Manually seeded nuclei segmentation/ Marker controlled watershed plugin

[MrScientistman2]

Hi all,


Here is what I am looking for:


I want to manually or semi-automatically mark the aproximate center of cell nuclei in 2D grey-scale images, then I want to find a plugin that could generate a segmentation of these nuclei, for example as ROIs in the ROI manager. Attached is an example image of nuclei that is similar to what I will be using. The end-goal is to measure the intensity of different antibody staining colors in individual nuclei (extra channels not in attached image).


It seems so hard to find a working plugin that takes MANUAL seed input!


Some options I've tried:


http://fiji.sc/Marker-controlled_Watershed

This plugin also claims to be fit for my problem, but I don't understand how it works. How should the marker input image be? Should it contain ROIs, 255-value pixels on a 0-value background, some RGB input markers or something else? It's really frustrating. Whatever I'm trying its just outputting a totally black image as the result.


http://imagej.nih.gov/ij/plugins/inserm514/Documentation/Watersheds_514_V2/Watersheds_514_V2.html

This plugin by Cutrona J. and Bonnet Noel seems to do excately what I want, but for some reason it doesnt work on my computer/ImageJ/FIJI. Is it due to not working on newer Javas or something?


Best wishes,

Anders


--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Ignacio Arganda-Carreras-2]

Hello Anders,

Regarding the marker-controlled watershed plugin, you need to introduce the
following images:

- Input image: usually the gradient version of the image you want to
segment (because the watershed floods from the low intensity values).
- Marker image: this is the seeds image. It needs to be of the same size as
the input image and containing the labeled seeds to start the flooding
from. For example, if you have a 512x512 grayscale input image, you can use
a 512x512 16-bit image with dots on each seed, where each dot needs to have
a different integer intensity value.

I'm attaching an example of the images so you can see what I mean (input
image = nuclei-Gradient.png, marker image = seeds-lbl.png, result =
nuclei-Gradient-watershed.png). Let me know if you need further help.

Best regards,

ignacio



On Sun, Sep 27, 2015 at 9:53 PM, Anders Lunde <[hidden email]>
wrote:

> Hi all,
>
>
> Here is what I am looking for:
>
>
> I want to manually or semi-automatically mark the aproximate center of
> cell nuclei in 2D grey-scale images, then I want to find a plugin that
> could generate a segmentation of these nuclei, for example as ROIs in the
> ROI manager. Attached is an example image of nuclei that is similar to what
> I will be using. The end-goal is to measure the intensity of different
> antibody staining colors in individual nuclei (extra channels not in
> attached image).
>
>
> It seems so hard to find a working plugin that takes MANUAL seed input!
>
>
> Some options I've tried:
>
>
> http://fiji.sc/Marker-controlled_Watershed
>
> This plugin also claims to be fit for my problem, but I don't understand
> how it works. How should the marker input image be? Should it contain ROIs,
> 255-value pixels on a 0-value background, some RGB input markers or
> something else? It's really frustrating. Whatever I'm trying its just
> outputting a totally black image as the result.
>
>
>
> http://imagej.nih.gov/ij/plugins/inserm514/Documentation/Watersheds_514_V2/Watersheds_514_V2.html
>
> This plugin by Cutrona J. and Bonnet Noel seems to do excately what I
> want, but for some reason it doesnt work on my computer/ImageJ/FIJI. Is it
> due to not working on newer Javas or something?
>
>
> Best wishes,
>
> Anders
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
Ignacio Arganda-Carreras, Ph.D.
Ikerbasque Research Fellow
Departamento de Ciencias de la Computacion e Inteligencia Artificial
Facultad de Informatica, Universidad del Pais Vasco
Paseo de Manuel Lardizabal, 1
20018 Donostia-San Sebastian
Guipuzcoa, Spain

Phone : [phone]
Website: http://sites.google.com/site/iargandacarreras/
<http://biocomp.cnb.csic.es/~iarganda/index_EN.html>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[MrScientistman2]

Thanks Ignacio! This looks very good. What would be a simple way to convert the segmentation result to ROIs in the ROI-manager? And how did you create the gradient image? Does this plugin work a lot better on gradient images, or might original grey scale work as good?

Best,
Anders

[MrScientistman2]

And also, what method did you use to create the markers? Manually?

[Ignacio Arganda-Carreras-2]

Hello again, Anders,

Here you are a macro with the steps I took to create the gradient image and
the labeled seeds, apply the seeded watershed and convert the result to
ROIs.

It assumes you have already opened the image "8 bit nuclei.tif" and you
have MorphoLibJ installed. I hope this makes things more clear :)

// find seeds in original image
run("Find Maxima...", "noise=20 output=[Point Selection]");

// create seed image
newImage("seeds", "8-bit black", 1024, 1024, 1);
setForegroundColor(255, 255, 255);
run("Restore Selection");
run("Draw");
// label seeds
run("Connected Components Labeling", "connectivity=4 type=[8 bits]");

// calculate morphological gradient of input image
selectWindow("8 bit nuclei.tif");
run("Morphological Filters", "operation=Gradient element=Octagon radius=2");

// apply marker-controlled watershed using the gradient image and the
labeled seeds
run("Marker-controlled Watershed", "input=8-Gradient marker=seeds-lbl
mask=None calculate");

// convert labels to ROIs
selectWindow("8-Gradient-watershed");
run("Grays");
setThreshold(0, 0);
run("Convert to Mask");
run("Create Selection");

On Mon, Sep 28, 2015 at 1:12 PM, MrScientistman2 <
[hidden email]> wrote:

> And also, what method did you use to create the markers? Manually?
>
>
>
> --
> View this message in context:
> http://imagej.1557.x6.nabble.com/Manually-seeded-nuclei-segmentation-Marker-controlled-watershed-plugin-tp5014479p5014483.html
> Sent from the ImageJ mailing list archive at Nabble.com.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
Ignacio Arganda-Carreras, Ph.D.
Ikerbasque Research Fellow
Departamento de Ciencias de la Computacion e Inteligencia Artificial
Facultad de Informatica, Universidad del Pais Vasco
Paseo de Manuel Lardizabal, 1
20018 Donostia-San Sebastian
Guipuzcoa, Spain

Phone : [phone]
Website: http://sites.google.com/site/iargandacarreras/
<http://biocomp.cnb.csic.es/~iarganda/index_EN.html>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html