ANALYSIS OF BIOLOGICAL OBJECTS BY DIGITAL OPTICAL MICROSCOPY USING NEURAL NETWORKS

An interdisciplinary (physics, biology, medicine, computer science) problem of recognizing bone marrow cells for the diagnosis of acute leukemia and minimal residual disease is considered. The use of neural networks for determining cell classes in a microscopic image of a bone marrow preparation is considered. Images are formed in transmitted light. The structure of the neural network and image sample for experimental studies are presented. The structure of the neural network is refined based on the experimental results. The accuracy characteristics of the developed system are determined.