DIAGNOSIS OF MALIGNAT NEIPLASMS OF THE CHEST USING NEURAL NETWORK

Abstract

The negative dynamics of the incidence of cancer gives high importance and relevance to the task of improving the effectiveness of diagnostic methods. Worldwide, more than 10 million cases of pathology are detected annually including 2.2 million cases of lung cancer of which 1.8 million cases are fatally. Early differential and accurate diagnosis of the disease is traditionally considered an important task of medicine. The aim of the work is to create an automated system for processing the results of objective control for the differential diagnosis of malignant neoplasms in the chests and to increase the accuracy and speed of diagnosis with its help. The resulting product is an artificial intelligence system based on a neural network that analyzes images and their multiple classification. Image analysis allows not only record the absence of presence of malignant neoplasms but also in the latter case to make a differential diagnosis of adenocarcinoma, large cell carcinoma and squamous cell lung cancer. The results of this product significantly exceed the achievements of other systems described and currently available the resulting product has an error of 3.5 % while the error of existing analogues is at least 7.1 % which is twice the error of the resulting system. The proposed product makes it possible to reduce the number of incorrectly diagnosed diagnoses by 2 times compared to currently existing analogues which is a significant achievement.