INTELLIGENT DATA ANALYSIS IN BIOMEDICAL RESEARCH: CONVOLUTIONAL ARTIFICIAL NEURAL NETWORKS

Convolutional neural networks are one of the modern tools that allow medical research to analyze and recognize not just a set of data about the objects under study or patients, but to use an image as the object under study. Due to the significant role of visual diagnostic methods in the provision of medical care, the use of intelligent recognition of the results of these methods becomes essential. At the moment, convolutional neural networks become widespread in research on the quality of diagnostics in various fields of medicine. However, complex mathematical apparatus behind convolutional neural networks function, and the tools for their construction limit implementation of these models into medical research and practice. This paper provides a gentle introduction to the methodology and application possibilities of convolutional neural networks in medical research. In this paper the reader will find methodological foundations behind convolutional neural networks, a description of a data set for building such models, an example of construction of a convolutional neural network model for classification of dermatoscopic images using TensorFlow and Keras libraries in Python as well as recommendations on how to present the results of building convolutional neural networks.