Re: How/where in the code are segmentations saved?

Posted by Sean Burke on Jun 07, 2011; 3:50pm
URL: http://imagej.273.s1.nabble.com/How-where-in-the-code-are-segmentations-saved-tp3684345p3684349.html

Along these lines is it possible to save the ROI or segmentation so that the positions and shapes of the segments can be reconstructed in another piece of software?




----------------------------------------------------
Sean Burke
Technical Applications Specialist
De Novo Software
[hidden email]
Phone: [phone] Ext. 806

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Johannes Schindelin
Sent: Sunday, June 05, 2011 10:29 AM
To: [hidden email]
Subject: Re: How/where in the code are segmentations saved?

Hi Kynn,

On Sun, 5 Jun 2011, Johannes Schindelin wrote:

> On Sun, 5 Jun 2011, Kynn Jones wrote:
>
> > On Sun, Jun 5, 2011 at 4:14 AM, Johannes Schindelin
> > <[hidden email]> wrote:
> >
> > > On Sat, 4 Jun 2011, Kynn Jones wrote:
> > >
> > >> I understand that ImageJ offers the possibility of saving to disk
> > >> the result of an image segmentation.
> > >
> > > Are you referring to ROIs?
> >
> > Not sure.  What I mean is this: suppose you run some image
> > segmentation.
>
> Well, if you give me a concrete example what plugin you run, then I
> can tell you the nature of the output. Frequently it is just binary
> (or
> indexed) images. In which case saving/loading is the same as with
> other images.
>
> > The result of this segmentation can be used in further analyses, so
> > one can think of them as "intermediate results".  It would be useful
> > to be able to save these intermediate segmentation results to disk,
> > and reload them later, so that various analyses can be performed on
> > the same intermediate segmentation results at different times.  If
> > these intermediate segmentation results are expressed, somehow, as a
> > collection of ROIs, then yes this is what I mean.  In this case, my
> > question reduces to "how are ROIs serialized?"
> >  I'll look at the links you posted.  Thanks!
>
> There is a quite thorough concept of ROIs (possibly overlapping,
> multiple classes, iterators, etc) in ImageJ2. Maybe you want to have a
> look:
>
> http://fiji.sc/cgi-bin/gitweb.cgi?p=imagej2/.git;a=tree;f=core/data/sr
> c/main/java/imagej/data/roi;h=620598150f9ea14ee4f6cf274c730680ea204a03
> ;hb=3d8fe33cd541d021b677fe4ccabde8977e25ca57
>
> Note, however, that this ROI framework is not yet in the form where
> you can expect things to Just Work. But it gives you an idea of the
> future direction.

Of course I forgot to add the link to the actual data structures:

http://fiji.sc/cgi-bin/gitweb.cgi?p=imglib.git;a=tree;f=imglib2/core/src/main/java/net/imglib2/roi;h=517700ada29c3ed1b3337ade8bc2a8dc0a913f8f;hb=464f3be5f7bddfc33069dc574f1afe4a8316a63c

Ciao,
Johannes