- M. Ušaj, D. Torkar, M. Kandušer, D. Miklavčič
- Journal of Microscopy, 2011, Vol. 2011, No. 3
pages: 303-314
- abstract:
In this paper a novel parameter optimization approach for cell
detection tool and counting cells procedure in phase contrast
images are presented. Manual counting of the attached cells
in phase contrast images is time-consuming and subjective.
For evaluation of electroporation efficiency of attached cells,
we often perform manual counting of the cells which is
needed to determine the percentage of electroporated cells
under different experimental conditions. Here we present an
automated cell counting procedure based on novel artificial
neural network optimizationof Image-based Tool forCounting
Nuclei algorithm parameters to fit the training image set based
on counts from an expert. Comparing the results of automated
cell counting to user manual counting a 90,31% average
agreement was achieved which is reasonably good especially
taking intoaccountinter-person errorwhichcanbeupto10%.
Evenmore, our procedure can also be used for fluorescent cell
imageswith similar counting accuracy (>90%) enabling us to
determine electroporation efficiency. In our experiments, the
electroporation efficiency determined by manual cell counting
was virtually the same as the one obtained by the automated
procedure.
|
