[Mrtrix-discussion] Analysis of multiple acquisitions?

[Donald Tournier]

Hi Dorian,

To answer your first question, you can concatenate them directly when
converting them. Just give mrconvert the name of the folder containing all
your DICOM data as the first argument, it'll then present you with a
breakdown of the contents, from which you can select an individual series.
If you want to concatenate multiple series, just list the series you're
interested as a comma-separated list, or using a dash for a range (i.e. a
number sequence, as per the doc
http://www.brain.org.au/software/mrtrix/general/cmdline.html#sequences ).

That said, it might not do what you need since concatenating 4D volumes
that way will by default give you a 5D volume. It would be fine if each
(3D) DW volume was stored as its own series, which doesn't apply to you, as
far as I can tell. It also assumes the data are scaled identically, which
isn't necessarily the case. You'd probably be better off converting each
series individually, then concatenating along the 4th dimension using mrcat
(-axis 3). I'd also recommend you use MRtrix's own .mif format, since that
will preserve much more of the header information than NIfTI, particularly
the DW gradient information, which you then don't need to supply at the
command line (although I can't remember if mrcat has been updated to merge
the gradient table properly, I'd have to check).

As to the second question, it doesn't really matter for CSD itself since
the b=0 volumes are ignored. It will matter for any tensor-derived metric
(FA, MD, etc). Personally I wouldn't average on the scanner unless you're
very confident that they perform proper realignment before averaging, and
that this alignment is as good as anything you could do yourself with post
processing. Another reason not to do it is that you would then have a b=0
image with a lower SNR than the rest of your data, which might bias any
subsequent fit that you do, since most fitting routines will assume
constant SNR across measurements (i.e. within each voxel). Not sure how big
a deal that might be in practice, but enough question marks that I would
avoid it if I had the option.

Hope that helps.
Cheers,

Donald.

--
Dr J-Donald Tournier (PhD)

Senior Lecturer, Biomedical Engineering
Division of Imaging Sciences & Biomedical Engineering
King's College London

A: Department of Perinatal Imaging & Health, 1st Floor South Wing, St
Thomas' Hospital, London. SE1 7EH
T: +44 (0)20 7188 7118 ext 53613
W:
http://www.kcl.ac.uk/medicine/research/divisions/imaging/departments/biomedengineering

On 25 Oct 2013 00:52, "Dorian P." <alb.net at gmail.com> wrote:

> Dear MTRtrixers,
>
> I have two questions. First, I am getting two HARDI sequences to improve
> SNR and final results. Each sequence is in dicom. Is there a way to feed
> both sequences to MRtrix to reach this goal? Or do I need to convert them
> in niftii, concatenate them, and input the new sequence to MRtrix?
>
> Second, Philips scanners have the option to get many b0 values (i.e. six)
> and average them to a single volume. Do you advise to use this option or
> shall I better get six separate b0 volumes for MRtrix CSD analysis?
>
> Thank you.
> Dorian
> TJU
>
>
>
> _______________________________________________
> Mrtrix-discussion mailing list
> Mrtrix-discussion at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20131025/f8883b04/attachment.html>