2518-1092Research result. Information technologies2518-109210.18413/2518-1092-2025-10-1-0-63748ARTIFICIAL INTELLIGENCE AND DECISION MAKING<strong>CORONARY ARTERY STENOSIS DETECTION BASED ON DEEP LEARNING MODELS</strong><strong>CORONARY ARTERY STENOSIS DETECTION BASED ON DEEP LEARNING MODELS</strong>ShchetininEugene YurievichShchetininEugene Yurievichriviera-molto@mail.ruTiutiunnikAnastasiia AlexandrovnaTiutiunnikAnastasiia Alexandrovnatyutyunnik_aa@pfur.ru202510100

Nowadays, cardiovascular diseases are the most common threat to human health, and coronary artery disease is a particularly serious disease. Coronary angiography is used to detect coronary artery disease. However, the high cost and complexity of analyzing its results have led to the need to automate the process of diagnosis of coronary artery stenosis. In this work, we considered popular models of deep learning-based stenosis detection. The models varied in their underlying neural network architecture and were pre-trained on publicly available data. The data consist of video sequences clinically obtained by invasive coronary angiography and labeled into separate frames for each video containing coronary artery stenosis with a resolution of (512x512) pixels. The paper presents a comparative analysis of the models based on the main performance indicators: average accuracy (mAP), image processing time, and the number of model parameters. The Faster R-CNN and EfficientDet D4 models showed the best performance. Compared to other models, they are characterized by relatively low parameter weights, high detection accuracy, and high image processing speed. The comparative analysis showed that the results of this study are superior to or comparable to those of other researchers.

Nowadays, cardiovascular diseases are the most common threat to human health, and coronary artery disease is a particularly serious disease. Coronary angiography is used to detect coronary artery disease. However, the high cost and complexity of analyzing its results have led to the need to automate the process of diagnosis of coronary artery stenosis. In this work, we considered popular models of deep learning-based stenosis detection. The models varied in their underlying neural network architecture and were pre-trained on publicly available data. The data consist of video sequences clinically obtained by invasive coronary angiography and labeled into separate frames for each video containing coronary artery stenosis with a resolution of (512x512) pixels. The paper presents a comparative analysis of the models based on the main performance indicators: average accuracy (mAP), image processing time, and the number of model parameters. The Faster R-CNN and EfficientDet D4 models showed the best performance. Compared to other models, they are characterized by relatively low parameter weights, high detection accuracy, and high image processing speed. The comparative analysis showed that the results of this study are superior to or comparable to those of other researchers.

deep learningcoronary artery stenosisneural networkX-ray coronary angiographydeep learningcoronary artery stenosisneural networkX-ray coronary angiographyСписок литературыMinistry of Health of the Russian Federation. URL: https://minzdrav.gov.ru/Falk, E., Pathogenesis of atherosclerosis. Journal of the American College of cardiology. 2006. 47(8S), p. C7-C12.Collet, C. et al. Coronary computed tomography angiography for heart team decision-making in multivessel coronary artery disease. Eur. Heart J.2008. 39(41), p. 3689&ndash;3698.Wan, T. et al. Automated identifcation and grading of coronary artery stenoses with X-ray angiography. Comput. Methods Progr. Biomed. 2018. 167, p. 13&ndash;22.Shchetinin E. Yu. Sovremennyye problemy tsifrovoy obrabotki i analiza bol&#39;shikh dannykh v meditsine i biologii. 2023. M.: Izdatel&#39;skiy Dom &laquo;Nauchnaya biblioteka&raquo;. 168 p.; [Modern problems of digital processing and analysis of big data in medicine and biology. 2023. M.: Publishing House &quot; Nauchnaya biblioteka &quot;. 168 p.] ISBN 978-5-907672-27-7.Shchetinin E. Yu. Heart coronary arteries detection by deep learning methods. Journal of applied informatics. 2024. Vol. 19. No. 4., p. 94&ndash;106. DOI: 10.37791/2687-06492024-19-4-94-106Ovalle-Magallanes, E., et al., Deep Learning-based Coronary Stenosis Detection in X-ray Angiography Images: Overview and Future Trends. Artificial Intelligence and Machine Learning for Healthcare. 2022. Vol. 2: Emerging Methodologies and Trends, p. 197-223.Pang, K., et al., Stenosis-DetNet: Sequence consistency-based stenosis detection for X-ray coronary angiography. Computerized Medical Imaging and Graphics. 2021. 89, p.101900.Danilov, V.V., et al., Real-time coronary artery stenosis detection based on modern neural networks. Scientific reports. 2021. 11(1), p. 7582.Moon, J.H., et al., Automatic stenosis recognition from coronary angiography using convolutional neural networks. Computer methods and programs in biomedicine. 2021. 198, p. 105819.Liu W., Anguelov D., Erhan D., Szegedy C., Reed S., Fu C.-Y., Berg A.C. SSD: Single Shot MultiBox Detector. 2016. URL: https://arxiv.org/pdf/1512.02325.pdfJ. Dai, Asia Yi Li, K. He, J. Sun. R-FCN: Object Detection via Region-based Fully Convolutional Networks. 2023. URL: https://arxiv.org/abs/1605.06409Ren S. et al., Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. 2016. URL: https://arxiv.org/abs/1506.01497Lin, T.-Y., et al. Feature pyramid networks for object detection. IEEE Conference on Computer Vision and Pattern Recognition. 2017. pp. 936-944, doi: 10.1109/CVPR.2017.106.Tan M., R. Pang, and Q.V. Le. Efficientdet: Scalable and efficient object detection. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2020. pp. 10778-10787, doi: 10.1109/CVPR42600.2020.01079.Wang C., Zhong C. Adaptive Feature Pyramid Networks for Object Detection.&nbsp;IEEE Access.&nbsp;2021. V.&nbsp;9, pp. 107024&ndash;107032. doi: 10.1109/ACCESS.2021.3100369Lin T.-Y., et al. Focal loss for dense object detection. IEEE International Conference on Computer Vision (ICCV). 2017. pp. 2999-3007, doi: 10.1109/ICCV.2017.324.National Heart, Lung, and Blood Institute: Ischemic Heart Disease (2013). URL: https://www.nhlbi.nih.gov/health-topics/ischemic-heart-disease.Microsoft COCO: Common Objects in Context. URL: http://cocodataset.orgKerasHub Modeling API. URL: https://keras.io/api/keras_cv/models/tasks/retinanet/.TensorFlow 2 Detection Model Zoo. URL:&nbsp;Rodrigues Dinis L., Menezes Miguel Nobre, Fausto J. Pinto, and Arlindo L. Oliveira. Automated Detection of Coronary Artery Stenosis in X-ray Angiography using Deep Neural Networks. 2021. URL: https://arxiv.org/abs/2103.02969Wu W, Zhang J, Xie H, et al. Automatic detection of coronary artery stenosis by convolutional neural network with temporal constraint. Comput. Biol. Med. 2020. 118, p. 103657. doi:10.1016/j. compbiomed.2020.103657.Xin Jin,&nbsp;Yuze Li&nbsp;,&nbsp;Fei Yan&nbsp;,&nbsp;Ye Liu&nbsp;,&nbsp;Xinghua Zhang&nbsp;,&nbsp;Tao Li&nbsp;,&nbsp;Li Yang&nbsp;,&nbsp;Huijun Chen.&nbsp;Automatic coronary plaque detection, classification, and stenosis grading using deep learning and radiomics on computed tomography angiography images: a multi-center multi-vendor study. Eur. Radiol. 2022. 32(8), p. 5276-5286.