Sickle Cell Anemia Detection

[Benjamin Grant]

Hi all,
I am trying to implement sickle cell anemia detection in imageJ. I don't
need programming advice currently, but just method ideas. Sickle cells are
jagged shaped compared to smooth, round normal cells. Any method available
to detect these sharp edges? Any thoughts? Thanks a lot
Ben

[Daniel Hornung]

Benjamin Grant wrote:
> Hi all,
> I am trying to implement sickle cell anemia detection in imageJ. I don't
> need programming advice currently, but just method ideas. Sickle cells are
> jagged shaped compared to smooth, round normal cells. Any method available
> to detect these sharp edges? Any thoughts? Thanks a lot
> Ben

Hello Ben,

you could try to automatically segment the cells, fit an ellipse to the mask
(http://rsb.info.nih.gov/ij/developer/api/ij/process/EllipseFitter.html) and
compare the length (major axis) or circumference to the results you got from
your segmentation.  You could also try to fit curves (e.g. Bezier curves or a
spline) to the segmentation boundary and search for high local curvatures there.

These are just some ideas, I'm sure you'll find more yourself.
HTH, Daniel

--
Daniel Hornung

Biomedical Physics Group
Max-Planck-Institute for Dynamics and Self-Organization
Bunsenstr. 10
D-37073 Goettingen

([phone]

[Joel Sheffield]

If sickle cells have a different axial ratio from normals, you could
use the analyze particles routine and select for particles of a given
value.

Joel


> Hi all,
> I am trying to implement sickle cell anemia detection in imageJ. I
> don't
> need programming advice currently, but just method ideas. Sickle
> cells are
> jagged shaped compared to smooth, round normal cells. Any method
> available
> to detect these sharp edges? Any thoughts? Thanks a lot
> Ben


--
Joel B. Sheffield, Ph.D.
Biology Department, Temple University
1900 North 12th Street
Philadelphia, PA 19122
[hidden email]  
[phone], fax [phone]
http://astro.temple.edu/~jbs

[David Webster]

Ben,

The ImageJ Plugins page has a reference to something called "Fractal
Dimension and Lacunarity" which seems oriented toward cell analysis. You
might also want to look at contour Fourier Descriptors by Thomas Boudier
on the same page.

Hope this helps. If I think of anymore, I will email you.

David Webster

[David Webster]

Ben,

A couple of more ideas. Check the plugins page for "Iterative Radial Voting*
" .* You might want to look into using local point location covariance
eigenvalues. Eigenvalue ratios can differentiate between long skinny and
bloby point distributions. This is somewhat akin to Medioni's Tensors for
Tensor voting.

I was wondering, do you have any non-proprietary cell imagery that I might
have access to. I am working (slowly) on using the Weka machine learning
utility with ImageJ, and could use some good data sets (hopefully lableled)
for testing ideas.

David Webster

[hidden email]




On Thu, Oct 22, 2009 at 6:34 AM, Benjamin Grant <[hidden email]> wrote:

> Hi all,
> I am trying to implement sickle cell anemia detection in imageJ. I don't
> need programming advice currently, but just method ideas. Sickle cells are
> jagged shaped compared to smooth, round normal cells. Any method available
> to detect these sharp edges? Any thoughts? Thanks a lot
> Ben
>