APPLICATION OF NEURAL NETWORKS IN MEDICAL DIAGNOSTICS

Abstract

This article discusses the possibilities of application of artificial neural networks to solve problems of increasing the diagnostic outcomes in clinical laboratory examination. High diagnostic sensitivity (96 %) and diagnostic accuracy (89.5 %) of the results were shown on a large amount of cellular material digitized by artificial intelligence microscopy automation system like the Vision Cyto Pap. The high resolution and sharpness of digital slides, the mode of viewing objects (cells) in the gallery, quick access to the results of preclassification, all of these factors together allow to reduce turnearound time in more than 2.5 times reducing disadvantages of the microscopy.Application of artificial neural networks does not substitute a doctor’s skills. The role in validation of reports eligible only for cytopathologist. This concept indicates a carefully approach for staff working with a microscope, respectful attitude to them professional skills, and highlights a personalized approach to patients.