Versatile Wand for Multispectral Image Stacks?

[Bill Beggs]

Hi everyone.

I use ImageJ for segmentation of 8-big grayscale multispectral images produced from energy dispersive x-ray analysis of heterogeneous inorganic powders.   So for example, I collect a set of images of the same scene, but each individual image is produced from characteristic xrays for Ca, Si, Al, Fe, Mg etc.

The individual particles that compose these powders have variable chemical composition and I classify the particles into families of similar composition for quantification and other analyses.

I'm familiar with the various segmentation plugins and I find Weka
Segmentation cumbersome unfortunately.

What I would really like to see is a tool similar to Versatile Wand, but
that operates on a stack of images.  Essentially, from a selection, it
would calculate something like a z-axis profile and select all other
locations that exibited a similar profile within adjustable tolerances.

Does anyone know if something like this exists?

Many thanks.

-Bill.

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

[Marcel]

I don't know if that could help you because the WEKA plugin is already an easy to use classification tool but I developed some functions in my application to transfer selections in opened image layers (opened images or stacks) to R and then classify them. From the results you can produce a plot in R or an image (map) in ImageJ.

Here are some examples:

Unsupervised classification GUI (similar to kmeans with GUI):

http://www.uni-bielefeld.de/biologie/Oekosystembiologie/bio7app/flashtut/cluster.htm

Supervised classification (R, G, B spectral bands):

https://www.youtube.com/watch?v=t3OOEZ5sBLY

Cluster Analysis (Particles):

https://www.youtube.com/watch?v=gUQRqfh7ZNc

[jerie]

Hi Bill,

How about creating a montage and working on all images simultaneously?

Greetings, Jens

Visiting Scientist
Fundação Oswaldo Cruz, Centro de Desenvolvimento Tecnológico em Saúde
(CDTS), Rio de Janeiro,  Brazil.
Am 08.03.2016 22:46 schrieb "Bill Beggs" <[hidden email]>:

> Hi everyone.
>
> I use ImageJ for segmentation of 8-big grayscale multispectral images
> produced from energy dispersive x-ray analysis of heterogeneous inorganic
> powders.   So for example, I collect a set of images of the same scene, but
> each individual image is produced from characteristic xrays for Ca, Si, Al,
> Fe, Mg etc.
>
> The individual particles that compose these powders have variable chemical
> composition and I classify the particles into families of similar
> composition for quantification and other analyses.
>
> I'm familiar with the various segmentation plugins and I find Weka
> Segmentation cumbersome unfortunately.
>
> What I would really like to see is a tool similar to Versatile Wand, but
> that operates on a stack of images.  Essentially, from a selection, it
> would calculate something like a z-axis profile and select all other
> locations that exibited a similar profile within adjustable tolerances.
>
> Does anyone know if something like this exists?
>
> Many thanks.
>
> -Bill.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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

[Christian Liebig-3]

Hi Bill,

maybe spectral similarity mapping would do the trick. You give a
reference spectrum and then each pixel in the hyperspectral image is
assigned a similarity value. If you then threshiold it should give you
the desired output.

I have a macroset doing this. It is functional but still under
development and has no documentation. But if you think it is what you
need I can help you setting it up.
You can download the set here
<https://owncloud.tuebingen.mpg.de/public.php?service=files&t=fe31e6513aeb00a8c3070cf5f60cdd83>.
What you need is the zNormaliser and SpectralSimilarity.

Best,
Christian


On 09.03.2016 02:34, Bill Beggs wrote:

> Hi everyone.
>
> I use ImageJ for segmentation of 8-big grayscale multispectral images produced from energy dispersive x-ray analysis of heterogeneous inorganic powders.   So for example, I collect a set of images of the same scene, but each individual image is produced from characteristic xrays for Ca, Si, Al, Fe, Mg etc.
>
> The individual particles that compose these powders have variable chemical composition and I classify the particles into families of similar composition for quantification and other analyses.
>
> I'm familiar with the various segmentation plugins and I find Weka
> Segmentation cumbersome unfortunately.
>
> What I would really like to see is a tool similar to Versatile Wand, but
> that operates on a stack of images.  Essentially, from a selection, it
> would calculate something like a z-axis profile and select all other
> locations that exibited a similar profile within adjustable tolerances.
>
> Does anyone know if something like this exists?
>
> Many thanks.
>
> -Bill.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
Christian Liebig, PhD

Light Microscopy Facility
Max-Planck-Institute for Developmental Biology
Spemannstrasse 35
D-72076 Tuebingen
Germany

Phone:  [phone]
e-mail: [hidden email]


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

[Bill Beggs]

Thanks Christian.  That sounds promising.

I'll check it out.

-Bill

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Christian Liebig
Sent: Thursday, March 10, 2016 5:19 AM
To: [hidden email]
Subject: Re: Versatile Wand for Multispectral Image Stacks?

Hi Bill,

maybe spectral similarity mapping would do the trick. You give a reference spectrum and then each pixel in the hyperspectral image is assigned a similarity value. If you then threshiold it should give you the desired output.

I have a macroset doing this. It is functional but still under development and has no documentation. But if you think it is what you need I can help you setting it up.
You can download the set here
<https://owncloud.tuebingen.mpg.de/public.php?service=files&t=fe31e6513aeb00a8c3070cf5f60cdd83>.
What you need is the zNormaliser and SpectralSimilarity.

Best,
Christian


On 09.03.2016 02:34, Bill Beggs wrote:

> Hi everyone.
>
> I use ImageJ for segmentation of 8-big grayscale multispectral images produced from energy dispersive x-ray analysis of heterogeneous inorganic powders.   So for example, I collect a set of images of the same scene, but each individual image is produced from characteristic xrays for Ca, Si, Al, Fe, Mg etc.
>
> The individual particles that compose these powders have variable chemical composition and I classify the particles into families of similar composition for quantification and other analyses.
>
> I'm familiar with the various segmentation plugins and I find Weka
> Segmentation cumbersome unfortunately.
>
> What I would really like to see is a tool similar to Versatile Wand,
> but that operates on a stack of images.  Essentially, from a
> selection, it would calculate something like a z-axis profile and
> select all other locations that exibited a similar profile within adjustable tolerances.
>
> Does anyone know if something like this exists?
>
> Many thanks.
>
> -Bill.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
Christian Liebig, PhD

Light Microscopy Facility
Max-Planck-Institute for Developmental Biology Spemannstrasse 35
D-72076 Tuebingen
Germany

Phone:  [phone]
e-mail: [hidden email]


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

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