nucleus counting with mixed size and intensity

[Jason Bailey-2]

Hello everyone,

I have had much success in counting Hoechst-stained nuclei in images of
various types of cultured cells in the past using the ITCN plugin, but I am
having much difficulty with a current set of images. The difficulty seems to
be that there are two populations present, one set is small and brightly
stained, and the other is large and more faint. Here is a link to a sample
image:

https://www.slashtmp.iu.edu/public/download.php?FILE=jasabail/42663HrCy7o

I have been unable to get ITCN to reliably count the large nuclei with any
settings I've tried, or with masking the small bright nuclei. I've tried a
few other things, such as watershed (both the built-in process and a plugin
written by Daniel Sage I found searching the archives), but these have not
worked.

Can anyone suggest a better strategy?

Thanks

Jason

[BenTupper]

Hi,

On Mar 26, 2010, at 5:17 PM, Jason Bailey wrote:

> Hello everyone,
>
> I have had much success in counting Hoechst-stained nuclei in images  
> of
> various types of cultured cells in the past using the ITCN plugin,  
> but I am
> having much difficulty with a current set of images. The difficulty  
> seems to
> be that there are two populations present, one set is small and  
> brightly
> stained, and the other is large and more faint. Here is a link to a  
> sample
> image:
>
> https://www.slashtmp.iu.edu/public/download.php?FILE=jasabail/42663HrCy7o
>
> I have been unable to get ITCN to reliably count the large nuclei  
> with any
> settings I've tried, or with masking the small bright nuclei. I've  
> tried a
> few other things, such as watershed (both the built-in process and a  
> plugin
> written by Daniel Sage I found searching the archives), but these  
> have not
> worked.
>

If you only want the brighter population you might try using the RATS  
segmentation plugin followed by a watershed. We have used it quite a  
bit with similar images.  Unfortunately RATS doesn't handle multiple  
thresholds so if you want the fainter objects I am not sure what to  
suggest.

You can find it here... http://rsb.info.nih.gov/ij/plugins/rats/index.html

Cheers,
Ben

[Gabriel Landini]

On Friday 26 March 2010, you wrote:

> I have had much success in counting Hoechst-stained nuclei in images of
> various types of cultured cells in the past using the ITCN plugin, but I am
> having much difficulty with a current set of images. The difficulty seems
> to be that there are two populations present, one set is small and
> brightly stained, and the other is large and more faint. Here is a link to
> a sample image:
>
> https://www.slashtmp.iu.edu/public/download.php?FILE=jasabail/42663HrCy7o

Have a look at the CLAHE plugin
http://rsb.info.nih.gov/ij/plugins/clahe/index.html
That brings up the dark nuclei and might make it easier to segment them.

G.

[Jason Bailey-2]

Thanks, all!

The RATS plugin did very well at pulling up the bright nuclei, but it cut out
the big, darker ones that I was having trouble counting. It will actually work
very nicely for another difficult set of images I need to quantify.

Using the CLAHE plugin to pre-process images before running ITCN improved the
counts quite a bit. I still need to play with the settings a little, but
that's a big step in the right direction. And as a bonus, it is fun to watch.

Thanks again!

Jason