A machine learning classification model for primary open-angle glaucoma based on optical coherence tomography optic disc parameters and interocular asymmetry features_Chinese Journal of Ocular Fundus Diseases

Authors：

Zhang Kaiqing ¹ , Liang Xuemei ^1,2 ,  Li Li ^1,2 , Liu Yunsheng ³

1. Aier Eye Hospital Affiliated to Jinan University, Guangzhou 510071, China;
2. Nanning Aier Eye Hospital, Nanning 530012, China;
3. Cenxi Aier Eye Hospital, Wuzhou 543200, China;

Corresponding?author：

Li Li, Email: 356588873@qq.com

Keywords：

Primary open angle glaucoma; Optical coherence tomography; Machine learning; Bruch's membrane opening-minimum rim width; Retinal nerve fiber layer; Asymmetry analysis

DOI：

10.3760/cma.j.cn511434-20251021-00461

Video：

Export PDF Favorites Scan Get Citation

Abstract Full text Figures/Tables Video References Cited by

Objective To develop machine learning classification models for primary open-angle glaucoma (POAG) using optical coherence tomography (OCT) optic disc parameters and interocular asymmetry features. Methods A prospective case-control study. From January 2023 to June 2025, 68 patients (136 eyes) diagnosed with POAG (POAG group) and 83 individuals (166 eyes) undergoing routine eye health examinations (healthy controls group) at the Department of Glaucoma of Nanning Aier Eye Hospital were enrolled. The optic disc was scanned using the Glaucoma Module Premium Edition mode of the Spectralis OCT to obtain peripapillary retinal nerve fiber layer (pRNFL) thickness and Bruch's membrane opening-minimum rim width (BMO-MRW) measurements for the global, temporal, superotemporal, superonasal, nasal, inferonasal, and inferotemporal quadrants. The normalized absolute difference in corresponding quadrant parameters between eyes was calculated as interocular asymmetry indix. Elastic Net feature selection was performed on 47 feature variables, and 11 machine learning classification models for POAG were constructed based on the selection results. The overall classification performance was evaluated using achieving an area under the curve (AUC), accuracy, and F1-score. The top five models with the best performance were selected based on AUC. The SHAP (SHapley Additive exPlanations) values were used to analyze the contribution of each feature to prediction outcome in the best-performing model. Results Compared with the healthy control group, both the BMO-MRW and pRNFL thickness in all quadrants were significantly reduced, and interocular asymmetry was significantly greater in POAG group, the differences were all statistically significant (P＜0.05). After feature selection and model comparison, the support vector machine model constructed based on 5 features performed the best. Its AUC was 0.981 (95% confidence interval 0.950-1.000), with an accuracy of 0.957 and an F1 score of 0.952. SHAP analysis indicated that interocular asymmetry in temporal and superotemporal BMO-MRW contributed highly to the model's classification decisions, second only to the inferonasal BMO-MRW thickness of the right eye. Conclusions Machine learning classification models incorporating OCT optic disc parameters and their interocular asymmetry features effectively distinguish POAG patients from healthy individuals. The asymmetry parameters of BMO-MRW between the two eyes play an important role in the model.

Citation： Zhang Kaiqing, Liang Xuemei, Li Li, Liu Yunsheng. A machine learning classification model for primary open-angle glaucoma based on optical coherence tomography optic disc parameters and interocular asymmetry features. Chinese Journal of Ocular Fundus Diseases, 2026, 42(2): 144-154. doi: 10.3760/cma.j.cn511434-20251021-00461 Copy

1.	Zhang N, Wang J, Li Y, et al. Prevalence of primary open angle glaucoma in the last 20 years: a meta-analysis and systematic review[J/OL]. Sci Rep, 2021, 11(1): 13762[2021-07-02]. https://pubmed.ncbi.nlm.nih.gov/34215769/. DOI: 10.1038/s41598-021-92971-w.
2.	Tham YC, Li X, Wong TY, et al. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis[J]. Ophthalmology, 2014, 121(11): 2081-2090. DOI: 10.1016/j.ophtha.2014.05.013.
3.	中華醫學會眼科學分會青光眼學組, 中國醫師協會眼科醫師分會青光眼學組. 中國青光眼指南(2020年)[J]. 中華眼科雜志, 2020, 56(8): 573-586. DOI: 10.3760/cma.j.cn112142-20200313-00182.Glaucoma Group of Ophthalmology Branch of Chinese Medical Association, Glaucoma Group of Chinese Ophthalmologist Association. The Chinese glaucoma guidelines (2020)[J]. Chin J Ophthalmol, 2020, 56(8): 573-586. DOI: 10.3760/cma.j.cn112142-20200313-00182.
4.	Yusof AMZ, Othman O, Tang SF, et al. Diagnostic evaluation of optical coherence tomography parameters in normal, preperimetric and perimetric glaucoma patients[J]. Int J Ophthalmol, 2022, 15(11): 1782-1790. DOI: 10.18240/ijo.2022.11.08.
5.	Hood DC, De Moraes CG. Challenges to the common clinical paradigm for diagnosis of glaucomatous damage with OCT and visual fields[J]. Invest Ophthalmol Vis Sci, 2018, 59(2): 788-791. DOI: 10.1167/iovs.17-23713.
6.	Hammel N, Belghith A, Weinreb RN, et al. Comparing the rates of retinal nerve fiber layer and ganglion cell-inner plexiform layer loss in healthy eyes and in glaucoma eyes[J]. Am J Ophthalmol, 2017, 178: 38-50. DOI: 10.1016/j.ajo.2017.03.008.
7.	Gmeiner JM, Schrems WA, Mardin CY, et al. Comparison of Bruch's membrane opening minimum rim width and peripapillary retinal nerve fiber layer thickness in early glaucoma assessment[J]. Invest Ophthalmol Vis Sci, 2016, 57(9): 575-584. DOI: 10.1167/iovs.15-18906.
8.	Gardiner SK, Boey PY, Yang H, et al. Structural measurements for monitoring change in glaucoma: comparing retinal nerve fiber layer thickness with minimum rim width and area[J]. Invest Ophthalmol Vis Sci, 2015, 56(11): 6886-6891. DOI: 10.1167/iovs.15-16701.
9.	Enders P, Adler W, Schaub F, et al. Novel Bruch's membrane opening minimum rim area equalizes disc size dependency and offers high diagnostic power for glaucoma[J]. Invest Ophthalmol Vis Sci, 2016, 57(15): 6596-6603. DOI: 10.1167/iovs.16-20561.
10.	Kwon YH, Fingert JH, Kuehn MH, et al. Primary open-angle glaucoma[J]. N Engl J Med, 2009, 360(11): 1113-1124. DOI: 10.1056/NEJMra0804630.
11.	Yousefi S. Clinical applications of artificial intelligence in glaucoma[J]. J Ophthalmic Vis Res, 2023, 18(1): 97-112. DOI: 10.18502/jovr.v18i1.12730.
12.	Thanapaisal S, Uttakit P, Ittharat W, et al. Machine learning technology in the classification of glaucoma severity using fundus photographs[J/OL]. Sci Rep, 2025, 15(1): 26151[2025-07-18]. https://pubmed.ncbi.nlm.nih.gov/40681609/. DOI: 10.1038/s41598-025-11697-1.
13.	Yang WYL, Wong HJ, Fu CE, et al. Artificial intelligence in the prediction of glaucoma development and progression: a systematic review[J]. Surv Ophthalmol, 2025, 71(1): 235-247. DOI: 10.1016/j.survophthal.2025.06.006.
14.	Huang X, Islam MR, Akter S, et al. Artificial intelligence in glaucoma: opportunities, challenges, and future directions[J/OL]. Biomed Eng Online, 2023, 22(1): 126[2023-12-16]. https://pubmed.ncbi.nlm.nih.gov/38102597/. DOI: 10.1186/s12938-023-01187-8.
15.	Hoang TT, Van Bui A, Nguyen V, et al. Comparison of perimetric glaucoma staging systems in Asians with primary glaucoma[J]. Eye (Lond), 2021, 35(3): 973-978. DOI: 10.1038/s41433-020-1012-z.
16.	Berenguer-Vidal R, Verdú-Monedero R, Morales-Sánchez J, et al. Decision trees for glaucoma screening based on the asymmetry of the retinal nerve fiber layer in optical coherence tomography[J/OL]. Sensors (Basel), 2022, 22(13): 4842[2022-06-27]. https://pubmed.ncbi.nlm.nih.gov/35808338/. DOI: 10.3390/s22134842.
17.	Chauhan BC, O'leary N, Almobarak FA, et al. Enhanced detection of open-angle glaucoma with an anatomically accurate optical coherence tomography-derived neuroretinal rim parameter[J]. Ophthalmology, 2013, 120(3): 535-543. DOI: 10.1016/j.ophtha.2012.09.055.
18.	Enders P, Adler W, Kiessling D, et al. Evaluation of two-dimensional Bruch's membrane opening minimum rim area for glaucoma diagnostics in a large patient cohort[J]. Acta Ophthalmol, 2019, 97(1): 60-67. DOI: 10.1111/aos.13698.
19.	Li R, Wang X, Wei Y, et al. Diagnostic capability of different morphological parameters for primary open-angle glaucoma in the Chinese population[J/OL]. BMC Ophthalmol, 2021, 21(1): 151[2021-03-25]. https://pubmed.ncbi.nlm.nih.gov/33765982/. DOI: 10.1186/s12886-021-01906-6.
20.	Sullivan-Mee M, Ruegg CC, Pensyl D, et al. Diagnostic precision of retinal nerve fiber layer and macular thickness asymmetry parameters for identifying early primary open-angle glaucoma[J]. Am J Ophthalmol, 2013, 156(3): 567-577. DOI: 10.1016/j.ajo.2013.04.037.
21.	張迪, 張水華, 肖錚, 等. Bruch膜開口-最小盤沿寬度在開角型青光眼中的應用[J]. 眼科學報, 2023, 38(7): 526-534. DOI: 10.12419/2301020001.Zhang D, Zhang SH, Xiao Z, et al. Application of Bruch's membrane opening minimum rim width in open-angle glaucoma[J]. Eye Science, 2023, 38(7): 526-534. DOI: 10.12419/2301020001.
22.	董雯潔, 張麗娜. BMO-MRW增強對原發性開角型青光眼患者的診斷能力[J]. 臨床醫學進展, 2025, 15(5): 2137-2148. DOI: 10.12677/acm.2025.1551602.Dong WJ, Zhang LN. Enhanced detection on optic neuropathy in patients with primary open-angle glucoma ysing BMO-MRW[J]. Adv Clin Med, 2025, 15(5): 2137-2148. DOI: 10.12677/acm.2025.1551602.
23.	Park D, Park SP, Na KI. Comparison of retinal nerve fiber layer thickness and Bruch's membrane opening minimum rim width thinning rate in open-angle glaucoma[J/OL]. Sci Rep, 2022, 12(1): 16069[2022-09-27]. https://pubmed.ncbi.nlm.nih.gov/36167787/. DOI: 10.1038/s41598-022-20423-0.
24.	Wu CW, Shen HL, Lu J, et al. Comparison of different machine learning classifiers for glaucoma diagnosis based on spectralis OCT[J/OL]. Diagnostics (Basel), 2021, 11(9): 1718[2021-09-19]. https://pubmed.ncbi.nlm.nih.gov/34574059/. DOI: 10.3390/diagnostics11091718.
25.	Lin PW, Chang HW, Lai IC, et al. Intraocular retinal thickness asymmetry in early stage of primary open angle glaucoma and normal tension glaucoma[J]. Int J Ophthalmol, 2018, 11(8): 1342-1351. DOI: 10.18240/ijo.2018.08.15.
26.	Kim SJ, Cho KJ, Oh S. Development of machine learning models for diagnosis of glaucoma[J/OL]. PLoS One, 2017, 12(5): e0177726[2017-05-23]. https://pubmed.ncbi.nlm.nih.gov/28542342/. DOI: 10.1371/journal.pone.0177726.
27.	Seo SB, Cho HK. Deep learning classification of early normal-tension glaucoma and glaucoma suspects using Bruch's membrane opening-minimum rim width and RNFL[J/OL]. Sci Rep, 2020, 10(1): 19042[2020-11-04]. https://pubmed.ncbi.nlm.nih.gov/33149191/. DOI: 10.1038/s41598-020-76154-7.
28.	Rodríguez-Robles F, Verdú-Monedero R, Berenguer-Vidal R, et al. Analysis of the asymmetry between both eyes in early diagnosis of glaucoma combining features extracted from retinal images and OCTs into classification models[J/OL]. Sensors (Basel), 2023, 23(10): 4737[2023-05-14]. https://pubmed.ncbi.nlm.nih.gov/37430650/. DOI: 10.3390/s23104737.

1. Zhang N, Wang J, Li Y, et al. Prevalence of primary open angle glaucoma in the last 20 years: a meta-analysis and systematic review[J/OL]. Sci Rep, 2021, 11(1): 13762[2021-07-02]. https://pubmed.ncbi.nlm.nih.gov/34215769/. DOI: 10.1038/s41598-021-92971-w.
2. Tham YC, Li X, Wong TY, et al. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis[J]. Ophthalmology, 2014, 121(11): 2081-2090. DOI: 10.1016/j.ophtha.2014.05.013.
3. 中華醫學會眼科學分會青光眼學組, 中國醫師協會眼科醫師分會青光眼學組. 中國青光眼指南(2020年)[J]. 中華眼科雜志, 2020, 56(8): 573-586. DOI: 10.3760/cma.j.cn112142-20200313-00182.Glaucoma Group of Ophthalmology Branch of Chinese Medical Association, Glaucoma Group of Chinese Ophthalmologist Association. The Chinese glaucoma guidelines (2020)[J]. Chin J Ophthalmol, 2020, 56(8): 573-586. DOI: 10.3760/cma.j.cn112142-20200313-00182.
4. Yusof AMZ, Othman O, Tang SF, et al. Diagnostic evaluation of optical coherence tomography parameters in normal, preperimetric and perimetric glaucoma patients[J]. Int J Ophthalmol, 2022, 15(11): 1782-1790. DOI: 10.18240/ijo.2022.11.08.
5. Hood DC, De Moraes CG. Challenges to the common clinical paradigm for diagnosis of glaucomatous damage with OCT and visual fields[J]. Invest Ophthalmol Vis Sci, 2018, 59(2): 788-791. DOI: 10.1167/iovs.17-23713.
6. Hammel N, Belghith A, Weinreb RN, et al. Comparing the rates of retinal nerve fiber layer and ganglion cell-inner plexiform layer loss in healthy eyes and in glaucoma eyes[J]. Am J Ophthalmol, 2017, 178: 38-50. DOI: 10.1016/j.ajo.2017.03.008.
7. Gmeiner JM, Schrems WA, Mardin CY, et al. Comparison of Bruch's membrane opening minimum rim width and peripapillary retinal nerve fiber layer thickness in early glaucoma assessment[J]. Invest Ophthalmol Vis Sci, 2016, 57(9): 575-584. DOI: 10.1167/iovs.15-18906.
8. Gardiner SK, Boey PY, Yang H, et al. Structural measurements for monitoring change in glaucoma: comparing retinal nerve fiber layer thickness with minimum rim width and area[J]. Invest Ophthalmol Vis Sci, 2015, 56(11): 6886-6891. DOI: 10.1167/iovs.15-16701.
9. Enders P, Adler W, Schaub F, et al. Novel Bruch's membrane opening minimum rim area equalizes disc size dependency and offers high diagnostic power for glaucoma[J]. Invest Ophthalmol Vis Sci, 2016, 57(15): 6596-6603. DOI: 10.1167/iovs.16-20561.
10. Kwon YH, Fingert JH, Kuehn MH, et al. Primary open-angle glaucoma[J]. N Engl J Med, 2009, 360(11): 1113-1124. DOI: 10.1056/NEJMra0804630.
11. Yousefi S. Clinical applications of artificial intelligence in glaucoma[J]. J Ophthalmic Vis Res, 2023, 18(1): 97-112. DOI: 10.18502/jovr.v18i1.12730.
12. Thanapaisal S, Uttakit P, Ittharat W, et al. Machine learning technology in the classification of glaucoma severity using fundus photographs[J/OL]. Sci Rep, 2025, 15(1): 26151[2025-07-18]. https://pubmed.ncbi.nlm.nih.gov/40681609/. DOI: 10.1038/s41598-025-11697-1.
13. Yang WYL, Wong HJ, Fu CE, et al. Artificial intelligence in the prediction of glaucoma development and progression: a systematic review[J]. Surv Ophthalmol, 2025, 71(1): 235-247. DOI: 10.1016/j.survophthal.2025.06.006.
14. Huang X, Islam MR, Akter S, et al. Artificial intelligence in glaucoma: opportunities, challenges, and future directions[J/OL]. Biomed Eng Online, 2023, 22(1): 126[2023-12-16]. https://pubmed.ncbi.nlm.nih.gov/38102597/. DOI: 10.1186/s12938-023-01187-8.
15. Hoang TT, Van Bui A, Nguyen V, et al. Comparison of perimetric glaucoma staging systems in Asians with primary glaucoma[J]. Eye (Lond), 2021, 35(3): 973-978. DOI: 10.1038/s41433-020-1012-z.
16. Berenguer-Vidal R, Verdú-Monedero R, Morales-Sánchez J, et al. Decision trees for glaucoma screening based on the asymmetry of the retinal nerve fiber layer in optical coherence tomography[J/OL]. Sensors (Basel), 2022, 22(13): 4842[2022-06-27]. https://pubmed.ncbi.nlm.nih.gov/35808338/. DOI: 10.3390/s22134842.
17. Chauhan BC, O'leary N, Almobarak FA, et al. Enhanced detection of open-angle glaucoma with an anatomically accurate optical coherence tomography-derived neuroretinal rim parameter[J]. Ophthalmology, 2013, 120(3): 535-543. DOI: 10.1016/j.ophtha.2012.09.055.
18. Enders P, Adler W, Kiessling D, et al. Evaluation of two-dimensional Bruch's membrane opening minimum rim area for glaucoma diagnostics in a large patient cohort[J]. Acta Ophthalmol, 2019, 97(1): 60-67. DOI: 10.1111/aos.13698.
19. Li R, Wang X, Wei Y, et al. Diagnostic capability of different morphological parameters for primary open-angle glaucoma in the Chinese population[J/OL]. BMC Ophthalmol, 2021, 21(1): 151[2021-03-25]. https://pubmed.ncbi.nlm.nih.gov/33765982/. DOI: 10.1186/s12886-021-01906-6.
20. Sullivan-Mee M, Ruegg CC, Pensyl D, et al. Diagnostic precision of retinal nerve fiber layer and macular thickness asymmetry parameters for identifying early primary open-angle glaucoma[J]. Am J Ophthalmol, 2013, 156(3): 567-577. DOI: 10.1016/j.ajo.2013.04.037.
21. 張迪, 張水華, 肖錚, 等. Bruch膜開口-最小盤沿寬度在開角型青光眼中的應用[J]. 眼科學報, 2023, 38(7): 526-534. DOI: 10.12419/2301020001.Zhang D, Zhang SH, Xiao Z, et al. Application of Bruch's membrane opening minimum rim width in open-angle glaucoma[J]. Eye Science, 2023, 38(7): 526-534. DOI: 10.12419/2301020001.
22. 董雯潔, 張麗娜. BMO-MRW增強對原發性開角型青光眼患者的診斷能力[J]. 臨床醫學進展, 2025, 15(5): 2137-2148. DOI: 10.12677/acm.2025.1551602.Dong WJ, Zhang LN. Enhanced detection on optic neuropathy in patients with primary open-angle glucoma ysing BMO-MRW[J]. Adv Clin Med, 2025, 15(5): 2137-2148. DOI: 10.12677/acm.2025.1551602.
23. Park D, Park SP, Na KI. Comparison of retinal nerve fiber layer thickness and Bruch's membrane opening minimum rim width thinning rate in open-angle glaucoma[J/OL]. Sci Rep, 2022, 12(1): 16069[2022-09-27]. https://pubmed.ncbi.nlm.nih.gov/36167787/. DOI: 10.1038/s41598-022-20423-0.
24. Wu CW, Shen HL, Lu J, et al. Comparison of different machine learning classifiers for glaucoma diagnosis based on spectralis OCT[J/OL]. Diagnostics (Basel), 2021, 11(9): 1718[2021-09-19]. https://pubmed.ncbi.nlm.nih.gov/34574059/. DOI: 10.3390/diagnostics11091718.
25. Lin PW, Chang HW, Lai IC, et al. Intraocular retinal thickness asymmetry in early stage of primary open angle glaucoma and normal tension glaucoma[J]. Int J Ophthalmol, 2018, 11(8): 1342-1351. DOI: 10.18240/ijo.2018.08.15.
26. Kim SJ, Cho KJ, Oh S. Development of machine learning models for diagnosis of glaucoma[J/OL]. PLoS One, 2017, 12(5): e0177726[2017-05-23]. https://pubmed.ncbi.nlm.nih.gov/28542342/. DOI: 10.1371/journal.pone.0177726.
27. Seo SB, Cho HK. Deep learning classification of early normal-tension glaucoma and glaucoma suspects using Bruch's membrane opening-minimum rim width and RNFL[J/OL]. Sci Rep, 2020, 10(1): 19042[2020-11-04]. https://pubmed.ncbi.nlm.nih.gov/33149191/. DOI: 10.1038/s41598-020-76154-7.
28. Rodríguez-Robles F, Verdú-Monedero R, Berenguer-Vidal R, et al. Analysis of the asymmetry between both eyes in early diagnosis of glaucoma combining features extracted from retinal images and OCTs into classification models[J/OL]. Sensors (Basel), 2023, 23(10): 4737[2023-05-14]. https://pubmed.ncbi.nlm.nih.gov/37430650/. DOI: 10.3390/s23104737.

Chinese Journal of Ocular Fundus Diseases

A machine learning classification model for primary open-angle glaucoma based on optical coherence tomography optic disc parameters and interocular asymmetry features

Abstract Full text Figures/Tables Video References Cited by

Previous Article

Next Article

Format

Content