bsuirДоклады БГУИРDoklady BGUIR1729-76482708-0382БГУИР10.35596/1729-7648-2025-23-1-74-82bsuir-4066Research ArticleСтатьиАрхитектура многоветвевой сверточной нейронной сети для диагностики глаукомы на основе биомаркеров оптической когерентной томографии и симуляции синтетических изображенийMulti-Branch Convolutional Neural Network Architecture for Glaucoma Diagnosis Using Optical Coherence Tomography Biomarkers and Synthetic Image SimulationУсенкоФ. В.UsenkoPh. V.

Минск

Postgraduate at the Engineering Psychology and Ergonomics Department

Minsk

ПрудникА. М.PrudnikA. M.

Минск

Prudnik Aleksander Mikhailovich, Cand. of Sci., Associate Professor, Associate Professor at the Engineering Psychology and Ergonomics Department

220013, Minsk, P. Brovki St., 6

aleksander.prudnik@bsuir.byБелорусский государственный университет информатики и радиоэлектроникиBelarusian State University of Informatics and Radioelectronics2025170220252317482Copyright © Усенко Ф.В., Прудник А.М., 20252025Усенко Ф.В., Прудник А.М.Usenko P.V., Prudnik A.M.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://doklady.bsuir.by/jour/article/view/4066

В статье представлена многоветвевая сверточная нейронная сеть, разработанная для диагностики глаукомы с использованием биомаркеров оптической когерентной томографии и симуляции синтетических изображений. Сеть включает шесть ветвей, каждая из которых нацелена на ключевые анатомические особенности. Обученная на синтетическом наборе данных, модель показала точность проверки 94,2 % и потери при обучении 0,162, демонстрируя эффективность в различении разных типов глаукомы. Результаты также подчеркивают потенциал модели для дальнейшего повышения точности, особенно в части уменьшения ошибок классификации между близкими состояниями.

This paper presents a multi-branch convolutional neural network designed for glaucoma diagnosis using optical coherence tomography biomarkers and synthetic image simulations. The network includes six branches, each targeting key anatomical features. Trained on a synthetic dataset, the model achieved a validation accuracy of 94.2 % and a training loss of 0.162, demonstrating effectiveness in distinguishing between different glaucoma types. The results also highlight the potential for further accuracy improvement, particularly in reducing classification errors between closely related conditions.

диагностика глаукомыоптическая когерентная томографиясверточная нейронная сетьбиомаркеры оптической когерентной томографиигенерация синтетических данныхсегментация изображенийглубокое обучение в офтальмологииglaucoma diagnosisoptical coherence tomographyconvolutional neural networkoptical coherence tomography biomarkerssynthetic data generationimage segmentationdeep learning in ophthalmologyВыражаем искреннюю благодарность сотрудникам кафедры офтальмологии Института повышения квалификации и переподготовки кадров здравоохранения Белорусского государственного медицинского университета кандидату медицинских наук, доценту Оксане Николаевне Дудич и доктору медицинских наук, профессору Виктории Леонидовне Красильниковой за ценное обсуждение и конструктивные замечания, которые способствовали совершенствованию данной статьи.We express our sincere gratitude to the staff of the Department of Ophthalmology of the Institute for Advanced Training and Retraining of Healthcare Personnel of the Belarusian State Medical University, candidate of medical sciences, Associate Professor Oksana Nikolaevna Dudich and Doctor of Medical Sciences, Professor Victoria Leonidovna Krasilnikova for valuable discussions and constructive comments that contributed to the improvement of this article.ReferencesWorld Health Organization (2024) ICD-11 for Mortality and Morbidity Statistics. https://icd.who.int/browse/2024-01/mms/en#499924848.World Health Organization (2024) ICD-11 for Mortality and Morbidity Statistics. https://icd.who.int/browse/2024-01/mms/en#499924848.Weinreb R. N., Aung T., Medeiros F. A. (2016) The Pathophysiology and Treatment of Glaucoma: A Review. JAMA. DOI: 10.1001/jama.2016.4697.Weinreb R. N., Aung T., Medeiros F. A. (2016) The Pathophysiology and Treatment of Glaucoma: A Review. JAMA. DOI: 10.1001/jama.2016.4697.Caprioli J., Coleman A. L. (2017) Intraocular Pressure Fluctuation: A Risk Factor for Visual Field Progression at Low Intraocular Pressures in the Advanced Glaucoma Intervention Study. Ophthalmology. DOI: 10.1016/j.ophtha.2008.10.013.Caprioli J., Coleman A. L. (2017) Intraocular Pressure Fluctuation: A Risk Factor for Visual Field Progression at Low Intraocular Pressures in the Advanced Glaucoma Intervention Study. Ophthalmology. DOI: 10.1016/j.ophtha.2008.10.013.Chan K., Lee T. W., Sample P. A., Goldbaum M. H. (2019) AI and Glaucoma. Progress in Retinal and Eye Research. DOI: 10.1016/j.preteyeres.2018.07.004.Chan K., Lee T. W., Sample P. A., Goldbaum M. H. (2019) AI and Glaucoma. Progress in Retinal and Eye Research. DOI: 10.1016/j.preteyeres.2018.07.004.Medeiros F. A., Jammal A. A., Thompson A. C. (2018) From Data to Diagnosis: New Techniques for Machine Learning and Image Analysis. British Journal of Ophthalmology. DOI: 10.1136/bjophthalmol-2018-312068.Medeiros F. A., Jammal A. A., Thompson A. C. (2018) From Data to Diagnosis: New Techniques for Machine Learning and Image Analysis. British Journal of Ophthalmology. DOI: 10.1136/bjophthalmol-2018-312068.Leung C. K., Cheung C. Y.-L., Weinreb R. N., Qiu Q., Liu Shu, Li H., et al. (2012) Retinal Nerve Fiber Layer Imaging with Spectral-Domain Optical Coherence Tomography: A Variability and Diagnostic Performance Study. Ophthalmology. DOI: 10.1016/j.ophtha.2011.12.030.Leung C. K., Cheung C. Y.-L., Weinreb R. N., Qiu Q., Liu Shu, Li H., et al. (2012) Retinal Nerve Fiber Layer Imaging with Spectral-Domain Optical Coherence Tomography: A Variability and Diagnostic Performance Study. Ophthalmology. DOI: 10.1016/j.ophtha.2011.12.030.Quigley H. A., McCormick T. A., Nicolela M. T., LeBlanc R. P. (2015) Optic Disc and Visual Field Changes in a Prospective Longitudinal Study of Patients with Glaucoma. Archives of Ophthalmology. DOI: 10.1001/archopht.1996.01100130104019.Quigley H. A., McCormick T. A., Nicolela M. T., LeBlanc R. P. (2015) Optic Disc and Visual Field Changes in a Prospective Longitudinal Study of Patients with Glaucoma. Archives of Ophthalmology. DOI: 10.1001/archopht.1996.01100130104019.Wollstein G., Schuman J. S., Price L. L., Aydin A., Stark P. C., Hertzmark E., et al. (2011) Optical Coherence Tomography Longitudinal Evaluation of Retinal Nerve Fiber Layer Thickness in Glaucoma. Archives of Ophthalmology. DOI: 10.1001/archophthalmol.2011.73.Wollstein G., Schuman J. S., Price L. L., Aydin A., Stark P. C., Hertzmark E., et al. (2011) Optical Coherence Tomography Longitudinal Evaluation of Retinal Nerve Fiber Layer Thickness in Glaucoma. Archives of Ophthalmology. DOI: 10.1001/archophthalmol.2011.73.Hood D. C., Raza A. S. (2011) Method for Comparing Visual Field Defects to Local RNFL and RGC Damage Seen on Frequency Domain OCT in Patients with Glaucoma. Biomed Opt. Express. 2 (5), 1097–1105. DOI: 10.1364/BOE.2.001097.Hood D. C., Raza A. S. (2011) Method for Comparing Visual Field Defects to Local RNFL and RGC Damage Seen on Frequency Domain OCT in Patients with Glaucoma. Biomed Opt. Express. 2 (5), 1097–1105. DOI: 10.1364/BOE.2.001097.Killer H. E., Pircher A. (2018) Normal Tension Glaucoma: Review of Current Understanding And Mechanisms of the Pathogenesis. Eye. 32 (5), 924–930. DOI: 10.1038/s41433-018-0042-2.Killer H. E., Pircher A. (2018) Normal Tension Glaucoma: Review of Current Understanding And Mechanisms of the Pathogenesis. Eye. 32 (5), 924–930. DOI: 10.1038/s41433-018-0042-2.Tan O., Chopra V., Lu A. T.-H., Schuman J. S., Ishikawa H., Wollstein G., et al. (2009) Detection of Macular Ganglion Cell Loss in Glaucoma by Fourier-Domain Optical Coherence Tomography. Ophthalmology. DOI: 10.1016/j.ophtha.2009.05.025.Tan O., Chopra V., Lu A. T.-H., Schuman J. S., Ishikawa H., Wollstein G., et al. (2009) Detection of Macular Ganglion Cell Loss in Glaucoma by Fourier-Domain Optical Coherence Tomography. Ophthalmology. DOI: 10.1016/j.ophtha.2009.05.025.Anderson D. R., Drance S. M., Schulzer M. (2001) Natural History of Normal-Tension Glaucoma. Ophthalmology. DOI: 10.1016/S0161-6420(00)00564-9.Anderson D. R., Drance S. M., Schulzer M. (2001) Natural History of Normal-Tension Glaucoma. Ophthalmology. DOI: 10.1016/S0161-6420(00)00564-9.Bussel I. I., Wollstein G., Schuman J. S. (2014) OCT for Glaucoma Diagnosis, Screening and Detection of Glaucoma Progression. British Journal of Ophthalmology. 98, ii15–ii19. DOI: 10.1136/bjophthalmol-2013-304326.Bussel I. I., Wollstein G., Schuman J. S. (2014) OCT for Glaucoma Diagnosis, Screening and Detection of Glaucoma Progression. British Journal of Ophthalmology. 98, ii15–ii19. DOI: 10.1136/bjophthalmol-2013-304326.Hood D. C., Fortune B., Arthur S. N., Xing D., Salant J. A., Ritch R., et al. (2017) Blood Vessel Contributions to Retinal Nerve Fiber Layer Thickness Profiles Measured with Optical Coherence Tomography. Journal of Glaucoma. 17 (7), 519–528. DOI: 10.1097/IJG.0b013e3181629a02.Hood D. C., Fortune B., Arthur S. N., Xing D., Salant J. A., Ritch R., et al. (2017) Blood Vessel Contributions to Retinal Nerve Fiber Layer Thickness Profiles Measured with Optical Coherence Tomography. Journal of Glaucoma. 17 (7), 519–528. DOI: 10.1097/IJG.0b013e3181629a02.Jonas J. B., Aung T., Bourne R. R., Bron A. M., Ritch R., Panda-Jonas S. (2018) Glaucoma. Lancet. 391 (10108), 2183–2193. DOI: 10.1016/S0140-6736(17)31469-1.Jonas J. B., Aung T., Bourne R. R., Bron A. M., Ritch R., Panda-Jonas S. (2018) Glaucoma. Lancet. 391 (10108), 2183–2193. DOI: 10.1016/S0140-6736(17)31469-1.Quigley H. A., Hohman R. M., Addicks E. M., Massof R. W., Green W. R. (2019) Morphologic Changes in the Lamina Cribrosa Correlated with Neural Loss in Open-Angle Glaucoma. American Journal of Ophthalmology. 95 (5), 673–691. DOI: 10.1016/0002-9394(83)90389-6.Quigley H. A., Hohman R. M., Addicks E. M., Massof R. W., Green W. R. (2019) Morphologic Changes in the Lamina Cribrosa Correlated with Neural Loss in Open-Angle Glaucoma. American Journal of Ophthalmology. 95 (5), 673–691. DOI: 10.1016/0002-9394(83)90389-6.Budenz D. L., Anderson D. R., Varma R., Schuman J., Cantor L., Savell J., et al. (2007) Determinants of Normal Retinal Nerve Fiber Layer Thickness Measured by Stratus OCT. Ophthalmology. 114 (6), 1046–1052. DOI: 10.1016/j.ophtha.2006.08.046.Budenz D. L., Anderson D. R., Varma R., Schuman J., Cantor L., Savell J., et al. (2007) Determinants of Normal Retinal Nerve Fiber Layer Thickness Measured by Stratus OCT. Ophthalmology. 114 (6), 1046–1052. DOI: 10.1016/j.ophtha.2006.08.046.Liu D., Li Z., Zhang X., Zhao L., Jia J., Sun F., et al. (2017) Assessment of Intracranial Pressure with Ultrasonographic Retrobulbar Optic Nerve Sheath Diameter Measurement. BMC Neurology. 17 (1). DOI: 10.1186/s12883-017-0964-5.Liu D., Li Z., Zhang X., Zhao L., Jia J., Sun F., et al. (2017) Assessment of Intracranial Pressure with Ultrasonographic Retrobulbar Optic Nerve Sheath Diameter Measurement. BMC Neurology. 17 (1). DOI: 10.1186/s12883-017-0964-5.Drexler W., Fujimoto J. G. (2008) Optical Coherence Tomography: Technology and Applications. Springer.Drexler W., Fujimoto J. G. (2008) Optical Coherence Tomography: Technology and Applications. Springer.Prudnik A., Usenko P. (2025) Synthetic OCT Glaucoma Dataset. Kaggle. https://doi.org/10.34740/KAGGLE/DS/6470196.Prudnik A., Usenko P. (2025) Synthetic OCT Glaucoma Dataset. Kaggle. https://doi.org/10.34740/KAGGLE/DS/6470196.

The authors declare that there are no conflicts of interest present.