Closing cells and Water-shedding

[Clay Morton]

 Hey guys,?


I am currently working on a python pipeline to identify?sickle cells in photos of blood samples. Im trying to use Fiji for the preprocessing steps. My idea is to first convert the images to binary and then to fill any holes left in the cells. I then use watershedding to correctly separate cells that are in clusters. As you can see from the attached image Fiji does a pretty good job. The only issues?arise when healthy blood cells take on a "C"-shape due to curvature of their membrane. This prevents such cells from being filled in and then leads to incorrect watershedding. Do y'all have any ideas as to how I can accurately close such cells and fill them, which out?interrupting?anything else?


Thanks,
Clay



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

[ctrueden]

Hi Clay,

> Do y'all have any ideas as to how I can accurately close such
> cells and fill them, which out interrupting anything else?

Without seeing your original data, I can only suggest checking out:
   http://imagej.net/Segmentation
and in particular
   http://imagej.net/Trainable_Weka_Segmentation

Regards,
Curtis

On Fri, Jan 16, 2015 at 12:24 PM, Clay Morton <
[hidden email]> wrote:

>  Hey guys,?
>
>
> I am currently working on a python pipeline to identify?sickle cells in
> photos of blood samples. Im trying to use Fiji for the preprocessing steps.
> My idea is to first convert the images to binary and then to fill any holes
> left in the cells. I then use watershedding to correctly separate cells
> that are in clusters. As you can see from the attached image Fiji does a
> pretty good job. The only issues?arise when healthy blood cells take on a
> "C"-shape due to curvature of their membrane. This prevents such cells from
> being filled in and then leads to incorrect watershedding. Do y'all have
> any ideas as to how I can accurately close such cells and fill them, which
> out?interrupting?anything else?
>
>
> Thanks,
> Clay
>
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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