[Mrtrix-discussion] mrtrix v3 precisions

[Robert Smith]

Romain

There is a class in the code back-end that receives the diffusion weighting
gradient table, and attempts to *cluster* the b-values into discrete shells.
It does so in a way that allows for some level of variance in the actual
b-values as reported by the gradient table (such as that you have
provided), so that even if the b-values are not precisely identical, they
will still be grouped into the same shell.
When this process is completed, each b-value shell is characterised
internally by the mean / stdev / minimum / maximum b-value in the shell.
When you provide a 'desired' b-value using the -shell option, this value is
compared to the shell classification as produced by the clustering
algorithm, such that you don't have to specify precisely the mean b-value
at the command-line in order to select the shell you want. If however there
are multiple shells in your data, and your b-value request cannot be
unambiguously assigned to one of those shells (e.g. the data has
b=0,1000,2000,3000 shells and you specify -shell 2500) the command should
give you an error.
Also note that if you try to feed a DSI acquisition to such a command, the
clustering algorithm *should* also identify the fact that a shell-like
classification is not appropriate for these data, and provide an error
message accordingly.

Regarding the use of multiple shells: We now have a neat command
construction framework that allows a standardised command-line option to be
accessed in multiple commands (so we don't have to repeatedly duplicate
them). This is used to define the -shell option for all commands that make
use of it, and all of those commands construct the class described above
for the b-value clustering. Since there is at least one command
(dwiextractfrom memory) that allows the use of multiple b-value
shells, the
command-line option states 'one or more'. However you should find that if
you provide more than one b-value to a command such as dwi2fod (e.g. -shell
1000,2000,3000), it should give you an appropriate error message.

Best regards
Rob


--

*Robert Smith, Ph.D*
Research Officer, Imaging Division

The Florey Institute of Neuroscience and Mental Health
Melbourne Brain Centre - Austin Campus
245 Burgundy Street
Heidelberg Vic 3084
Ph: +61 3 9035 7128
Fax: +61 3 9035 7301
www.florey.edu.au


On Wed, May 21, 2014 at 5:37 PM, romain valabregue <
romain.valabregue at upmc.fr> wrote:

>  Hi Robert
>
> Thanks for the clarification. Just one more precision then
> what happend if I have  a list a bvalue like
>
> 0 0 0  3005 2995 3000 2990 2995 3010 3005 2985 3005 2990 300
>
> It should not be consider as a multishell. What is then the range you
> consider as being the same b value ?
>
>
>
> Note that the documentation :
> -shell list
>      specify one or more diffusion-weighted gradient shells to use during
>      processing ...
>
> this suggest that you can take into account more than one diffusion shells
> ...
>
>
> romain
>
>
>
> Le 21/05/2014 09:10, Robert Smith a écrit :
>
> Hi Romain
>
>    1. The previous functionality of filter_tracks can now be found in the
>    tckedit command. This command is a bit of a track-editing workhorse;
>    check the -help page to see all of its functionality.
>    2. We currently do not have genuine multi-shell processing, though we
>    hope to introduce it in the future. The -shell option in dwi2fod (the
>    same option also exists in other dwi-related commands) simply specifies the
>    b-value in a multi-shell acquisition to be extracted for use in subsequent
>    processing. If a multi-shell acquisition is received as input, and the
>    -shell option is *not* provided, the command will extract and use only
>    the highest b-value shell by default.
>
> Cheers
>
> Rob
>
>
> --
>
> *Robert Smith, Ph.D*
> Research Officer, Imaging Division
>
> The Florey Institute of Neuroscience and Mental Health
> Melbourne Brain Centre - Austin Campus
> 245 Burgundy Street
> Heidelberg Vic 3084
> Ph: +61 3 9035 7128
> Fax: +61 3 9035 7301
> www.florey.edu.au
>
>
> On Wed, May 21, 2014 at 6:25 AM, romain valabregue <
> romain.valabregue at upmc.fr> wrote:
>
>> Hello
>>
>> Two short questions about the very nice new version (I just started to
>> test)
>>
>> 1_ I did not find the equivalent of the previous function filter_tracks
>>
>> 2_ In the dwi2fod function I do not understand the -sehlll options. Can I
>> use it with a multishell dataset not already ? (the bvalue are already in
>> the gradient table)
>>
>>
>> Many thanks (and congratulation for the nice viewer)
>>
>> Romain
>> _______________________________________________
>> Mrtrix-discussion mailing list
>> Mrtrix-discussion at www.nitrc.org
>> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>>
>
>
>
> _______________________________________________
> 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/20140521/abc89b6a/attachment-0001.html>