Analyze Skeleton - largest branch question

[Jan Eglinger-5]

Dear all,

As far as I see, the AnalyzeSkeleton plugin measures the maximum branch
length as the length of the largest branch between any two junctions, or
between any junction and an end-point.

Is there a way to measure the longest continuous length between any two
end-points as well? (that means over several junctions, if necessary)

Jan

[Ignacio Arganda-Carreras]

Hello Jan,

As it is now, the code does not provide an easy way of calculating the
longest branch including junctions. However, i made some changes
lately and i store now the skeleton as an undirected graph (set of
vertices and edges). You could maybe play around with some graph
algorithms to calculate that measure...

If more people are interested, i can add an option to store the
skeleton as a JGrapht graph (http://jgrapht.sourceforge.net/).

Just let me know!

Cheers!

ignacio

On Mon, Sep 7, 2009 at 12:16 PM, Jan Eglinger<[hidden email]> wrote:

> Dear all,
>
> As far as I see, the AnalyzeSkeleton plugin measures the maximum branch
> length as the length of the largest branch between any two junctions, or
> between any junction and an end-point.
>
> Is there a way to measure the longest continuous length between any two
> end-points as well? (that means over several junctions, if necessary)
>
> Jan
>

--
Ignacio Arganda-Carreras, Ph.D.
Institute of Neuroinformatics
Uni/ETH Zurich
Winterthurerstrasse 190
Bau 55
Zurich 8057
[phone]

Website: http://bioweb.cnb.csic.es/~iarganda/

[Gabriel Landini]

On Monday 07 September 2009  12:05:02 Jan Eglinger wrote:
> Is there a way to measure the longest continuous length between any two
> end-points as well? (that means over several junctions, if necessary)

Yes there is, using mathematical morphology to get the geodesic distance via
the so called "geodesic dilation" of one of the starting points as seed and
the rest of the skeleton as mask.

The drawback is that you need to repeat this for all possible starting points
to know which two points are furthest away.

Cheers

Gabriel

[gerrit]

> Hello Jan,
>
> As it is now, the code does not provide an easy way of
> calculating the longest branch including junctions. However,
> i made some changes lately and i store now the skeleton as an
> undirected graph (set of vertices and edges). You could maybe
> play around with some graph algorithms to calculate that measure...
>
> If more people are interested, i can add an option to store
> the skeleton as a JGrapht graph (http://jgrapht.sourceforge.net/).
>
> Just let me know!

In order to measure the largest branch, you need to implement the
Dijkstra algorithm.
Since this already is implemented in jgrapht, it would be very
convenient when the skeleton can be stored as jgrapht.

I also need this functionality.

So ignacio, yes please!!!!!!!!!!!!!!

Kindly,

Gerrit.

[Ignacio Arganda-Carreras]

Hello everyone again,

As a workaround for everything we talked here, i just updated
AnalyzeSkeleton with an option to display all the branch information
needed to create a graph. At the same time, I removed the log window
because the information that was being displayed there is now included
in the new "Branch information" result table:

http://pacific.mpi-cbg.de/wiki/index.php/AnalyzeSkeleton

Hope it is useful!!!

ignacio


On Tue, Sep 8, 2009 at 9:32 AM, Polder, Gerrit <[hidden email]> wrote:

>> Hello Jan,
>>
>> As it is now, the code does not provide an easy way of
>> calculating the longest branch including junctions. However,
>> i made some changes lately and i store now the skeleton as an
>> undirected graph (set of vertices and edges). You could maybe
>> play around with some graph algorithms to calculate that measure...
>>
>> If more people are interested, i can add an option to store
>> the skeleton as a JGrapht graph (http://jgrapht.sourceforge.net/).
>>
>> Just let me know!
>
> In order to measure the largest branch, you need to implement the
> Dijkstra algorithm.
> Since this already is implemented in jgrapht, it would be very
> convenient when the skeleton can be stored as jgrapht.
>
> I also need this functionality.
>
> So ignacio, yes please!!!!!!!!!!!!!!
>
> Kindly,
>
> Gerrit.
>

--
Ignacio Arganda-Carreras, Ph.D.
Institute of Neuroinformatics
Uni/ETH Zurich
Winterthurerstrasse 190
Bau 55
Zurich 8057
[phone]

Website: http://bioweb.cnb.csic.es/~iarganda/

[David Webster]

I'm curious why the Fifi and ImageJ versions are different (or at least the
wiki descriptions).

David Webster

On Sun, Sep 13, 2009 at 6:07 AM, Ignacio Arganda
<[hidden email]>wrote:

> Hello everyone again,
>
> As a workaround for everything we talked here, i just updated
> AnalyzeSkeleton with an option to display all the branch information
> needed to create a graph. At the same time, I removed the log window
> because the information that was being displayed there is now included
> in the new "Branch information" result table:
>
> http://pacific.mpi-cbg.de/wiki/index.php/AnalyzeSkeleton
>
> Hope it is useful!!!
>
> ignacio
>
>
> On Tue, Sep 8, 2009 at 9:32 AM, Polder, Gerrit <[hidden email]>
> wrote:
> >> Hello Jan,
> >>
> >> As it is now, the code does not provide an easy way of
> >> calculating the longest branch including junctions. However,
> >> i made some changes lately and i store now the skeleton as an
> >> undirected graph (set of vertices and edges). You could maybe
> >> play around with some graph algorithms to calculate that measure...
> >>
> >> If more people are interested, i can add an option to store
> >> the skeleton as a JGrapht graph (http://jgrapht.sourceforge.net/).
> >>
> >> Just let me know!
> >
> > In order to measure the largest branch, you need to implement the
> > Dijkstra algorithm.
> > Since this already is implemented in jgrapht, it would be very
> > convenient when the skeleton can be stored as jgrapht.
> >
> > I also need this functionality.
> >
> > So ignacio, yes please!!!!!!!!!!!!!!
> >
> > Kindly,
> >
> > Gerrit.
> >
>
>
>
> --
> Ignacio Arganda-Carreras, Ph.D.
> Institute of Neuroinformatics
> Uni/ETH Zurich
> Winterthurerstrasse 190
> Bau 55
> Zurich 8057
> [phone]
>
> Website: http://bioweb.cnb.csic.es/~iarganda/
>

[Ignacio Arganda-Carreras]

That's because i didn't have time to update both websites... I'll do
it soon though. The code is the same.

Cheers!

ignacio

On Sun, Sep 13, 2009 at 6:53 PM, David Webster <[hidden email]> wrote:

> I'm curious why the Fifi and ImageJ versions are different (or at least the
> wiki descriptions).
>
> David Webster
>
> On Sun, Sep 13, 2009 at 6:07 AM, Ignacio Arganda
> <[hidden email]>wrote:
>
>> Hello everyone again,
>>
>> As a workaround for everything we talked here, i just updated
>> AnalyzeSkeleton with an option to display all the branch information
>> needed to create a graph. At the same time, I removed the log window
>> because the information that was being displayed there is now included
>> in the new "Branch information" result table:
>>
>> http://pacific.mpi-cbg.de/wiki/index.php/AnalyzeSkeleton
>>
>> Hope it is useful!!!
>>
>> ignacio
>>
>>
>> On Tue, Sep 8, 2009 at 9:32 AM, Polder, Gerrit <[hidden email]>
>> wrote:
>> >> Hello Jan,
>> >>
>> >> As it is now, the code does not provide an easy way of
>> >> calculating the longest branch including junctions. However,
>> >> i made some changes lately and i store now the skeleton as an
>> >> undirected graph (set of vertices and edges). You could maybe
>> >> play around with some graph algorithms to calculate that measure...
>> >>
>> >> If more people are interested, i can add an option to store
>> >> the skeleton as a JGrapht graph (http://jgrapht.sourceforge.net/).
>> >>
>> >> Just let me know!
>> >
>> > In order to measure the largest branch, you need to implement the
>> > Dijkstra algorithm.
>> > Since this already is implemented in jgrapht, it would be very
>> > convenient when the skeleton can be stored as jgrapht.
>> >
>> > I also need this functionality.
>> >
>> > So ignacio, yes please!!!!!!!!!!!!!!
>> >
>> > Kindly,
>> >
>> > Gerrit.
>> >
>>
>>
>>
>> --
>> Ignacio Arganda-Carreras, Ph.D.
>> Institute of Neuroinformatics
>> Uni/ETH Zurich
>> Winterthurerstrasse 190
>> Bau 55
>> Zurich 8057
>> [phone]
>>
>> Website: http://bioweb.cnb.csic.es/~iarganda/
>>
>

--
Ignacio Arganda-Carreras, Ph.D.
Institute of Neuroinformatics
Uni/ETH Zurich
Winterthurerstrasse 190
Bau 55
Zurich 8057
[phone]

Website: http://bioweb.cnb.csic.es/~iarganda/