P. Belhomme, P. Herlin, A. Elmoataz, O. Rougereau, C. Boudry & D. Bloyet

Automatic segmentation for DNA ploidy measurements: methodological choices

Attached document(s)

Annexes

• AS_18_n3_415_BEL (3.4M)

Abstract

Image cytometry (ICM) makes it possible to measure DNA content of cancer cell nuclei in a fully automatic and reproducible way. DNA ICM relies on a three step procedure: segmentation of all the nuclei collected in a series of images, followed by computation of integrated optical density (IOD) of DNA specific stain and morphometric parameters, then on elimination of unwanted elements, thanks to a selective sorting of parameters. Quality of sorting is closely linked to the accuracy of the computed parameters, these later depending of course on the quality of segmentation. The present paper studies the impact of the segmentation step on the number of nuclei to be measured, on their size and their IOD. The study reveals that the choice of one segmentation method over another may involve a great disparity in measurements, e.g. in the percentage of aneuploid cells and in the size of the proliferating cell compartment.

To cite this article

About: P. Belhomme

About: P. Herlin

About: A. Elmoataz

About: O. Rougereau

About: C. Boudry

About: D. Bloyet