Article 7423

Title of the article

DIABETIC RETINOPATHY DIAGNOSIS USING A CONVOLUTIONAL NEURAL NETWORK 

Authors

Daria S. Shishikina, Master degree student of the sub-department of computer technologies, Penza State University (40 Krasnaya street, Penza, Russia), shishikina.2560@mail.ru
Vladimir I. Gorbachenko, Doctor of technical sciences, professor, head of the sub-department of computer technologies, Penza State University (40 Krasnaya street, Penza, Russia), gorvi@mail.ru
Mikhail A. Shcherbakov, Doctor of technical sciences, professor, head of the sub-department of automation and telemechanics, Penza State University (40 Krasnaya street, Penza, Russia), mashcherbakov@yandex.ru

Abstract

Background. The paper deals with the problem of diagnosing diabetic retinopathy. The aim of the work is to create software to facilitate the diagnosis of diabetic retinopathy. Materials and methods. A publicly available set of ocular fundus images from Kaggle website was used as initial data for diabetic retinopathy diagnosis. Inception v3 convolutional neural network was used to work with these images. Quality metrics (precision, completeness and F1-measure) were calculated to assess the quality of the network. Results. Training of a convolutional neural network on ocular fundus images with signs of diabetic retinopathy for image-based diagnosis of the disease and its degree of development was carried out. The network was tested and its quality was evaluated. Conclusions. The network classifies images according to the generally accepted classification of diabetic retinopathy depending on the degree of its development. The network has the ability to classify and can be finalized experimentally for the possibility of its further implementation in professional medical activity.

Key words

diabetic retinopathy, diagnosis of diabetic retinopathy, image recognition, neural networks, convolutional neural networks, classification task

 Download PDF
For citation

Shishikina D.S., Gorbachenko V.I., Shcherbakov M.A. Diabetic retinopathy diagnosis using a convolutional neural network. Modeli, sistemy, seti v ekonomike, tekhnike, prirode i obshchestve = Models, systems, networks in economics, technology, nature and society. 2023;(4):114–132. (In Russ.). doi: 10.21685/2227-8486-2023-4-7

 

Дата создания: 14.12.2023 10:31
Дата обновления: 14.12.2023 11:21