Segmentation

[Sylma]

Hello
I work on histological slices of meat products (minced meat...). I look for
a way to separate intact muscle fibers and destructured fibers. Thresholding
don’t give me good results. I also like isolate lipid vesicles but the
outline is not always well defined, and the function "watershed" don’t
satisfy me. Could you give me some ideas? what plugins or macros I can use.

Thank you and happy new year

[Johan Henriksson-2]

Sylma wrote:
> Hello
> I work on histological slices of meat products (minced meat...). I look for
> a way to separate intact muscle fibers and destructured fibers. Thresholding
> don’t give me good results. I also like isolate lipid vesicles but the
> outline is not always well defined, and the function "watershed" don’t
> satisfy me. Could you give me some ideas? what plugins or macros I can use.
>
> Thank you and happy new year
>  
supply sample images

/Johan

[Sylma]

Hello

I send you an image to solve my problem : lipid vesicles are white

Best regards

[Sylma]

Hello Mario

It's not really what I need.
In fact, I want to isolate each lipid vesicle to count and measure them.
I tried to use Watershed but vesicles are too fragmented. Can I modify some
parameters of this fonction to obtain a better result?

Thanks

[Mario Guarracino]

Attached you find the segmentation obtained with Yawi3d plugin (http://yawi3d.sourceforge.net).
All the best,
Mario

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Sylma
Sent: Thursday, January 07, 2010 9:31 AM
To: [hidden email]
Subject: Re: Segmentation


Hello

I send you an image to solve my problem : lipid vesicles are white

Best regards

[Michael Schmid]

Hi Sylma (Marie?),

you can use the standard thresholding tool if you subtract the  
background.
E.g.
   run("8-bit");
   run("Subtract Background...", "rolling=500 light sliding");
   setThreshold(0, 240);

For removing residual noise from the result, you can use, e.g.,  
"Analyze Particles" with "Mask" output and a certain minimum size of  
the background objects. An alternative is
   run("Minimum...", "radius=1.5");
   run("Maximum...", "radius=1.5");

Finally, invert the images to get the bright objects. If required,  
use a further "Minimum, Maximum" sequence for smoother borders.

Michael
________________________________________________________________

On 7 Jan 2010, at 09:31, Sylma wrote:

>
> Hello
>
> I send you an image to solve my problem : lipid vesicles are white
>
> Best regards<1.jpg>

[Michael Schmid]

Hi Sylma,

look at the following thread - it describes how to have an equivalent  
to Watershed with an additional parameter:

   https://list.nih.gov/cgi-bin/wa?A2=ind0909&L=IMAGEJ&P=R28684&I=-3

Michael
________________________________________________________________

On 7 Jan 2010, at 13:29, Sylma wrote:

> Hello Mario
>
> It's not really what I need.
> In fact, I want to isolate each lipid vesicle to count and measure  
> them.
> I tried to use Watershed but vesicles are too fragmented. Can I  
> modify some
> parameters of this fonction to obtain a better result?
>
> Thanks

[Daniel Hornung]

Sylma wrote:
> Hello Mario
>
> It's not really what I need.
> In fact, I want to isolate each lipid vesicle to count and measure them.
> I tried to use Watershed but vesicles are too fragmented. Can I modify some
> parameters of this fonction to obtain a better result?

Hello,
I must admit that I'm neither a physician nor a biologist, but not even I (as a
human) could alway tell, where as structure should be classified as one
connected or two separate components, as sketched in the attachment.

Daniel