Samorodov

Classification of digital microscopy systems and elements of the automated microscopy complexes with an external automation set are considered. A new method and variants of corresponding device for automation of microscope stage, suggesting, in contrast to existing analogs, the use of own microscope stage without its disassembling is proposed. The stated solution can significantly reduce the cost of microscope automation set and make it available for wide practical application.

pp. 91-104

pp. 78-89

The paper presents results of developing an automated complex for determination of the hormonal status of breast cancer using an immunocytochemistry method. The complex performs automated scanning of medication and image capturing by means of an automated computer-driven microscope, as well as automated image analysis which results in estimation of Allred’s scores; the developed complex is capable of visual verification and correction of analysis results if it is necessary. Experimental studies demonstrated high repeatable accuracy of automated analysis results and agreement of automatically drawn conclusions on the hormonal status of breast cancer with the ones obtained by cytologist during visual microscopic examination.

This article presents results of development of an automatic segmentation algorithm for fluorescent microscopic images of cell culture preparations in microbiology tasks. The segmentation algorithm is based on the Niblack local adaptive threshold  inarization of cells images in the V channel of the HSV color space, and images of inclusions of intracellular parasites formed with the use of the modified method of color deconvolution. Practical usage of the segmentation algorithm became possible due to the development of the fast algorithm for the Niblack local adaptive binarization. The fast algorithm is based on construction of two integral representations of the original image: a classical integral image and an integral squared image used for quick calculation of local standard deviation of intensity values. The proposed method provides reduction of binarization running time by 5 orders on average. Introduction of the algorithm into microbiological laboratory practice would eliminate the negative impact of fluorescence on doctor’s eyes caused by visual microscopic analysis of cell culture preparations, significantly reduce complexity and time of the analysis of one preparation and increase overall accuracy of the cultural method.

This paper presents results of developing a segmentation method and algorithm for microscopic images of breast immunocytochemical smears; this method is the key stage in creating a system of smears’ automated analysis. The developed segmentation method was adapted to the currently used method of preparing immunocytochemical smears and provided an opportunity to work with microscopic images of smears with different color-luminance characteristics. The obtained results will allow to objectify quantitative assessment of immunocytochemical reactions and contribute to widespread usage of this type of analysis in clinical practice.

Problems of analysis of renal epithelium cytological specimens’ images, stained by silver nitrate (AgNOR-stained) are considered. Attribute space to describe the nuclei of cells by their images is formed. Nuclei classes, providing the differentiation of norm, reactive, benign and malignant modifications are defined. The nuclei geometric and textural characteristics on AgNOR-stained cytological specimens were used for morphofunctional state analysis of the cells and for the diagnostics.