Thread Subject: counting white blood cell image-filtering doubt

hi!i m working with matlab software to count white blood cell in color image...i had already performed segmentation(k-means clustering) n image enhancement..results seems ok but the problem is the image did really left only white blood cells with some particles at the background after segmentation...can d background particles be filtered out???what type of filtering should be used to provide a better result n at d same time produce back color image after the filtering stage???please help?
thanks a lot.

Poh Choo:
Employ better filtering than you already have. Seriously, how can we
comment on pictures that you haven't shown us yet?
Post your original pictures and segmentation results to http://drop.io
and then we can help you better.
-ImageAnalyst

"Poh Choo " <swpohchoo@yahoo.com> wrote in message <hk8ia2$rjm$1@fred.mathworks.com>...
> hi!i m working with matlab software to count white blood cell in color image...i had already performed segmentation(k-means clustering) n image enhancement..results seems ok but the problem is the image did really left only white blood cells with some particles at the background after segmentation...can d background particles be filtered out???what type of filtering should be used to provide a better result n at d same time produce back color image after the filtering stage???please help?
> thanks a lot.

hi is there any body to give repky for this message,.. pls help us.. v much inneed of ur reply

It is genuinely impossible to make specific suggestions without more information about the problem, and a sight of an example image. However, you could make a good start yourself (yourselves?) by looking through the demos for the Image Processing Toolbox to get ideas. You might find that one of these answers your question.

On Mar 20, 10:18 am, "Mohamed Hani" <catchmd_h...@hotmail.com> wrote:
> "Poh Choo " <swpohc...@yahoo.com> wrote in message <hk8ia2$rj...@fred.mathworks.com>...
>
> > hi!i m working with matlab software to count white blood cell in color image...i had already performed segmentation(k-means clustering)  n image enhancement..results seems ok but the problem is the image did really left only white blood cells with some particles at the background after segmentation...can d background particles be filtered out???what type of filtering should be used to provide a better result n at d same time produce back color image after the filtering stage???please help?
> > thanks a lot.
>
> hi is there any body to give repky for this message,.. pls help us.. v much inneed of ur reply

------------------------------------------------------------------------------------------------
Why don't you start your own thread? You saw how Poh Choo got nowhere
with a post that was useless, and then you ask about his/her post
again even though you saw that he/she didn't post any additional
information. How about this: You ignore Poh Choo and all of the
mistakes he/she made and post your own message with enough information
for people to give you meaningful help. Will that work for you?

"David Young" <d.s.young.notthisbit@sussex.ac.uk> wrote in message <ho2ogf$8rb$1@fred.mathworks.com>...
> It is genuinely impossible to make specific suggestions without more information about the problem, and a sight of an example image. However, you could make a good start yourself (yourselves?) by looking through the demos for the Image Processing Toolbox to get ideas. You might find that one of these answers your question.

Sir I am able t separate the nucleus part from the blood microscope images but we require also the cytoplasm along with the blue nucleus which the kmeans is unable to separate. The cytoplasm remains along with the red blood cells and we are unable to get it,

Subrajeet

"subrajeet Mohapatra" <subrajeets@gmail.com> wrote in message <hqe7f0$lsu$1@fred.mathworks.com>...
> "David Young" <d.s.young.notthisbit@sussex.ac.uk> wrote in message <ho2ogf$8rb$1@fred.mathworks.com>...
> > It is genuinely impossible to make specific suggestions without more information about the problem, and a sight of an example image. However, you could make a good start yourself (yourselves?) by looking through the demos for the Image Processing Toolbox to get ideas. You might find that one of these answers your question.
>
> Sir I am able t separate the nucleus part from the blood microscope images but we require also the cytoplasm along with the blue nucleus which the kmeans is unable to separate. The cytoplasm remains along with the red blood cells and we are unable to get it,
>
> Subrajeet


Are you able to share an image so we might be able to better understand what you are trying to do, its tough to get ideas on how to better process an image without actually seeing an image. If your research prevents you from disclosing the data then I dont know how we can help.

"Poh Choo " <swpohchoo@yahoo.com> wrote in message <hk8ia2$rjm$1@fred.mathworks.com>...
> hi!i m working with matlab software to count white blood cell in color image...i had already performed segmentation(k-means clustering) n image enhancement..results seems ok but the problem is the image did really left only white blood cells with some particles at the background after segmentation...can d background particles be filtered out???what type of filtering should be used to provide a better result n at d same time produce back color image after the filtering stage???please help?
> thanks a lot.

David Young,

 We dont want to hide any thing. We want to get the WBC separated from the blood microscope images. So that we can clearly do meidical analysis on the WBC for disorders.

> We dont want to hide any thing. We want to get the WBC separated from the blood microscope images. So that we can clearly do meidical analysis on the WBC for disorders.

I didn't say you wanted to hide anything! But I still can't see an image!

"Poh Choo " <swpohchoo@yahoo.com> wrote in message <hk8ia2$rjm$1@fred.mathworks.com>...
> hi!i m working with matlab software to count white blood cell in color image...i had already performed segmentation(k-means clustering) n image enhancement..results seems ok but the problem is the image did really left only white blood cells with some particles at the background after segmentation...can d background particles be filtered out???what type of filtering should be used to provide a better result n at d same time produce back color image after the filtering stage???please help?
> thanks a lot.

Ok Sorry miss interpretation
Plz find the links
http://www.healthsystem.virginia.edu/internet/hematology/images/AML-M7-immature-megakaryocytes-bilobed-website.jpg

http://www.healthsystem.virginia.edu/internet/hematology/images/Acute-lymphoblastic-leukemia-L1-subtype-100x-website.jpg

So we need to extract WBC only and remove RBc and background

OK, thanks for posting the link to the images. It's interesting. I've just realised, though, that this somewhat duplicates the discussion on the other thread, at

http://www.mathworks.com/matlabcentral/newsreader/view_thread/278122

where ImageAnalyst's reply looks good to me, so I'll leave this one.

can you plz tell me the way you've applied k-means clustering to the image of wbc.
i am also working on a similar project based on detection of tuberoclosis bacili.
plz reply....

priyanka