Microsc. Microanal. Microstruct. 7, 331-337 (1996)
DOI: 10.1051/mmm:1996130

Automated Segmentation of Cytological and Histological Images for the Nuclear Quantification: an Adaptive Approach based on Mathematical Morphology

Abderrahim Elmoataz¹, Philippe Belhomme¹, Paulette Herlin², Sophie Schüpp¹, Marinette Revenu¹ et Daniel Bloyet<sup>1

1</sup>  GREYC-UPRESA CNRS 6072, ISMRA, 6 bd Maréchal Juin, 14050 Caen Cedex, France
²  Laboratoire d'Anatomie Pathologique, Centre F. Baclesse, Route de Lion/mer, 14021 Caen Cedex, France

Abstract
A general segmentation strategy allowing to blend multiple criteria such as contour-region or color information in a region growing process derived from the watershed transformation is proposed. It is applied onto different types of cytological and histological microscopic images.

Résumé
Une stratégie générale de segmentation combinant des critères locaux et globaux dans un processus de croissance dérivé de la ligne de partage des eaux est proposée. Les informations prises en compte sont de type contour-région ou couleur. La méthode est appliquée à différents cas d'images microscopiques de cytologie et d'histologie.

PACS
8760G - Microwaves and other electromagnetic waves medical uses.
8725F - Physics of subcellular structures.
8770E - Patient diagnostic methods and instrumentation.
7510B - Radiation and radioactivity applications in biomedicine.
6140C - Optical information, image and video signal processing.
7330 - Biology and medical computing.
5260B - Computer vision and image processing techniques.

Key words
adaptive signal processing -- cellular biophysics -- image segmentation -- medical image processing -- optical microscopy -- histological images -- cytological images -- nuclear quantification -- automated segmentation -- mathematical morphology -- contour region information -- color information -- region growing process -- watershed transformation -- microscopic images -- oncology


© EDP Sciences 1996