Article 8423
Title of the article |
THE USE OF DEEP MACHINE LEARNING METHODS IN THE TASK OF DETECTING THE DOME OF THE CECUM |
Authors |
Vladimir V. Khryashchev, Candidate of technical sciences, associate professor, associate professor of the sub-department of digital technologies and machine learning, Yaroslavl State University named after P.G. Demidov (14 Sovetskaya street, Yaroslavl, Russia), v.khryashchev@uniyar.ac.ru |
Abstract |
Background. Quality control of colonoscopic examinations of the intestine based on the analysis of relevant videodata is considered. To solve this problem, it is proposed to detect frames on the video stream containing images of the dome of the cecum. These fragments in the video stream are a key point in the process of conducting a colonoscopy examination. Materials and methods. To solve this problem, methods and algorithms of deep machine learning and computer vision are used. An algorithm is proposed for detecting the cecal dome region in video images using a convolutional neural network of the Yolo architecture. Results. Metrics for assessing the quality of work of the proposed algorithm were obtained using a test image database containing 1561 frames of the cecal dome region. The results of the neural network algorithm are analyzed in comparison with the popular approach based on the neural network architecture of SSD. The proposed neural network algorithm shows stable operation and outperforms its analogues in terms of standard metrics. Conclusions. The results will be the basis for building, based on the developed algorithm, a videostream analysis module in a real medical system for conducting colonoscopy studies. |
Key words |
colonoscopic video images, cecum localization, deep learning, neural network object detection |
![]() |
Download PDF |
For citation |
Khryashchev V.V. The use of deep machine learning methods in the task of detecting the dome of the cecum on the video data of colonoscopic examination. Modeli, sistemy, seti v ekonomike, tekhnike, prirode i obshchestve = Models, systems, networks in economics, technology, nature and society. 2023;(4):133–141. (In Russ.). doi: 10.21685/2227-8486-2023-4-8 |
Дата обновления: 14.12.2023 11:24