Spematozoid segmentation

[Juanma Pascual Gaspar]

Hello, I am developing a tool for counting spermatozoids based on the ImageJ
ParticleAnalizer plugin. The thing is the PA plugin works well with a very
defined binary image, but in the binary segmentation process (based on pixel
filtering wihin a range of gray scale levels) I can't diferentiate between
genuine spermatozoid and another cells with the same size and gray scale
intensity.

The difference between these cells and the indesired cells
(denominated round cells) is the absence of tail. Does anybody knows how I
can do this with (preferably with a ImageJ plugin or similar).

Many thanks in advance.
 Juan

[jiho]

On 2 févr. 06, at 15:56, Juanma Pascual Gaspar wrote:

> Hello, I am developing a tool for counting spermatozoids based on  
> the ImageJ
> ParticleAnalizer plugin. The thing is the PA plugin works well with  
> a very
> defined binary image, but in the binary segmentation process (based  
> on pixel
> filtering wihin a range of gray scale levels) I can't diferentiate  
> between
> genuine spermatozoid and another cells with the same size and gray  
> scale
> intensity.
>
> The difference between these cells and the indesired cells
> (denominated round cells) is the absence of tail. Does anybody  
> knows how I
> can do this with (preferably with a ImageJ plugin or similar).

if you see the tail in the binary image you could filter your results  
based on the ratio circumference/area of the particles. The round  
ones will probably have a much smaller circumference than the  
spermatozoids, for a the same area. It will require some calibration  
to find the good range of values but the analyze particle function  
can give you both the area and the circumference of particles with no  
problem.

> Many thanks in advance.
>  Juan

JiHO
---
   Windows, c'est un peu comme le beaujolais nouveau :
a chaque nouvelle cuvee on sait que ce sera degueulasse,
     mais on en prend quand meme par masochisme.
---

[Gary Chinga]

You may also try the shape descriptor plugin which remove particles  
based on specific shape values. Try with a Form Factors lower than  
0.95 or a combination of the available descriptors.

http://www.gcsca.net/IJ/Shapes.html

Gary.


On Feb 2, 2006, at 3:56 PM, Juanma Pascual Gaspar wrote:

> Hello, I am developing a tool for counting spermatozoids based on  
> the ImageJ
> ParticleAnalizer plugin. The thing is the PA plugin works well with  
> a very
> defined binary image, but in the binary segmentation process (based  
> on pixel
> filtering wihin a range of gray scale levels) I can't diferentiate  
> between
> genuine spermatozoid and another cells with the same size and gray  
> scale
> intensity.
>
> The difference between these cells and the indesired cells
> (denominated round cells) is the absence of tail. Does anybody  
> knows how I
> can do this with (preferably with a ImageJ plugin or similar).
>
> Many thanks in advance.
>  Juan
>
>

[Angel Cruz-3]

Hola, I also have been developing a tool of analysis of sperm, so much of
morphology as mobility. I it have been employed at bovine sperm, and really
to realize a good segmentation it must use two criteria, the roundness or
compacity, the area, and the color (level of grey for example).
Generally the sperms are elliptical and his detection is more or less easy.
Despues it must purify it with a good classifier. To eliminate the tails it
can use the erosion and the dilate.
I have more info. if you need

2006/2/2, jiho <[hidden email]>:

>
> On 2 févr. 06, at 15:56, Juanma Pascual Gaspar wrote:
>
> > Hello, I am developing a tool for counting spermatozoids based on
> > the ImageJ
> > ParticleAnalizer plugin. The thing is the PA plugin works well with
> > a very
> > defined binary image, but in the binary segmentation process (based
> > on pixel
> > filtering wihin a range of gray scale levels) I can't diferentiate
> > between
> > genuine spermatozoid and another cells with the same size and gray
> > scale
> > intensity.
> >
> > The difference between these cells and the indesired cells
> > (denominated round cells) is the absence of tail. Does anybody
> > knows how I
> > can do this with (preferably with a ImageJ plugin or similar).
>
> if you see the tail in the binary image you could filter your results
> based on the ratio circumference/area of the particles. The round
> ones will probably have a much smaller circumference than the
> spermatozoids, for a the same area. It will require some calibration
> to find the good range of values but the analyze particle function
> can give you both the area and the circumference of particles with no
> problem.
>
> > Many thanks in advance.
> >  Juan
>
> JiHO
> ---
>   Windows, c'est un peu comme le beaujolais nouveau :
> a chaque nouvelle cuvee on sait que ce sera degueulasse,
>     mais on en prend quand meme par masochisme.
> ---
>

--
Cordial Saludo,


ANGEL ALFONSO CRUZ ROA
Ing. de Sistemas
Universidad de los Llanos
Villavicencio - Meta,  COLOMBIA
Tel. + 57 8 6624769
Movil: 300 - 586 0237
E-mail:  [hidden email]
E-mail2: [hidden email]