Abstract
this study discusses a system for analyzing noninvasive examination results to support the decision-making by a cardiovascular surgeon/phlebologist. The software helps the phlebologist in making decisions to determine the CEAP classification code in controversial and complicated cases. The system recognizes uploaded DICOM format images with a convolutional neural network.
Contrast enhancement of b/w DICOM images was applied for the neural network training. It improves the image handling and increases the recognition accuracy. The average recognition rate is from 86.1 to 97.4 %.
References
Таршхоева Ж. Т. Язык программирования Python. Библиотеки Python. Молодой ученый. 2021;5:20–21. Режим доступа: https://elibrary.ru/item.asp?id=44667958.
Ле Мань Ха. Сверточная нейронная сеть для решения задачи классификации. Труды МФТИ.2016;8(3):91–97.
Mayank Mishra. Сверточная нейронная сеть (CNN). Режим доступа: https://www.helenkapatsa.ru/sviortochnaia-nieironnaia-siet/.
TensorFlow. Режим доступа: https://www.tensorflow.org/.