SYSTEM AND METHOD FOR IMPLEMENTING QUANTITATIVE HISTOPATHOLOGICAL ASSESSMENT OF NON-CLASSIFIED TISSUE IMAGE ON SLIDE, TISSUE CLASSIFIER FOR IMPLEMENTING QUANTITATIVE HISTOPATHOLOGICAL ASSESSMENT AND METHOD FOR ITS TRAINING Russian patent published in 2023 - IPC A61B5/00 G06V10/40 G06V30/18 G06T7/00 G01N33/483 C12M1/34 C12N5/00 

FIELD: medicine.

SUBSTANCE: group of inventions relates to quantitative histomorphological analysis of digital images of cells in a tissue sample to determine viable candidate tissues for transplantation, namely: to a system and method for quantitative histopathological evaluation of an unclassified tissue image on a slide, a tissue classifier for such evaluation and a method for its training. At the same time, the image of the glass slide is converted into a binary image. The slide image is selected from a library of reference images, where each image is associated with a pass or fail classification. Nuclei are detected within a binary image. A feature is extracted from the detected core, which contains the area, perimeter, integral density and roundness of the core. For each detected nucleus, a feature associated with the binary image of the slide is generated, which is a numerical value. A binary image is included in the cluster. An upper limit value is applied to a tree cluster scheme to form a plurality of groups. The upper limit minimizes the number of groups within the set of groups based on the multivariate analysis of variance of the cluster. The first group is categorized as a positive control group if the first group contains the first slide images associated with a satisfactory classification. The second group is categorized as a negative control group if the second group contains second slide images associated with an unsatisfactory classification.

EFFECT: increase in the efficiency of qualitative analyses is achieved with the possibility of converting them into semi-quantitative or quantitative ones.

60 cl, 57 dwg, 7 tbl, 8 ex