3 THE SEGMENTATION METHOD

4 CONCLUSIONS

REFERENCES

Acknowledgments

Group of Digital Signal Processing (GUPDIS)
(+) Department of Pathology , Hospital "Dr. Carlos J. Finlay"
Institute of Cibernetics, Mathematics and Physics (ICIMAF)
Academy of Sciences of Cuba
Calle 15 No. 551 e/ C y D Código Postal 10400, La Habana, CUBA

Fax: (537) 33-3373, Tel.: 32-5026, 32-7764, 32-2688

E mail: rrm@cidet.icmf.inf.cu

The obtained results, in color image, using differents segmentation methods with the objective to isolate and to identify blood vessels in histological cut of malign tumors of soft parts are presented.

For comparison purpose was realized experiments in divers color spaces, and it was concluded, that for this type of application and according to the original image, the RGB space is the one which offers better results. It was realized a study of the image for the red, green and blue channel, and is proposed a strategy of segmentation where the experimental results demonstrate that the same is rapid and robust, all this in the environment of a personal computer. In the work was evidenced that red channel is the more noisy and that the green channel is the one which offers greater quantity of information.