METHOD FOR PERFORMING AUTOMATIC ADAPTIVE CELL SEGMENTATION AND RECOGNITION ON CYTOLOGICAL PREPARATION IMAGES

FIELD: medicine. SUBSTANCE: method involves processing microscopic images in several stages. First of all, cellular structures (nucleoli, nuclei and cytoplasm) outline model parameters adjustment is carried out using small education cell images sample based on cellular structures outline data shown by an operator. Parameters adjustment is carried out in automatic mode by building education sample of neighboring facet (single-connected image fragments uniform in color and brightness) boundaries declared as mergeable or non-mergeable ones. Digitized frame containing cell image is divided into facets during the segmentation process and cellular structure boundaries are built. A segment of common boundary separating two facets belonging the perimeter of one of them, mean curvatures of facet boundaries and their optic densities, predefined ranges of permissible color index values and areas of each structure type are thought to be the factors allowing facets merging. EFFECT: small computation complexity. 2 cl, 1 dwg