On parameterization of cell colonies images: ART classifier approach

Abstract

Cell colonies counting is an important subset of cytometry research. That is why the article investigates possibility to use binary ART classifier for the flexible automatic parameterization of biological sample images. In order to evaluate influence of electroporation semi automatic cell colonies detection system is created. For the investigation developed cell colonies image parameterization technique is used. Developed technique deals with specifics in biological sample image registration, Petri dish area determination, cell colonies detection and parameterization, and incorporation of ART classifier. Experimental investigation is carried out in order to check reliability of cell colony count as also as efficiency of cell colony parameterization. Results of experimental verification confirm that the use of ART1 classifier increases trustworthiness of the parameterization of images with cell colonies in terms of sensitivity and specificity. ART1 classifier outperforms SVM and heuristic classifiers in terms of Fl, accuracy and precision.