Particle Analysis

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Particle Analysis

Rohitesh Gupta
Hi Guys,

I was following a document related to imagej earlier for particle analysis.

https://www.unige.ch/medecine/bioimaging/files/1914/1208/6000/Quantification.pdf

Somehow this document seems to give inaccurate results. Could anyone help me calculate the right particle counts on the attached image. Any help will be much appreciated.

Thanks,
Rohitesh


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

Picture1.jpg (38K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Particle Analysis

Rohitesh Gupta
Sometimes I am getting this error as well. Checked online and found that this problem has existed on imagej since ages.

“No particles were detected. The threshold (255-255) may not be correct”

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: Particle Analysis

Rohitesh Gupta
In reply to this post by Rohitesh Gupta
Sorry for the trouble but the imagej -> analyze particles on windows 10 is giving error but the one on my macbook is working fine. I don't understand if I need to reinstall the software as I have upgraded my imagej but still the problem persists. Is there a way to look into the code of analyze particles and correct it ?

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: Particle Analysis

John Minter
In reply to this post by Rohitesh Gupta
Rohitesh, how do you really know the particle count in this image? What is
your model for this system? Is it a distribution of spheroidal particles
that are agglomerated? What is really important here: count or size (and
size distribution)?

I ask this because the particles look like agglomerated spheroidal
particles to me. I saw a lot of similar problems in my career. Many of your
"agglomerates" do not have significant intrusions. These will not be
separated by a watershed algorithm. If that occurs, your "particle count"
will need to be corrected. How do you plan on doing this?

Many of the boundaries appear diffuse and particles of similar diameter
have different contrast. This could imply a depth of field problem or
perhaps the colorant is not present at the same concentration in all
particles. How do you plan on handling this question? You probably also
want to record an "empty field" image to test the uniformity of your
illumination. I would also suggest submitting images in png format instead
of jpeg because png is a lossless compression.

The particles look magenta to me and so extracting the green channel (the
complementary color) helps contrast.

With image analysis one always gets an answer. Sometimes that answer is
misleading. The key is to anticipate sources of bias and deviation from the
model for the system to minimize errors in interpretation.

If you want to continue dialog with the community, I suggest you move the
discussion to the ImageJ Forum [https://forum.image.sc/] so participants
can better follow the discussion.

Best Regards,
John Minter
Retired from Kodak Analytical Sciences

On Fri, May 10, 2019 at 2:54 PM Rohitesh Gupta <[hidden email]>
wrote:

> Hi Guys,
>
> I was following a document related to imagej earlier for particle
> analysis.
>
>
> https://www.unige.ch/medecine/bioimaging/files/1914/1208/6000/Quantification.pdf
>
> Somehow this document seems to give inaccurate results. Could anyone help
> me calculate the right particle counts on the attached image. Any help will
> be much appreciated.
>
> Thanks,
> Rohitesh
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: Particle Analysis

Rohitesh Gupta
In reply to this post by Rohitesh Gupta
Thank you for your email.

I am trying to convey that the analyze particles feature on windows 10  in my office is underestimating the particle counts. In this case, certainly the particle count will be based on the spheroids generated in the picture. Also, using "adjustable watershed" further demarcations on the image can be introduced. That will improve the results but certainly won't give 100% accurate results. I have followed a document on particle count and have found it to work well. When I have to compare two different systems where one is control and the other is knockout or knockdown, then one can perform the analysis and get promising results. Infact, even if there are marginal differences between the two samples, one can analyze using "particle analyzer" provided the image is crystal clear.

Thanks,
Rohitesh

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