FIELD: medicine; computer engineering.;SUBSTANCE: invention relates to computer engineering and medicine. Image containing blood cells is obtained; method includes detecting, on the obtained image, blood cells; distinguished are normal and peripheral blood cells; isolating normal blood cells and cutting them from the image, and blood boundary cells are excluded from further analysis; followed by classification of blood cells by types, wherein: obtaining for each image of cut blood cells using augmentation method, which consists in obtaining from a single image a set of images, by turns, images and cuts of part of image; method includes analysing a set of images obtained for each cell and classifying each blood cell by type according to the given set.;EFFECT: technical result consists in automatic detection and classification of blood cell types using deep convolutional neural networks.;5 cl, 1 tbl, 7 dwg
展开▼