with matlab
> (and spm) code if you dont know where to start looking it's not at all
> clear what is what. Could you give us a quick overview of the various
> routines? In particular, which routines are likely to be generally
> useful (spm_diffusion_display, spm_diffusion_orthoviews?) and which are
> site and protocol specific (spm_diffusion_convert,
> combine_diff_series?) Also, what would you like to see developed?
>
> The usual initial problem with processing diffusion data is that
> everyone uses different protocols, and that you need to work out the
> data acquisition order and also do a full bvalue calculation for each
> different pulse sequence/protocol/whatever. My initial quick look
> suggests that you're using a Siemens system ('.ima' files) and that
> from the comments in the files you do a fairly standard 7 axis tensor
> measurement with equal averaging and diffusion weighting on each
> diffusion step (2 b values only)? Any more details you can give us so
> that anyone doing other protocols can work out what they need to
> change?
>
> Thanks in advance,
>
> _________________
>
> Gareth J. Barker, Institute of Neurology, Queen Square,
> London WC1N 3BG, UK.
>
> email : [log in to unmask]
> SMS (phone/pager) : [log in to unmask] (< 100 characters, please!)
> phone : +44 20 7837 3611 Xn 4269
> fax : +44 20 7278 5616
--
Russell A. Poldrack, Ph. D.
Assistant Professor of Radiology, Harvard Medical School
MGH-NMR Center
Building 149, 13th St.
Charlestown, MA 02129
Phone: 617-726-4060
FAX: 617-726-7422
Email: [log in to unmask]
Web Page: http://www.nmr.mgh.harvard.edu/~poldrack
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|