Decoding the Glaucomatous Optic Nerve Head: AI-Driven Structural Phenotype Exploration

 Corresponding author: Namrata Srivastava Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution License 4.0. Decoding the Glaucomatous Optic Nerve Head: AI-Driven Structural Phenotype Exploration Namrata Srivastava 1, * and Mohd. Javed Akhtar 2 1 Department of Optometry ERA University of Health Science and Research, Lucknow, U.P India. 2 Department of Ophthalmology, Sharp Sight eye hospital New Delhi. World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 Publication history: Received on 01 September 2025; revised on 06 October 2025; accepted on 09 October 2025 Article DOI: https://doi.org/10.30574/wjarr.2025.28.1.3483 Abstract Glaucoma, a leading cause of irreversible blindness, is characterized by progressive optic nerve head (ONH) damage and subsequent visual field loss. Early diagnosis and precise monitoring are critical to preventing vision impairment. Traditional ONH assessment methods, though valuable, often rely on subjective interpretation, risking oversight of early glaucomatous changes. Optical coherence tomography (OCT) has revolutionized ONH evaluation by delivering highresolution, quantitative structural data. This review explores the pivotal role of Artificial Intelligence (AI) in analysing complex OCT datasets to improve glaucoma diagnosis and progression monitoring. AI techniques, including machine learning and deep learning, provide automated, accurate detection of pathological ONH changes, offering superior accuracy and efficiency compared to conventional methods. These tools decode intricate structural features, enabling timely interventions and reducing diagnostic variability. We examine the limitations of traditional approaches, including their subjectivity and inconsistency, and highlight advancements in OCT imaging that provide detailed, reproducible data. AI integration with OCT facilitates objective, reliable assessments, potentially enhancing patient outcomes by minimizing diagnostic delays. Convolutional neural networks and predictive modeling are highlighted for their ability to identify early glaucomatous changes and forecast disease progression. This paper emphasizes AI-driven ONH analysis as a solution to unmet needs in glaucoma management, offering a pathway to personalized, data-driven management. Future directions include integrating these technologies into routine clinical practice to optimize early detection and treatment, ultimately improving quality of life for glaucoma patients. Keywords: Glaucoma; Optic Nerve Head; Optical Coherence Tomography; Artificial Intelligence; Machine Learning; Deep Learning; Structural Analysis; Early Diagnosis; Disease Progression; Predictive Modelling 1. Introduction Glaucoma is a multifactorial optic neuropathy affecting millions worldwide. Its insidious onset and asymptomatic progression in early stages often lead to delayed diagnosis and treatment, resulting in irreversible vision loss. The hallmark of glaucoma is characteristic damage to the optic nerve head, specifically the retinal ganglion cell axons that form the optic nerve and the overlying neuroretinal rim. This structural damage is typically accompanied by corresponding visual field defects [1, 2]. Understanding the intricate relationship between ONH structure and visual function is paramount for glaucoma management [3, 4]. Traditional methods for assessing ONH damage include direct ophthalmoscopy, stereoscopic fundus photography, and visual field perimetry. While these methods have been instrumental, they are prone to observer variability and may not be sensitive enough to detect early structural changes [5]. The development of high-resolution imaging modalities, such as spectral-domain optical coherence tomography (SD-OCT), has provided unprecedented insights into the World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 775 microanatomy of the ONH [6]. OCT allows for cross-sectional imaging of the retina and ONH, enabling quantitative measurement of crucial parameters like retinal nerve fibre layer (RNFL) thickness, neuroretinal rim area, and importantly, detailed assessment of the lamina cribrosa (LC) and surrounding scleral structures [7, 8]. Figure 1 Schematic Diagram of the Optic Nerve Head (ONH) Despite the wealth of information provided by OCT, analyzing these complex, multidimensional datasets for subtle glaucomatous changes remains a significant challenge for clinicians. This is where Artificial Intelligence (AI), particularly machine learning and deep learning, is emerging as a transformative tool. AI algorithms can learn intricate patterns and correlations within large OCT datasets that may be imperceptible to the human eye [9]. This paper aims to provide a comprehensive overview of AI applications in the structural analysis of the optic nerve head for glaucoma detection and management. 1.1. The Optic Nerve Head: A Complex Anatomical Structure: The optic nerve head is a complex anatomical region where the axons of retinal ganglion cells converge to form the optic nerve. It is characterized by several key structures: • Neuroretinal Rim: The area between the optic disc margin and the central cup. Glaucomatous damage typically leads to thinning and excavation of the neuroretinal rim [10]. • Lamina Cribrosa (LC): A sieve-like structure composed of connective tissue beams and pores through which the optic nerve fibres pass from the eye to the brain. The LC plays a crucial role in supporting the optic nerve and has been implicated in glaucoma pathogenesis. Its morphology, including depth and excursion, can be significantly altered in glaucoma [11, 12, 13]. Studies have shown that the LC's anterior migration and posterior displacement are associated with glaucomatous damage [14, 15, 16]. Furthermore, the overall connective tissue phenotype of the glaucomatous cup, including the scleral canal and LC, is a subject of intense research [17, 18]. • Bruch's Membrane Opening (BMO): The opening in Bruch's membrane through which the optic nerve exits the eye. The BMO-minimum rim width is a sensitive parameter for detecting early glaucomatous loss, as it defines the innermost extent of the neural rim [19, 20]. • Peripapillary Sclera: The scleral tissue surrounding the optic disc. Recent research suggests that the sclera itself can exhibit characteristic deformations in glaucoma, potentially offering additional diagnostic clues [21]. The interplay between these structures and their changes over time are central to understanding glaucoma progression. Accurately quantifying and interpreting these complex geometric and structural alterations is essential for diagnosis. 1.2. Advancements in Optic Nerve Head Imaging: The Role of OCT: Optical coherence tomography (OCT) has revolutionized ONH imaging by providing high-resolution, cross-sectional images of the retina and ONH. Early OCT systems provided primarily en-face imaging and limited cross-sectional views. The advent of spectral-domain OCT (SD-OCT) offered significant improvements in speed and resolution, enabling detailed analysis of RNFL thickness and neuroretinal rim area [6]. World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 776 Table 1 OCT Advancements in ONH Imaging OCT Modality Key Features Relevance to Glaucoma SD-OCT Cross-sectional imaging, RNFL & rim analysis Baseline structural evaluation 3D OCT Volumetric ONH datasets Detect subtle abnormalities High-Resolution OCT Microstructure visualization (LC pores, sclera) Early structural biomarkers OCTA Microvascular imaging Vascular role in glaucoma Biomechanical OCT Inference of ONH/scleral stiffness Predictive of damage risk More recent advancements in OCT technology are focusing on: • 3D OCT: Capturing volumetric datasets of the ONH, allowing for more comprehensive structural analysis and the identification of subtle abnormalities. • High-Resolution OCT: Achieving resolutions on the order of micrometers, enabling visualization of finer microstructures within the ONH, including the pores of the lamina cribrosa and the dural border tissue. • Advanced OCT Angiography (OCTA): Visualizing microvascular changes in the ONH and peripapillary region, which are increasingly recognized as playing a role in glaucoma pathogenesis [22]. • Biomechanical OCT: Investigating the mechanical properties of the ONH and peripapillary sclera. While direct biomechanical testing is invasive, AI models are being developed to infer biomechanical robustness from structural OCT data [23]. These imaging advancements generate vast amounts of intricate structural data, making manual interpretation challenging and time-consuming. This is where AI excels. 1.3. AI Applications in ONH Structural Analysis for Glaucoma: AI, particularly deep learning, has shown remarkable promise in analyzing OCT data for glaucoma detection and progression monitoring. The ability of AI models to learn complex patterns and features from large datasets makes them ideally suited for this task. Key AI applications in ONH structural analysis include: • Automated Segmentation and Feature Extraction: o Deep learning models, such as Convolutional Neural Networks (CNNs), are adept at automatically segmenting critical ONH structures (e.g., optic disc margin, cup, RNFL, Bruch's membrane opening) from OCT volumes [9, 24]. Accurate segmentation is the foundational step for all subsequent quantitative analysis. o These models can then extract a wide array of quantitative structural features, including those related to the neuroretinal rim, RNFL thickness, and, more recently, the intricate geometry of the lamina cribrosa [25, 26]. PointNet and similar architectures are being explored for analyzing 3D point cloud data derived from OCT, offering novel ways to capture geometric nuances of the ONH [27]. • Glaucoma Detection and Classification: o AI models can be trained on large datasets of OCT scans from healthy individuals and glaucoma patients to learn the subtle structural patterns indicative of the disease. o These models can classify ONH structures as healthy, suspect, or glaucomatous with high accuracy, potentially outperforming traditional clinical assessments in certain scenarios, especially for earlystage or preperimetric glaucoma [9, 26]. o AI can also assist in staging glaucoma based on structural damage, providing a more objective measure than existing clinical staging systems that heavily rely on visual field loss [28]. World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 777 • Progression Monitoring: o One of the most critical challenges in glaucoma management is detecting and quantifying disease progression. Serial OCT scans over time can reveal subtle changes in ONH structure. o AI algorithms can be trained to compare longitudinal OCT scans and identify statistically significant structural changes in RNFL thickness, neuroretinal rim area, or LC morphology that indicate progression [6, 9]. o This automated progression analysis can alert clinicians to patients who require closer monitoring or intensified treatment. • Structure-Function Relationship Analysis: o A fundamental aspect of glaucoma is the correlation between structural damage at the ONH and visual field deficits. However, this relationship can be complex and vary with disease severity [4, 29]. o AI can help unravel these complex structure-function relationships by analyzing large, integrated datasets that include both OCT structural data and visual field test results [30]. This can lead to more accurate predictions of functional loss based on structural findings. o For instance, AI might identify specific structural biomarkers of the ONH that are more predictive of certain types of visual field defects, improving diagnostic precision. . Figure 2 Flowchart of AI-based ONH Analysis • Predictive Modeling and Risk Stratification: o Beyond diagnosis and progression, AI can be used to develop predictive models for future glaucoma development or progression, especially in at-risk individuals. o By analyzing a combination of structural ONH features, genetic predispositions, and other risk factors, AI could stratify patients based on their likelihood of developing or progressing in glaucoma. World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 778 Table 2 AI Applications in ONH Structural Analysis AI Task Technique(s) Used Clinical Relevance Segmentation CNNs, PointNet Automated ONH/RNFL analysis Disease Detection CNNs, Deep Ensemble Models Early glaucoma screening Progression Monitoring RNNs, Longitudinal CNNs Tracking subtle changes Structure-Function Analysis Multimodal Deep Learning Predicts visual field loss Risk Stratification Predictive Modeling + Genetics Identifies high-risk patients Specific AI Methodologies and Their Relevance: • Convolutional Neural Networks (CNNs): Particularly effective for analyzing image-based data like OCT scans. They can learn hierarchical features from raw pixel data, enabling automated segmentation, classification, and detection of abnormalities [9, 24]. • Recurrent Neural Networks (RNNs) and LSTMs: Useful for analyzing sequential data, such as serial OCT scans for progression monitoring. They can capture temporal dependencies and identify trends indicating worsening disease. • Graph Neural Networks (GNNs) and Geometric Deep Learning: With the increasing focus on the 3D geometry of the ONH, including the intricate structures of the lamina cribrosa, GNNs and related geometric deep learning approaches are becoming highly relevant. These methods can directly operate on irregular 3D data (like point clouds or meshes) derived from high-resolution OCT, allowing for the identification of critical 3D structural features relevant to glaucoma pathogenesis that might be missed by traditional image-based CNNs [27, 26]. For example, they can analyze the complex curvature and connectivity of the LC beams. • Ensemble Methods: Combining the predictions of multiple AI models can often lead to improved robustness and accuracy, reducing the risk of over-reliance on a single model's potential biases. 2. Challenges and Future Directions Despite the immense potential, several challenges need to be addressed: • Data Requirements: AI models, especially deep learning ones, require large, diverse, and well-annotated datasets for training. Acquiring such datasets with consistent imaging protocols and clinical outcomes is a significant undertaking. • Generalizability and Bias: AI models trained on data from one population or imaging device may not perform as well on data from different sources. Ensuring generalizability and mitigating biases related to race, ethnicity, and age is critical [30]. • Interpretability (Explainable AI - XAI): "Black box" AI models can be a concern in clinical decision-making. Developing explainable AI methods that can articulate why a certain diagnosis or prediction is made is crucial for clinician trust and adoption. Understanding which specific structural features contribute most to an AI's decision is vital for clinical validation. • Integration into Clinical Workflow: Seamless integration of AI tools into existing ophthalmology workflows is necessary for practical implementation. This includes user-friendly interfaces and efficient data processing pipelines. • Regulatory Approval: AI algorithms used for medical diagnosis require rigorous validation and regulatory approval from health authorities to ensure safety and efficacy. • Focus on the Lamina Cribrosa and Biomechanics: While RNFL and rim area analysis are well-established; AI's ability to precisely characterize the 3D structure and mechanical properties of the lamina cribrosa and peripapillary sclera for glaucoma diagnosis remains an active area of research [12, 13, 23, 26]. Geometric deep learning holds significant promise here. • Early Detection of Preperimetric Glaucoma: AI's ability to identify subtle structural changes might be particularly valuable in detecting preperimetric glaucoma, where visual field defects are not yet apparent [7, 8]. World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 779 Table 3 Key Challenges and Proposed Solutions Challenge Problem Description Proposed Solutions Data Limitations Small, non-uniform datasets Multi-center collaborations, federated learning Generalizability Bias across populations/devices Domain adaptation, large diverse datasets Interpretability Black-box nature of deep learning Explainable AI (saliency maps, SHAP, LIME) Integration Workflow disruptions Plug-in AI tools within OCT platforms Regulatory Approval Clinical validation required Prospective trials, FDA/EMA approvals 3. Conclusion The synergy between advanced OCT imaging and Artificial Intelligence is poised to transform the landscape of glaucoma diagnosis and management. AI algorithms offer the potential to objectively and efficiently analyze the complex structural data of the optic nerve head, moving beyond the limitations of traditional assessment. By automating segmentation, extracting intricate structural features, detecting subtle abnormalities, and monitoring progression, AI can empower clinicians with more precise diagnostic tools, leading to earlier intervention and improved patient outcomes. Continued research into explainable AI, robust datasets, and novel geometric deep learning approaches will further unlock the potential of AI in decoding the structural intricacies of the optic nerve head, ultimately contributing to the fight against this sight-threatening disease. Compliance with ethical standards Acknowledgments The authors express their sincere gratitude to all researchers and clinicians whose work has contributed to the advancement of glaucoma imaging and Artificial Intelligence applications in ophthalmology. Special thanks are extended to the institutions and laboratories that provided access to scientific resources and reference materials essential for this review. The authors also acknowledge the guidance and constructive feedback of peers during manuscript preparation. Disclosure of conflict of interest No conflict of interest to be disclosed. Funding No specific funding was received for this study and the authors declare no conflicts of interest. References [1] Mills RP, Budenz DL, Lee PP, et al. categorizing the stage of glaucoma from pre-diagnosis to end-stage disease. Am J Ophthalmol. 2006; 141(1):24-30. doi:10.1016/j.ajo.2005.07.044 [2] Robin TA, Müller A, Rait J, Keeffe JE, Taylor HR, Mukesh BN. Performance of community-based glaucoma screening using Frequency Doubling Technology and Heidelberg Retinal Tomography. Ophthalmic Epidemiol. 2005; 12(3):167-178. doi:10.1080/09286580590969716 [3] Gonzalez-Hernandez M, Pablo LE, Armas-Dominguez K, de la Vega RR, Ferreras A, de la Rosa MG. Structurefunction relationship depends on glaucoma severity. Br J Ophthalmol. 2009; 93(9):1195-1199. doi:10.1136/bjo.2008.154815 [4] Medeiros FA, Zangwill LM, Bowd C, Mansouri K, Weinreb RN. The structure and function relationship in glaucoma: implications for detection of progression and measurement of rates of change. Invest Ophthalmol Vis Sci. 2012; 53(11):6939-6946. doi:10.1167/iovs.12-10345 [5] Bussel II, Wollstein G, Schuman JS. OCT for glaucoma diagnosis, screening and detection of glaucoma progression. Br J Ophthalmol. 2014; 98(Suppl 2)(suppl 2):ii15-ii19. doi:10.1136/bjophthalmol-2013-304326 World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 780 [6] Lavinsky F, Wollstein G, Tauber J, Schuman JS. The future of imaging in detecting glaucoma progression. Ophthalmology. 2017; 124(12S):S76-S82. doi:10.1016/j.ophtha.2017.10.011 [7] Kim NR, Lee ES, Seong GJ, Kim JH, An HG, Kim CY. Structure-function relationship and diagnostic value of macular ganglion cell complex measurement using Fourier-domain OCT in glaucoma. Invest Ophthalmol Vis Sci. 2010; 51(9):4646-4651. doi:10.1167/iovs.09-5053 [8] Kim YJ, Kang MH, Cho HY, Lim HW, Seong M. Comparative study of macular ganglion cell complex thickness measured by spectral-domain optical coherence tomography in healthy eyes, eyes with preperimetric glaucoma, and eyes with early glaucoma. Jpn J Ophthalmol. 2014;58(3):244-251. doi:10.1007/s10384-014-0315-7 [9] Hood DC, La Bruna S, Tsamis E, et al. Detecting glaucoma with only OCT: implications for the clinic, research, screening, and AI development. Prog Retin Eye Res. 2022; 90:101052. doi:10.1016/j.preteyeres.2022.101052 [10] Danthurebandara VM, Sharpe GP, Hutchison DM, et al. Enhanced structure-function relationship in glaucoma with an anatomically and geometrically accurate neuroretinal rim measurement. Invest Ophthalmol Vis Sci. 2014; 56(1):98-105. doi:10.1167/iovs.14-15375 [11] Quigley HA, Hohman RM, Addicks EM, Massof RW, Green WR. Morphologic changes in the lamina cribrosa correlated with neural loss in open-angle glaucoma. Am J Ophthalmol. 1983; 95(5):673-691. doi:10.1016/00029394(83)90389-6 [12] Park SC, Brumm J, Furlanetto RL, et al. Lamina cribrosa depth in different stages of glaucoma. Invest Ophthalmol Vis Sci. 2015; 56(3):2059-2064. doi:10.1167/iovs.14-15540 [13] Lee SH, Kim TW, Lee EJ, Girard MJ, Mari JM. Diagnostic power of lamina cribrosa depth and curvature in glaucoma. Invest Ophthalmol Vis Sci. 2017; 58(2):755-762. doi:10.1167/iovs.16-20802 [14] Tan NYQ, Tham YC, Thakku SG, et al. Changes in the anterior lamina cribrosa morphology with glaucoma severity. Sci Rep. 2019; 9(1):6612. doi:10.1038/s41598-019-42649-1 [15] Vianna JR, Lanoe VR, Quach J, et al. Serial changes in lamina cribrosa depth and neuroretinal parameters in glaucoma: impact of choroidal thickness. Ophthalmology. 2017; 124(9):1392-1402. doi:10.1016/j.ophtha.2017.03.048 [16] Yang H, Williams G, Downs JC, et al. Posterior (outward) migration of the lamina cribrosa and early cupping in monkey experimental glaucoma. Invest Ophthalmol Vis Sci. 2011; 52(10):7109-7121. doi:10.1167/iovs.11-7448 [17] Yang H, Reynaud J, Lockwood H, et al. The connective tissue phenotype of glaucomatous cupping in the monkey eye: clinical and research implications. Prog Retin Eye Res. 2017; 59:1-52. doi:10.1016/j.preteyeres.2017.03.001 [18] Bellezza AJ, Rintalan CJ, Thompson HW, Downs JC, Hart RT, Burgoyne CF. Deformation of the lamina cribrosa and anterior scleral canal wall in early experimental glaucoma. Invest Ophthalmol Vis Sci. 2003;44(2):623-637. doi:10.1167/iovs.01-1282 [19] Park K-H, Lee JW, Kim JM, Nouri-Mahdavi K, Caprioli J. Bruch’s membrane opening-minimum rim width and visual field loss in glaucoma: a broken stick analysis. Int J Ophthalmol. 2018; 11(5):828-834. [20] Jonas JB, Gründler AE. Correlation between mean visual field loss and morphometric optic disk variables in the open-angle glaucomas. Am J Ophthalmol. 1997;124(4):488-497. doi:10.1016/S0002-9394(14)70864-5 [21] Wang X, Tun TA, Nongpiur ME, et al. Peripapillary sclera exhibits a v-shaped configuration that is more pronounced in glaucoma eyes. Br J Ophthalmol. 2022;106(4):491-496. doi:10.1136/bjophthalmol-2020-317900 [22] Brooks DE, Källberg ME, Cannon RL, et al. Functional and structural analysis of the visual system in the rhesus monkey model of optic nerve head ischemia. Invest Ophthalmol Vis Sci. 2004;45(6):1830-1840. doi:10.1167/iovs.03-0950 [23] Braeu FA, Chuangsuwanich T, Tun TA, et al. AI-based clinical assessment of optic nerve head robustness superseding biomechanical testing. Br J Ophthalmol. 2023; BJO-2022-322374. doi:10.1136/bjo-2022-322374 [24] Devalla SK, Pham TH, Panda SK, et al. Towards label-free 3D segmentation of optical coherence tomography images of the optic nerve head using deep learning. Biomed Opt Express. 2020;11(11):6356-6378. doi:10.1364/BOE.395934 [25] Amini N, Daneshvar R, Sharifipour F, et al. Structure-function relationships in perimetric glaucoma: comparison of minimum-rim width and retinal nerve fiber layer parameters. Invest Ophthalmol Vis Sci. 2017;58(11):46234631. doi:10.1167/iovs.17-21936 World Journal of Advanced Research and Reviews, 2025, 28(01), 774-781 781 [26] Braeu FA, Thiéry AH, Tun TA, et al. Geometric deep learning to identify the critical 3D structural features of the optic nerve head for glaucoma diagnosis. Am J Ophthalmol. 2023;250:38-48. doi:10.1016/j.ajo.2023.01.008 [27] Qi CR, Su H, Mo K, Guibas LJ. PointNet: deep learning on point sets for 3D classification and segmentation. Presented at: IEEE Conference on Computer Vision and Pattern Recognition (CVPR); July 21, 2017; Honolulu, Hawaii. [28] De Moraes CG, Sun A, Jarukasetphon R, et al. Association of macular visual field measurements with glaucoma staging systems. JAMA Ophthalmol. 2019;137(2):139-145. doi:10.1001/jamaophthalmol.2018.5398 [29] Wollstein G, Kagemann L, Bilonick RA, et al. Retinal nerve fibre layer and visual function loss in glaucoma: the tipping point. Br J Ophthalmol. 2012;96(1):47-52. doi:10.1136/bjo.2010.196907 [30] Tham YC, Li X, Yu M, et al. Artificial intelligence for glaucoma detection: generalizability across different populations. Ophthalmology. 2020;127(12):1632-1639. doi:10.1016/j.ophtha.2020.05.003