3D roi extension

[Paola Lepanto]

Hello,

I would like to quantify somehow the fluorescence intensity around particles (0.5um or so around the particle surface), for example lysosomal membrane protein mark around bacteria in 3D confocal stacks.

Our initial idea was to detect particles (bacteria) using 3D Objects Counter, add particles to the 3D Roi Manager, somehow extend the selection border of the particle to include nearby areas and finally using this new extended selection to measure intensity in the lysosomal channel.

My question is: is it possible to extend a 3D selection and to manage it as a new 3D object?

Any other idea to quantify fluorescence around 3D particles will be welcome.

Thanks in advance

Paola

[dscho]

Hi Paola,

On Mon, 6 Feb 2012, Paola Lepanto wrote:

> I would like to quantify somehow the fluorescence intensity around
> particles (0.5um or so around the particle surface), for example
> lysosomal membrane protein mark around bacteria in 3D confocal stacks.
>
> Our initial idea was to detect particles (bacteria) using 3D Objects
> Counter, add particles to the 3D Roi Manager, somehow extend the
> selection border of the particle to include nearby areas and finally
> using this new extended selection to measure intensity in the lysosomal
> channel.
>
> My question is: is it possible to extend a 3D selection and to manage it
> as a new 3D object?

In the short-term, I would recommend using TrakEM2: http://fiji.sc/TrakEM2
since it has powerful tools for multi-dimensional regions of interest.

In the long-term, ImageJ2 will solve the problem for you.

Ciao,
Johannes

[Thomas Boudier]

Hi Paola,

In case your objects are not touching you can simply use a dilatation,
i.e. use a maximum filter of the radius you want to extend your objects,
before adding the image to 3Dmanager.

http://imagejdocu.tudor.lu/doku.php?id=plugin:filter:3d_filters:start

Thomas


Le 06/02/2012 20:14, Paola Lepanto a écrit :

> Hello,
>
> I would like to quantify somehow the fluorescence intensity around particles (0.5um or so around the particle surface), for example lysosomal membrane protein mark around bacteria in 3D confocal stacks.
>
> Our initial idea was to detect particles (bacteria) using 3D Objects Counter, add particles to the 3D Roi Manager, somehow extend the selection border of the particle to include nearby areas and finally using this new extended selection to measure intensity in the lysosomal channel.
>
> My question is: is it possible to extend a 3D selection and to manage it as a new 3D object?
>
> Any other idea to quantify fluorescence around 3D particles will be welcome.
>
> Thanks in advance
>
> Paola
>

--
   /**********************************************************/
      Thomas Boudier, MCU Université Pierre et Marie Curie,
      Modélisation Cellulaire et Imagerie Biologique (EE1),
      IFR 83, Bat B 7ème étage, porte 709, Campus Jussieu.
      Tel : [phone]   Fax : [phone]
/*******************************************************/

[Thomas Boudier]

Hi Johannes,

About ImageJ2 ca you give us some ideas about how you are planning  to
implement 3D selection ? I would be more than happy to contibute to this
task or use your libraries.

best,

Thomas


Le 06/02/2012 20:55, Johannes Schindelin a écrit :

> Hi Paola,
>
> On Mon, 6 Feb 2012, Paola Lepanto wrote:
>
>> I would like to quantify somehow the fluorescence intensity around
>> particles (0.5um or so around the particle surface), for example
>> lysosomal membrane protein mark around bacteria in 3D confocal stacks.
>>
>> Our initial idea was to detect particles (bacteria) using 3D Objects
>> Counter, add particles to the 3D Roi Manager, somehow extend the
>> selection border of the particle to include nearby areas and finally
>> using this new extended selection to measure intensity in the lysosomal
>> channel.
>>
>> My question is: is it possible to extend a 3D selection and to manage it
>> as a new 3D object?
> In the short-term, I would recommend using TrakEM2: http://fiji.sc/TrakEM2
> since it has powerful tools for multi-dimensional regions of interest.
>
> In the long-term, ImageJ2 will solve the problem for you.
>
> Ciao,
> Johannes
>

--
   /**********************************************************/
      Thomas Boudier, MCU Université Pierre et Marie Curie,
      Modélisation Cellulaire et Imagerie Biologique (EE1),
      IFR 83, Bat B 7ème étage, porte 709, Campus Jussieu.
      Tel : [phone]   Fax : [phone]
/*******************************************************/

[dscho]

Hi Thomas,

On Tue, 7 Feb 2012, Thomas Boudier wrote:

> About ImageJ2 ca you give us some ideas about how you are planning  to
> implement 3D selection ? I would be more than happy to contibute to this
> task or use your libraries.

This is indeed a hot topic. At our cross-project hackathon in December, we
discussed this subject quite intensely.

One point I found striking was that "selections" is not really one thing.
There are

- regions of interest; possibly more than 2-dimensional, and characterized
  by a vast range of methods:

        - thresholding
        - explicit regions (i.e. having binary mask images)
        - explicit programmatic regions (e.g. thresholding a second
          channel)
        - implicit regions (e.g. defining a sphere by center and radius)
        - filtered selections (e.g. filter connected components of
          selected pixels by morphological features, think: Particle
          Analyzer's filter, but in a way that can be passed on to other
          processing modules)
        - ...

- labeling; every pixel can belong to more than one label, think
  hierarchies of labels e.g. "cell", "membrane", "mitochondrion", etc

- overlays; these also might need to be higher-dimensional; think of the
  isosurface mode in the 3D Viewer.

- renderings (i.e. when you want to "burn" the outlines into an image,
  e.g. for publication)

Now, as we try to come up with a clean architecture (one that will not
limit us later), what we really want to have is interfaces that kind of
describe the features semantically, and that are certainly not mingled
with UI-specific code.

One distinction that needs to be reflected in that architecture is that
sub-pixel resolution makes sense for overlays, not at all for renderings,
and only for some regions of interest. (So there needs to be some sort of
hierarchy, think: interface SampledRegionOfInterest extends
RealRandomAccessible<BitType>)

And that discussion does not even address the issue of hyperplanes (i.e.
selections that are of a lower dimension than the image, such as linear
selections).

In any case, the state of the art in ImgLib2 (not really being updated
after the discussions from the hackathon) can be found here:

http://fiji.sc/cgi-bin/gitweb.cgi?p=imglib.git;a=tree;f=imglib2/core/src/main/java/net/imglib2/roi;h=92c81c20be96176a64cce8285242bdebaf627f38;hb=6358ca00a9b388a3be62f28b50b588707a33e67e

and here:

http://fiji.sc/cgi-bin/gitweb.cgi?p=imglib.git;a=tree;f=imglib2/core/src/main/java/net/imglib2/labeling;h=04b1806e6b7376689ab0973ebc1f2fa7f257f595;hb=6358ca00a9b388a3be62f28b50b588707a33e67e

Curtis summarized the discussions from the hackathon better than me, and
he put the information (pretty technical, so no IJ1 mailing list material;
oops, I think this mail falls into that category, too...) here:

https://groups.google.com/forum/#!topic/scijava/Gp5C10NySFs

Ciao,
Johannes