particle analyzer

[Christian Scheit]

Hi,

I am using "Analyze Particles" to count particles in my images.

I have now images with particles which are in focus and particles which
are not.
I only want to count the particles which are in focus. With a normal
threshold image I am not able to separate. With "Find edges" it is also
not possible to distinguish very well.

Is there any plugin to find focused particles in an image?
What else can I do to distinguish between particles in focus and out of
focus?

I have attached a sample image.

Thanks a lot in advance.

Christian

[John Minter-3]

Christian, as I examined your image the question that came to mind
was, 'Would any particle analysis on this image provide reliable,
meaningful results?' Without more specific information on the material
and imaging conditions, I can't give you definitive advice. I will
instead pose some questions for your consideration.

You need to convince yourself and any client or referee that you can
explain all the morphologies present in the image and how your choices
affect the results. As one who has measured size distributions of many
colloidal materials, I would like to know clearly what is a "single
particle" and what is an "agglomerate" and how that decision
influences your results. In the materials I analyze, in most (but not
all) cases single particles are convex. In those cases, blobs with
large inclusions are agglomerates. These can often be distinguished by
considering the difference in the area of the blob and the convex
hull. In your images I detect many features that are largely convex
and many that would be classified as agglomerates in our materials.
How does your decision of how to handle these "agglomerates" influence
the results? How is this relevant to your particular problem?

The second issue that I see is the broad range of areas of the
features I see. In such cases, I have found it very difficult to get
reliable segmentation of large and small features in the same image.
There are some image processing techniques that can help, but in most
cases I found that working to improve specimen preparation provides
superior results. If you can achieve it without changing morphology
(you need to understand the chemistry), I would consider trying a more
dilute sample.

You mention some features are "out of focus". You need to consider
whether this is produced by a limited depth of field (are these just
thick regions?) and whether some contaminate (i.e. a covering of some
residual binder) is producing this (I've seen this in backscattered
electron images of electron dense materials covered with polymeric
binders.) In either case, you will need to convince yourself and your
client/reviewer that ignoring these is not biasing the measurement.

In any case, replicates are your friend...

I hope this helps,

Best regards,
John Minter
Kodak Analytical Sciences - Microscopy & Spectroscopy Lab


On Sat, Feb 25, 2012 at 4:30 PM, Christian Scheit <[hidden email]> wrote:

> Hi,
>
> I am using "Analyze Particles" to count particles in my images.
>
> I have now images with particles which are in focus and particles which are
> not.
> I only want to count the particles which are in focus. With a normal
> threshold image I am not able to separate. With "Find edges" it is also not
> possible to distinguish very well.
>
> Is there any plugin to find focused particles in an image?
> What else can I do to distinguish between particles in focus and out of
> focus?
>
> I have attached a sample image.
>
> Thanks a lot in advance.
>
> Christian

[Gabriel Landini]

On Saturday 25 Feb 2012 21:30:09 Christian Scheit wrote:
> I have now images with particles which are in focus and particles which
> are not.
> I only want to count the particles which are in focus. With a normal
> threshold image I am not able to separate. With "Find edges" it is also
> not possible to distinguish very well.

If some are in and others out of focus, it means that they are in different
positions in the z axis. I do not think you could resolve *overlapping* semi
transparent particles at different z positions in a single shot.

> What else can I do to distinguish between particles in focus and out of
> focus?

Use a confocal microscope to get the z positions of objects.

Or reduce the thickness of the specimen. It seems to me that you are trying to
do whole particle analysis by their semi transparent silhouettes.

Or, if particles are *not* overlapping, you could take a number of shots at
different z positions and use the extending depth of focus plugin to get
everything focused, but semi-transparent specimens are the ones that cannot be
processed well with that approach.

In any case, you might want to improve image quality. There is a lot of high
frequency noise (maybe courtesy of the jpeg encoding) and the image data uses
just about 1/3 of the channel capacity. You could improve this quite a bit.

Cheers
Gabriel