Monitoring Age-related Macular Degeneration Progression In Optical Coherence Tomography

Age-related macular degeneration (AMD) is a progressive deterioration of the macula, the central portion of the retina, affecting approximately 196 million individuals worldwide [1]. It typically manifests from the age of 50, becoming more prevalent after 65, leading to a significant decline in visual acuity without completely impairing
vision. AMD is a complex and multifactorial disease, influenced by a combination of genetic and environmental risk factors. Advanced stages of the disease (atrophy and neovascularization) impact nearly 20% of patients, serving as the primary cause of severe vision loss and blindness in developed countries.

Since their introduction in 2007, anti-VEGF treatments have demonstrated remarkable efficacy in mitigating disease progression and even enhancing visual function in neovascular forms of AMD [2]. This effectiveness is optimized by minimizing the time between diagnosis and initiating treatment, along with regular follow-up examinations and retreatment as needed [3]. The indication for anti-VEGF therapy is now widely acknowledged as the presence of exudative signs (subretinal and intraretinal fluid, intraretinal hyperreflective spots, etc.) evident on optical coherence tomography (OCT) [4], a 3D imaging modality. The utilization of AI for AMD prediction [5] primarily focuses on the initial onset of early/intermediate (iAMD), atrophic (GA), and neovascular (nAMD) stages. Notably, there is a lack of research on forecasting AMD progression in patients undergoing anti-VEGF treatment monitoring.

Therefore, reliably detecting changes in neovascularization activity [6] by monitoring exudative signs is crucial for the precise implementation of individualized anti-VEGF treatment strategies. The primary objective of this challenge is to assess existing and novel algorithms [7] for recognizing the evolution of neovascularization activity
in OCT scans of patients with exudative AMD, ultimately aiming to improve the planning of anti-VEGF treatments.

The challenge will address two main tasks:
Task 1: This task focuses on pairs of 2D slices (B-scans) from two consecutive OCT acquisitions. The objective is to classify the evolution between these two slices (before and after), which are typically examined side-by-side on clinicians' screens.
Task 2: This task focuses on the level of individual 2D slices. The goal is to predict the future evolution within three months for patients undergoing close monitoring as part of an anti-VEGF treatment plan. In essence, Task 1 aims to automate the initial step of the analysis (useful for decision support), while Task 2 aims to automate the complete analysis process (valuable for autonomous AI).

Preliminary experiments utilizing a ResNet-50 Siamese [8] baseline algorithm on the dataset indicate the feasibility to predict AMD change using OCT images (Kappa score of 0.55 for task 1). To evaluate performance, the MARIO challenge will leverage independent data from the LAZOUNI Ophthalmology Clinic in Tlemcen, Algeria. The challenge's utilization of a comprehensive dataset comprising data from both French and Algerian populations and images captured using diverse OCT devices underscores its dedication to creating universally applicable solutions that overcome geographic and demographic limitations. This cross-population applicability is paramount to ensuring that the predictive models developed are robust and effective across a wide range of patient groups. This commitment to universality further fosters the adoption of pre-training strategies and promotes the development of population-agnostic approaches, aligning seamlessly with the objectives of this challenge and the MICCAI conference.

References
1 - Jonas, J. B., Cheung, C. M. G., & Panda-Jonas, S. (2017). Updates on the epidemiology of age-related macular degeneration. Asia Pacific Journal of Ophthalmology (Phila), 6(6), 493-497.

2 - Rosenfeld, P. J., Brown, D. M., Heier, J. S., Boyer, D. S., Kaiser, P. K., Chung, C. Y., ... & Macular Photocoagulation Study Group. (2006). Ranibizumab for neovascular age-related macular degeneration. New England Journal of Medicine, 355(14), 1419-1431.

3 - Rasmussen, A., & Sander, B. (2014). Long-term longitudinal study of patients treated with ranibizumab for neovascular age-related macular degeneration. Current Opinion in Ophthalmology, 25(3), 158-163.

4 - Freund, K. B., Korobelnik, J.-F., Devenyi, R., Framme, C., Galic, J., Herbert, E., ... & European Society for Intravitreal Implant and Surgery, Research Committee. (2015). Treat-and-extend regimens with anti-VEGF agents in retinal diseases: A literature review and consensus recommendations. Retina (Philadelphia, Pa), 35(8), 1489-1506.

5 - Bhuiyan A, Wong TY, Ting DSW, Govindaiah A, Souied EH, Smith RT. Artificial Intelligence to Stratify Severity of Age-Related Macular Degeneration (AMD) and Predict Risk of Progression to Late AMD. Transl Vis Sci Technol. 2020 Apr 24;9(2):25. doi: 10.1167/tvst.9.2.25. PMID: 32818086; PMCID: PMC7396183.

6 - Li E, Donati S, Lindsley KB, Krzystolik MG, Virgili G. Treatment regimens for administration of anti-vascular endothelial growth factor agents for neovascular age-related macular degeneration. Cochrane Database Syst Rev. 2020 May 5;5(5):CD012208. doi: 10.1002/14651858.CD012208.pub2. PMID: 32374423; PMCID: PMC7202375.

7 - Emre, T., Chakravarty, A., Rivail, A., Riedl, S., Schmidt-Erfurth, U., Bogunovic, H. (2022). TINC: Temporally Informed Non-contrastive Learning for Disease Progression Modeling in Retinal OCT Volumes. In: Wang, L., Dou, Q., Fletcher, P.T., Speidel, S., Li, S. (eds) Medical Image Computing and Computer Assisted Intervention MICCAI
2022. MICCAI 2022. Lecture Notes in Computer Science, vol 13432. Springer, Cham. https://doi.org/10.1007/978-3-031-16434-7 60

8 - Antoine Rivail, Ursula Schmidt-Erfurth, Wolf DieterVogel, Sebastian M. Waldstein, Sophie Riedl, Christoph Grechenig, Zhichao Wu, and Hrvoje Bogunovic, Modeling disease progression in retinal octs with longitudinal self-supervised learning, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11843 LNCS, pp. 44-52, 2019.