AccScience Publishing / AIH / Online First / DOI: 10.36922/AIH026220053
Submit to AIH
Cite this article
2
Download
80
Views
Related Info Links
Google Scholar
More by Authors Links
Fen Zhou
Jianguo Xu
Journal Browser
Volume | Year
Issue
Search
Forthcoming Issue
Current Issue
View All
News and Announcements
View All
ORIGINAL RESEARCH ARTICLE

A multimodal image-based automated method for macular hole boundary detection

Qingxin Yuan1† ,  Rong Tan1† ,  Jianxin Shen1 ,  Fen Zhou2* ,  Jianguo Xu1*
Show Less
1 College of Mechanical and Electrical Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu, China
2 The Affiliated Eye Hospital of Nanjing Medical University, Nanjing, Jiangsu, China
†These authors contributed equally to this work.
AIH, 026220053 https://doi.org/10.36922/AIH026220053
Received: 26 May 2026 | Revised: 4 July 2026 | Accepted: 8 July 2026 | Published online: 23 July 2026
© 2026 by the Author(s). This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution 4.0 International License ( https://creativecommons.org/licenses/by/4.0/ )
Download PDF
XML
Cite
Abstract

Accurate detection of the macular hole (MH) boundary is crucial for diagnosis and treatment planning. Manual annotation by ophthalmologists is time-consuming, poorly reproducible, and subject to substantial inter-observer variability. To address these challenges, this paper proposes an automated MH boundary detection method based on multimodal images. The method mainly consists of the following stages: (i) automatically localizing the scanning line center in the fundus infrared (IR) image using image processing techniques to provide a reference point for subsequent cross-modal mapping; (ii) introducing the YOLOv8-pose keypoint detection model into the task of MH highest point detection, achieving its automated localization in fundus optical coherence tomography (OCT) B-scan images; (iii) constructing a cross-modal mapping algorithm that, through a combination of position transformation and rotation, facilitates the mapping of the MH highest point from OCT B-scan images to IR images; (iv) designing a boundary fitting scheme that connects all mapped points with reference to adjacent scanning lines to form a closed polygon, thereby achieving MH boundary detection. Finally, a series of quantitative and qualitative experiments were conducted to evaluate the performance of the YOLOv8-pose-based keypoint detection model on public and clinical MH datasets for highest-point detection and to validate the effectiveness of the proposed method on a clinical dataset for MH boundary detection. This study presents a new approach to automated MH boundary detection and may facilitate the accurate diagnosis and treatment planning of MH.

Graphical abstract
Keywords
Macular hole
Multimodal image processing
Keypoint detection
Cross-modal mapping
YOLOv8-pose
Funding
This work was financially supported by the Nationaal Natural Science Foundation of China (62401259), the Fundamental Research Funds for the Central Universities (NZ2026006), the Postdoctoral Fellowship Program of CPSF under Grant Number GZC20242228, the General Clinical Trial Project of Nanjing Medical Science and Technology Development Fund (LYM25023), and the High Performance Computing Platform of Nanjing University of Aeronautics and Astronautics.
Conflict of interest
Jianguo Xu is the Youth Editorial Board Member of this journal, but he was not in any way involved in the editorial and peer-review process conducted for this paper, directly or indirectly. Other authors declare that they have no conflict of interest regarding this paper.
References
  1. Tozzi L, Gius I, Greggio A, et al. Case report: spontaneous closure of idiopathic full-thickness macular hole and early development of lamellar hole. Optom Vis Sci. 2023;100(11):804–809. doi: 10.1097/OPX.0000000000002070
  2. Zhang Y, Li J, Yue H, et al. Clinical observation of anatomical and visual outcomes of macular hole after inverted internal limiting membrane flap in patients with idiopathic macular. J Healthc Eng. 2023;2023:5816473. doi: 10.1155/2023/5816473
  3. Wolf S, Wolf-Schnurrbusch U. Spectral-domain optical coherence tomography use in macular diseases: a review. Ophthalmologica. 2010;224(6):333–340. doi: 10.1159/000313814
  4. Singh VK, Kucukgoz B, Murphy DC, Xiong X, Steel DH, Obara B. Benchmarking automated detection of the retinal external limiting membrane in a 3D spectral domain optical coherence tomography image dataset of full thickness macular holes. Comput Biol Med. 2022;140:105070. doi: 10.1016/j.compbiomed.2021.105070
  5. Elhusseiny AM, Schwartz SG, Flynn HW Jr, Smiddy WE. Long-term outcomes after macular hole surgery. Ophthalmol Retina. 2020;4(4):369–376. doi: 10.1016/j.oret.2019.09.015
  6. Singh A, Kendal A, Trivedi D, Cazabon S. Patient expectation and satisfaction after macular hole surgery. Optom Vis Sci. 2011;88(2):312–316. doi: 10.1097/OPX.0b013e3182058fc0
  7. Kucukgoz B, Zou K, Murphy DC, Steel DH, Obara B, Fu H. Uncertainty-aware regression model to predict postoperative visual acuity in patients with macular holes. Comput Med Imaging Graph. 2025;119:102461. doi: 10.1016/j.compmedimag.2024.102461
  8. Lu J, Cheng Y, Hiya FE, et al. Deep-learning-based automated measurement of outer retinal layer thickness for use in the assessment of age-related macular degeneration, applicable to both swept-source and spectral-domain OCT imaging. Biomed Opt Express. 2024;15(1):413–427. doi: 10.1364/BOE.512359
  9. Herath HMSS, Yasakethu SLP, Madusanka N, Yi M, Lee BI. Comparative analysis of deep learning architectures for macular hole segmentation in OCT images: a performance evaluation of U-Net variants. J Imaging. 2025;11(2):53. doi: 10.3390/jimaging11020053
  10. Gribova V, Shalfeeva E. Explainable solutions from artificial intelligence for health-care support systems. Artif Intell Health. 2025;2(3):138–153. doi: 10.36922/aih.5736
  11. Gholami P, Sheikh Hassani M, Parthasarathy MK, Zelek JS, Lakshminarayanan V. Classification of optical coherence tomography images for diagnosing different ocular diseases. In: Multimodal Biomedical Imaging XIII. Vol 10487. SPIE; 2018:1048705. doi: 10.1117/12.2292520
  12. Asani B, Holmberg O, Schiefelbein JB, et al. Evaluation of OCT biomarker changes in treatment-naive neovascular AMD using a deep semantic segmentation algorithm. Eye (Lond). 2024;38(16):3180–3186. doi: 10.1038/s41433-024-03264-1
  13. Volkov E, Averkin A. Modification of U-Net architecture based on Gabor filters with trainable parameters to improve biomarker segmentation on OCT images. In: 2025 VI International Conference on Neural Networks and Neurotechnologies (NeuroNT). IEEE; 2025:66–68. doi: 10.1109/NeuroNT66873.2025.11049967
  14. Alekseev A, Bobe A. GaborNet: Gabor filters with learnable parameters in deep convolutional neural network. In: 2019 International Conference on Engineering and Telecommunication (EnT). IEEE; 2019:1–4. doi: 10.1109/EnT47717.2019.9030571
  15. Jebaseeli TJ, Haldorai A, Kim HJ. Diagnosis of Purtscher retinopathy from spectral-domain OCT images using panoptic segmentation technique. Traitement du Signal. 2025;42(3):1393–1407. doi: 10.18280/ts.420315
  16. Mukherjee S, De Silva T, Duic C, et al. Validation of deep learning–based automatic retinal layer segmentation algorithms for age-related macular degeneration with 2 spectral-domain OCT devices. Ophthalmol Sci. 2025;5(3):100670. doi: 10.1016/j.xops.2024.100670
  17. Cao H, Lian X, Chen L, Duan Z. A computationally efficient and lightweight model for high-accuracy OCT image classification. World Wide Web. 2025;28:35. doi: 10.1007/s11280-025-01348-w
  18. Sharma R, Gangrade J, Gangrade S, Mishra A, Kumar G, Gunjan VK. Modified EfficientNetB3 deep learning model to classify colour fundus images of eye diseases. In: 2023 IEEE 5th International Conference on Cybernetics, Cognition and Machine Learning Applications (ICCCMLA). IEEE; 2023:632–638. doi: 10.1109/ICCCMLA58983.2023.10346769
  19. Phu DN, Tran AQ, Tran NQ, Dang TH. The classification of optical coherence tomography images using machine learning for macular hole and diabetic macular edema diagnoses. In: 2023 12th International Conference on Control, Automation and Information Sciences (ICCAIS). IEEE; 2023:634–638. doi: 10.1109/ICCAIS59597.2023.10382327
  20. Mall PK, Singh PK, Yadav D. GLCM-based feature extraction and medical X-ray image classification using machine learning techniques. In: 2019 IEEE Conference on Information and Communication Technology (CICT). IEEE; 2019:1–6. doi: 10.1109/CICT48419.2019.9066263
  21. Shijin PS, Dharun VS. Extraction of texture features using GLCM and shape features using connected regions. Int J Eng Technol. 2016;8(6):2926–2930. doi: 10.21817/ijet/2016/v8i6/160806254
  22. Wang M, Lin T, Peng Y, et al. Self-guided optimization semi-supervised method for joint segmentation of macular hole and cystoid macular edema in retinal OCT images. IEEE Trans Biomed Eng. 2023;70(7):2013–2024. doi: 10.1109/TBME.2023.3234031
  23. Keller B, Cunefare D, Grewal DS, Mahmoud TH, Izatt JA, Farsiu S. Length-adaptive graph search for automatic segmentation of pathological features in optical coherence tomography images. J Biomed Opt. 2016;21(7):076015. doi: 10.1117/1.JBO.21.7.076015
  24. Wang Y, Zhen L, Tan T, et al. Geometric correspondence-based multimodal learning for ophthalmic image analysis. IEEE Trans Med Imaging. 2024;43(5):1945–1957. doi: 10.1109/TMI.2024.3352602
  25. Zhang H, Yang B, Li S, et al. Retinal OCT image segmentation with deep learning: a review of advances, datasets, and evaluation metrics. Comput Med Imaging Graph. 2025;123:102539. doi: 10.1016/j.compmedimag.2025.102539
  26. Zhang Y, Zhang Y, Li T, Shi M, Huang Y. Analysis of a multimodal ocular ultrasound image classification algorithm based on YOLO and VanillaNet. IEEE Access. 2025;13:148374–148383. doi: 10.1109/ACCESS.2025.3598747
  27. Wen PC, Guan Y, Li JQ, Mahmood T, Zhao YZ. A fundus image myopia diagnosis model based on homogeneous multimodal feature fusion. In: Hung JC, Yen NY, Chang JW, eds. Frontier Computing. Lecture Notes in Electrical Engineering. Vol 1031. Springer; 2023:39–51. doi: 10.1007/978-981-99-1428-9_5
  28. Huang SC, Pareek A, Seyyedi S, Banerjee I, Lungren MP. Fusion of medical imaging and electronic health records using deep learning: a systematic review and implementation guidelines. NPJ Digit Med. 2020;3:136. doi: 10.1038/s41746-020-00341-z
  29. Li Q, Li S, Wu Y, et al. Orientation-independent feature matching (OIFM) for multimodal retinal image registration. Biomed Signal Process Control. 2020;60:101957. doi: 10.1016/j.bspc.2020.101957
  30. Hossein-Nejad Z, Nasri M. A-RANSAC: adaptive random sample consensus method in multimodal retinal image registration. Biomed Signal Process Control. 2018;45:325–338. doi: 10.1016/j.bspc.2018.06.002
  31. Mur-Artal R, Tardós JD. ORB-SLAM2: an open-source SLAM system for monocular, stereo, and RGB-D cameras. IEEE Trans Robot. 2017;33(5):1255–1262. doi: 10.1109/TRO.2017.2705103
  32. de Vos BD, Berendsen FF, Viergever MA, Sokooti H, Staring M, Išgum I. A deep learning framework for unsupervised affine and deformable image registration. Med Image Anal. 2019;52:128–143. doi: 10.1016/j.media.2018.11.010
  33. Çatal O, Jansen W, Verbelen T, Dhoedt B, Steckel J. LatentSLAM: unsupervised multi-sensor representation learning for localization and mapping. In: 2021 IEEE International Conference on Robotics and Automation (ICRA). IEEE; 2021:6739–6745. doi: 10.1109/ICRA48506.2021.9560768
  34. Xie L, Wisse LEM, Pluta J, et al. Automated segmentation of medial temporal lobe subregions on in vivo T1-weighted MRI in early stages of Alzheimer’s disease. Hum Brain Mapp. 2019;40(12):3431–3451. doi: 10.1002/hbm.24607
  35. Park K, Sinha U, Barron JT, et al. Nerfies: deformable neural radiance fields. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. IEEE/CVF; 2021:5865–5874.
  36. Gholami P, Roy P, Parthasarathy MK, Lakshminarayanan V. OCTID: optical coherence tomography image database. Comput Electr Eng. 2020;81:106532. doi: 10.1016/j.compeleceng.2019.106532
  37. Domingues NS. From engineering principles to healthcare practice: a hybrid reasoning framework for transparent clinical decision support. Artif Intell Health. 2026;3(2):025470102. doi: 10.36922/AIH025470102
  38. Abdusalomov A, Mirzakhalilov S, Umirzakova S, et al. Optimized lightweight architecture for coronary artery disease classification in medical imaging. Diagnostics. 2025;15(4):446. doi: 10.3390/diagnostics15040446
  39. Nambiar VB, Ramamurthy B, Veeresha P. Gender determination from periocular images using deep learning based EfficientNet architecture. Int J Math Comput Eng. 2024;2(1):59–70. doi: 10.2478/ijmce-2024-0005

 

Next article in this issue
Share
Back to top
Artificial Intelligence in Health, Electronic ISSN: 3029-2387 Print ISSN: 3041-0894, Published by AccScience Publishing
We use cookies on our website to ensure you get the best experience.
Read more about our cookies here
Accept