ImageJ

Weka Trainable Segmentation alternative

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

8 messages Options

Adrián Villalba

Weka Trainable Segmentation alternative

Dear all,

I am trying to use the Weka Trainable Segmentation plugin in order to
classify islets of Langerhans in Hematoxilin/Eosin tissues. There are four
different types of islets depending on insulitis (immune cell infiltration
within the islet) as i show you in an attached JPG-picture (just to show
you the expected result, not to manipulate).

My goal is to do it automatically in imageJ, rather tan manual scoring of
pictures. So i thought it would be a good idea to use the Trainable Weka
Segmentation plugin just to train the algorithm to do it automatically but
it fails. (I cannot attach the classifier.model archive because it is
rejected by the mailing list conditions).

I think that maybe it is not a proble for the Weka, instead being a
conceptual problem and that maybe you could know another imageJ tool to
pursuit that goal.

Thank you very much in advance!

--

- Adrián Villalba Felipe.
https://es.linkedin.com/in/adrianvillalba

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

th.jpg (18K) Download Attachment

Adrián Villalba

Re: Weka Trainable Segmentation alternative

wEKA_INSULITIS0004.tif
<https://drive.google.com/file/d/1cAtH2qKeDXdE9uEAg1lC9V2EP98JPbUD/view?usp=drive_web>

wEKA_INSULITIS0005.tif
<https://drive.google.com/file/d/15yHjug2I4gu8rXbRIcEEBGKRrOcgAO51/view?usp=drive_web>

wEKA_INSULITIS0006.tif
<https://drive.google.com/file/d/1VK5YmHAfTHMZwTW7XFnFXMPjkyeiUF72/view?usp=drive_web>

wEKA_INSULITIS0007.tif
<https://drive.google.com/file/d/1Bojzu-_vz8Nut13-zaGCw5CN5wyn1H1i/view?usp=drive_web>

wEKA_INSULITIS0008.tif
<https://drive.google.com/file/d/1uPdOlprtusdaXbsjdGj-6QK-0T6f9ctS/view?usp=drive_web>
Dear all,

Sorry for the example JPG-compressed file. Here I attach some examples of
tiff acquired images. Weka Trainable Segmentation plufin is not very
sensitive in order to split the islet (round shaped) versus the rest of
exocrine tissue. Indeed, every islet is scored different in base of the
surrounding cell layer. Do you know how can i approach this problem?

Thank you in advance,

2018-01-06 0:39 GMT+01:00 Adrián Villalba <[hidden email]>:

> Dear all,
>
> I am trying to use the Weka Trainable Segmentation plugin in order to
> classify islets of Langerhans in Hematoxilin/Eosin tissues. There are four
> different types of islets depending on insulitis (immune cell infiltration
> within the islet) as i show you in an attached JPG-picture (just to show
> you the expected result, not to manipulate).
>
> My goal is to do it automatically in imageJ, rather tan manual scoring of
> pictures. So i thought it would be a good idea to use the Trainable Weka
> Segmentation plugin just to train the algorithm to do it automatically but
> it fails. (I cannot attach the classifier.model archive because it is
> rejected by the mailing list conditions).
>
> I think that maybe it is not a proble for the Weka, instead being a
> conceptual problem and that maybe you could know another imageJ tool to
> pursuit that goal.
>
> Thank you very much in advance!
>
> --
>
> - Adrián Villalba Felipe.
> https://es.linkedin.com/in/adrianvillalba
>
>

--

- Adrián Villalba Felipe.
https://es.linkedin.com/in/adrianvillalba

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

Ignacio Arganda-Carreras-2

Re: Weka Trainable Segmentation alternative

Dear Adrián,

Can you please manually annotate one of your images with the result you
expect so we better understand what you are trying to achieve?

Best regards,

ignacio

On Sat, Jan 6, 2018 at 6:51 PM, Adrián Villalba <[hidden email]> wrote:

>
> wEKA_INSULITIS0004.tif
> <https://drive.google.com/file/d/1cAtH2qKeDXdE9uEAg1lC9V2EP98JP
> bUD/view?usp=drive_web>
>
> wEKA_INSULITIS0005.tif
> <https://drive.google.com/file/d/15yHjug2I4gu8rXbRIcEEBGKRrOcgA
> O51/view?usp=drive_web>
>
> wEKA_INSULITIS0006.tif
> <https://drive.google.com/file/d/1VK5YmHAfTHMZwTW7XFnFXMPjkyeiU
> F72/view?usp=drive_web>
>
> wEKA_INSULITIS0007.tif
> <https://drive.google.com/file/d/1Bojzu-_vz8Nut13-
> zaGCw5CN5wyn1H1i/view?usp=drive_web>
>
> wEKA_INSULITIS0008.tif
> <https://drive.google.com/file/d/1uPdOlprtusdaXbsjdGj-
> 6QK-0T6f9ctS/view?usp=drive_web>
> Dear all,
>
> Sorry for the example JPG-compressed file. Here I attach some examples of
> tiff acquired images. Weka Trainable Segmentation plufin is not very
> sensitive in order to split the islet (round shaped) versus the rest of
> exocrine tissue. Indeed, every islet is scored different in base of the
> surrounding cell layer. Do you know how can i approach this problem?
>
> Thank you in advance,
>
> 2018-01-06 0:39 GMT+01:00 Adrián Villalba <[hidden email]>:
>
> > Dear all,
> >
> > I am trying to use the Weka Trainable Segmentation plugin in order to
> > classify islets of Langerhans in Hematoxilin/Eosin tissues. There are
> four
> > different types of islets depending on insulitis (immune cell
> infiltration
> > within the islet) as i show you in an attached JPG-picture (just to show
> > you the expected result, not to manipulate).
> >
> > My goal is to do it automatically in imageJ, rather tan manual scoring of
> > pictures. So i thought it would be a good idea to use the Trainable Weka
> > Segmentation plugin just to train the algorithm to do it automatically
> but
> > it fails. (I cannot attach the classifier.model archive because it is
> > rejected by the mailing list conditions).
> >
> > I think that maybe it is not a proble for the Weka, instead being a
> > conceptual problem and that maybe you could know another imageJ tool to
> > pursuit that goal.
> >
> > Thank you very much in advance!
> >
> > --
> >
> > - Adrián Villalba Felipe.
> > https://es.linkedin.com/in/adrianvillalba
> >
> >
>
>
> --
>
> - Adrián Villalba Felipe.
> https://es.linkedin.com/in/adrianvillalba
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
Ignacio Arganda-Carreras, Ph.D.
Ikerbasque Research Fellow
Departamento de Ciencia 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 : +34 943 01 73 25
Website: http://sites.google.com/site/iargandacarreras/

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

Adrián Villalba

Re: Weka Trainable Segmentation alternative

In reply to this post by Adrián Villalba

Dear Ignacio,

This is an example image with the different types of islets and insulitis.
The legend is the following:
A = grade 0, no insulitis; B = grade 1, peri-insular; C = grade 2, mild
insulitis (<25%
of the islet infiltrated); D = grade 3, moderate insulitis (25–75% of the
islet
infiltrated); E = grade 4, severe insulitis (>75% islet infiltration).

I would like to compare my template images against a new image
automatically, so i do not have to do it manually in the microscope. But
Weka has not been useful, first of all because Hematoxilin/Eosin staining
is not very different between islets and the rest of tissue. Moreover, the
different score of the 5 types of islets (scored from 0 to 4) is also
tricky for the Weka algorithm. Do you think i can use a different approach?

Thank you very much for your time and attention,

<https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
Libre
de virus. www.avast.com
<https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
<#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

2018-01-06 0:39 GMT+01:00 Adrián Villalba <[hidden email]>:

--

- Adrián Villalba Felipe.
https://es.linkedin.com/in/adrianvillalba

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

Insulitis_example.tif (928K) Download Attachment

Herbie

Re: Weka Trainable Segmentation alternative

Good day Adrián Villalba Felipe,

yesterday I took a closer look at your sample image data and, as you've
already conceded, the staining is not specific enough.

I found that a reasonable contrast of the relevant islets is obtained by
the magenta channel after transforming from RGB to CMYK colour space. In
general, your sample images contain very little colour information, i.e.
if you find a suitable colour space, the information is mostly contained
in a single channel. In order to find the optimum colour space (CMYK
isn't bad but perhaps not optimum) I recommend to experiment with the
"Colour_Deconvolution"-plugin.

Using the magenta channel I wasn't able to find an automatic
thresholding scheme that worked with all of your five sample images. A
threshold-based segmentation of sample image "wEKA_INSULITIS0005.tif"
was impossible even with non-automatic thresholding.

WEKA-classification based on colour may fail due the reasons mentioned
above. Based on structural features, a segmentation may be possible if a
large and adequate training-set is used.

All of the above holds for segmentation of the islets from the surround,
not for the classification of the types of islets which is much more
involved and I doubt that the latter is possible with reasonable success
for the present staining.

In short: A more specific staining appears to be required.

Regards

Herbie

:::::::::::::::::::::::::::::::::::::::::::::
Am 06.01.18 um 23:24 schrieb Adrián Villalba:

> Dear Ignacio,
>
> This is an example image with the different types of islets and insulitis.
> The legend is the following:
> A = grade 0, no insulitis; B = grade 1, peri-insular; C = grade 2, mild
> insulitis (<25%
> of the islet infiltrated); D = grade 3, moderate insulitis (25–75% of the
> islet
> infiltrated); E = grade 4, severe insulitis (>75% islet infiltration).
>
> I would like to compare my template images against a new image
> automatically, so i do not have to do it manually in the microscope. But
> Weka has not been useful, first of all because Hematoxilin/Eosin staining
> is not very different between islets and the rest of tissue. Moreover, the
> different score of the 5 types of islets (scored from 0 to 4) is also
> tricky for the Weka algorithm. Do you think i can use a different approach?
>
>
> Thank you very much for your time and attention,
>
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
> Libre
> de virus. www.avast.com
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
> <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>
> 2018-01-06 0:39 GMT+01:00 Adrián Villalba <[hidden email]>:
>
>> Dear all,
>>
>> I am trying to use the Weka Trainable Segmentation plugin in order to
>> classify islets of Langerhans in Hematoxilin/Eosin tissues. There are four
>> different types of islets depending on insulitis (immune cell infiltration
>> within the islet) as i show you in an attached JPG-picture (just to show
>> you the expected result, not to manipulate).
>>
>> My goal is to do it automatically in imageJ, rather tan manual scoring of
>> pictures. So i thought it would be a good idea to use the Trainable Weka
>> Segmentation plugin just to train the algorithm to do it automatically but
>> it fails. (I cannot attach the classifier.model archive because it is
>> rejected by the mailing list conditions).
>>
>> I think that maybe it is not a proble for the Weka, instead being a
>> conceptual problem and that maybe you could know another imageJ tool to
>> pursuit that goal.
>>
>> Thank you very much in advance!

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

Ignacio Arganda-Carreras-2

Re: Weka Trainable Segmentation alternative

In reply to this post by Adrián Villalba

Dear Adrián,

From your description I understand that you are trying to perform image
classification not segmentation. The Trainable Weka Segmentation transforms
the segmentation problem into a pixel classification problem, so every
pixel (and not the entire image) gets assigned a class. You might want to
try a different approach.

Best,

ignacio

On Sat, Jan 6, 2018 at 11:24 PM, Adrián Villalba <[hidden email]>
wrote:

> Dear Ignacio,
>
> This is an example image with the different types of islets and insulitis.
> The legend is the following:
> A = grade 0, no insulitis; B = grade 1, peri-insular; C = grade 2, mild
> insulitis (<25%
> of the islet infiltrated); D = grade 3, moderate insulitis (25–75% of the
> islet
> infiltrated); E = grade 4, severe insulitis (>75% islet infiltration).
>
> I would like to compare my template images against a new image
> automatically, so i do not have to do it manually in the microscope. But
> Weka has not been useful, first of all because Hematoxilin/Eosin staining
> is not very different between islets and the rest of tissue. Moreover, the
> different score of the 5 types of islets (scored from 0 to 4) is also
> tricky for the Weka algorithm. Do you think i can use a different approach?
>
>
> Thank you very much for your time and attention,
>
>
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> Libre
> de virus. www.avast.com
> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
> <#m_-9005445674115851828_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>
> 2018-01-06 0:39 GMT+01:00 Adrián Villalba <[hidden email]>:
>
>> Dear all,
>>
>> I am trying to use the Weka Trainable Segmentation plugin in order to
>> classify islets of Langerhans in Hematoxilin/Eosin tissues. There are four
>> different types of islets depending on insulitis (immune cell infiltration
>> within the islet) as i show you in an attached JPG-picture (just to show
>> you the expected result, not to manipulate).
>>
>> My goal is to do it automatically in imageJ, rather tan manual scoring of
>> pictures. So i thought it would be a good idea to use the Trainable Weka
>> Segmentation plugin just to train the algorithm to do it automatically but
>> it fails. (I cannot attach the classifier.model archive because it is
>> rejected by the mailing list conditions).
>>
>> I think that maybe it is not a proble for the Weka, instead being a
>> conceptual problem and that maybe you could know another imageJ tool to
>> pursuit that goal.
>>
>> Thank you very much in advance!
>>
>> --
>>
>> - Adrián Villalba Felipe.
>> https://es.linkedin.com/in/adrianvillalba
>>
>>
>
>
> --
>
> - Adrián Villalba Felipe.
> https://es.linkedin.com/in/adrianvillalba
>
>

Stoyan Pavlov

Re: Weka Trainable Segmentation alternative

In reply to this post by Adrián Villalba

Dear Adrian,
As mentioned before Weka Segmentation performs pixel classification, so the
best result you can hope to achieve is to segment islets from the exocrine
tissue, but not to classify the islets. Ilastik ( the interactive learning
and segmentation toolkit; http://ilastik.org/ ) is a tool for image
classification and segmentation similar to the weka segmentation, but it
has multiple workflows. I think in your case might be worth to try the
object classification workflow: first you train the classifier to recognize
islets from exocrine tissue using pixel classification, in a second pass
the workflow is trained to classify the detected objects based on multiple
selectable features.

Best of luck,
Stoyan

На 6.01.2018 г. 1:56 пр.об. "Adrián Villalba" <[hidden email]> написа:

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

Jacqueline Ross

Re: Weka Trainable Segmentation alternative

In reply to this post by Adrián Villalba

Dear Adrián,

I suggest using an immunohistochemistry approach with fluorescence secondary antibodies so that you can specifically label the immune cells and also because segmentation of fluorescence images is much better.

Also, I'm a bit puzzled by your H&E images as they seem to be all the same shade of blue/purple whereas the eosin should be pink with the nuclei only stained in blue. Using the green channel seems like it coud potentially work for size measurements but that's not what you are looking for.

Kind regards,

Jacqui

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Adrián Villalba
Sent: Sunday, 7 January 2018 11:24 a.m.
To: [hidden email]
Subject: Re: Weka Trainable Segmentation alternative

Dear Ignacio,

This is an example image with the different types of islets and insulitis.
The legend is the following:
A = grade 0, no insulitis; B = grade 1, peri-insular; C = grade 2, mild insulitis (<25% of the islet infiltrated); D = grade 3, moderate insulitis (25–75% of the islet infiltrated); E = grade 4, severe insulitis (>75% islet infiltration).

I would like to compare my template images against a new image automatically, so i do not have to do it manually in the microscope. But Weka has not been useful, first of all because Hematoxilin/Eosin staining is not very different between islets and the rest of tissue. Moreover, the different score of the 5 types of islets (scored from 0 to 4) is also tricky for the Weka algorithm. Do you think i can use a different approach?

Thank you very much for your time and attention,

<https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
Libre
de virus. www.avast.com
<https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
<#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

2018-01-06 0:39 GMT+01:00 Adrián Villalba <[hidden email]>:

> Dear all,
>
> I am trying to use the Weka Trainable Segmentation plugin in order to
> classify islets of Langerhans in Hematoxilin/Eosin tissues. There are
> four different types of islets depending on insulitis (immune cell
> infiltration within the islet) as i show you in an attached
> JPG-picture (just to show you the expected result, not to manipulate).
>
> My goal is to do it automatically in imageJ, rather tan manual scoring
> of pictures. So i thought it would be a good idea to use the Trainable
> Weka Segmentation plugin just to train the algorithm to do it
> automatically but it fails. (I cannot attach the classifier.model
> archive because it is rejected by the mailing list conditions).
>
> I think that maybe it is not a proble for the Weka, instead being a
> conceptual problem and that maybe you could know another imageJ tool
> to pursuit that goal.
>
> Thank you very much in advance!
>
> --
>
> - Adrián Villalba Felipe.
> https://es.linkedin.com/in/adrianvillalba
>
>

--

- Adrián Villalba Felipe.
https://es.linkedin.com/in/adrianvillalba

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

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