Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Oct 9:11:1438768.
doi: 10.3389/fmed.2024.1438768. eCollection 2024.

Dry age-related macular degeneration classification from optical coherence tomography images based on ensemble deep learning architecture

Jikun Yang  1   2 , Bin Wu  2 , Jing Wang  2 , Yuanyuan Lu  2 , Zhenbo Zhao  1   2 , Yuxi Ding  2 , Kaili Tang  1   2 , Feng Lu  3 , Liwei Ma  1   2
Affiliations

Dry age-related macular degeneration classification from optical coherence tomography images based on ensemble deep learning architecture

Jikun Yang et al. Front Med (Lausanne). .

Abstract

Background: Dry age-related macular degeneration (AMD) is a retinal disease, which has been the third leading cause of vision loss. But current AMD classification technologies did not focus on the classification of early stage. This study aimed to develop a deep learning architecture to improve the classification accuracy of dry AMD, through the analysis of optical coherence tomography (OCT) images.

Methods: We put forward an ensemble deep learning architecture which integrated four different convolution neural networks including ResNet50, EfficientNetB4, MobileNetV3 and Xception. All networks were pre-trained and fine-tuned. Then diverse convolution neural networks were combined. To classify OCT images, the proposed architecture was trained on the dataset from Shenyang Aier Excellence Hospital. The number of original images was 4,096 from 1,310 patients. After rotation and flipping operations, the dataset consisting of 16,384 retinal OCT images could be established.

Results: Evaluation and comparison obtained from three-fold cross-validation were used to show the advantage of the proposed architecture. Four metrics were applied to compare the performance of each base model. Moreover, different combination strategies were also compared to validate the merit of the proposed architecture. The results demonstrated that the proposed architecture could categorize various stages of AMD. Moreover, the proposed network could improve the classification performance of nascent geographic atrophy (nGA).

Conclusion: In this article, an ensemble deep learning was proposed to classify dry AMD progression stages. The performance of the proposed architecture produced promising classification results which showed its advantage to provide global diagnosis for early AMD screening. The classification performance demonstrated its potential for individualized treatment plans for patients with AMD.

Keywords: NGA; dry age-related macular degeneration; early AMD detection; ensemble deep learning; optical coherence tomography.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

Figure 1
Figure 1
Different stages of dry AMD in OCT imaging. (A) Normal. (B) Drusen. (C) nGA. (D) GA.
Figure 2
Figure 2
Results comparison from image enhancement. (A) The original image. (B) The OCT image from (A) with diffusion filtering. (C) The OCT image from (B) with linear enhancement. (D) The OCT image from (C) with the exponential enhancement.
Figure 3
Figure 3
The global ensemble architecture.
Figure 4
Figure 4
The training process of transfer learning.
Figure 5
Figure 5
The combination strategy of the proposed ensemble-based architecture.
Figure 6
Figure 6
The confusion matrix of overall classification results. (A) Majority voting. (B) Stacking. (C) Simple averaging. (D) Weighting function.
Figure 7
Figure 7
The performance of accuracy in 400 epochs. (A) ResNet 50. (B) EfficientNetB4. (C) MobileNetV3. (D) Xception. (E) Proposed.
Figure 8
Figure 8
ROC comparison among different methods. (A) Normal. (B) Drusen. (C) nGA. (D) GA.
Figure 9
Figure 9
Heatmaps from dry AMD with pathological features.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Wu Z, Fletcher EL, Kumar H, Greferath U, Guymer RH. Reticular pseudodrusen: a critical phenotype in age-related macular degeneration. Prog Retin Eye Res. (2022) 88:101017. doi: 10.1016/j.preteyeres.2021.101017, PMID: - DOI - PubMed
    1. Yang S, Zhao J, Sun X. Resistance to anti-VEGF therapy in neovascular age-related macular degeneration: a comprehensive review. Drug Des Devel Ther. (2016):1857–67. doi: 10.2147/DDDT.S97653 - DOI - PMC - PubMed
    1. Mitchell P, Liew G, Gopinath B, Wong TY. Age-related macular degeneration. Lancet. (2018) 392:1147–59. doi: 10.1016/S0140-6736(18)31550-2 - DOI - PubMed
    1. Mrowicka M, Mrowicki J, Kucharska E, Majsterek I. Lutein and zeaxanthin and their roles in age-related macular degeneration—neurodegenerative disease. Nutrients. (2022) 14:827. doi: 10.3390/nu14040827, PMID: - DOI - PMC - PubMed
    1. Schmidt-Erfurth U, Waldstein SM. A paradigm shift in imaging biomarkers in neovascular age-related macular degeneration. Prog Retin Eye Res. (2016) 50:1–24. doi: 10.1016/j.preteyeres.2015.07.007, PMID: - DOI - PubMed

LinkOut - more resources