Diagnostics of retinal pathologies by optical coherence tomography images using artificial intelligence tools

The importance of early detection and monitoring of retinal diseases determines the relevance of the study devoted to the diagnosis of retinal pathologies by OCT images using artificial intelligence (AI) tools.
The purpose is to develop algorithms for diagnosing retinal pathologies from OCT images by machine learning methods.
Material and methods. The study used a dataset (20,000 eyes), publicly available on the Internet, which contains OCT images of healthy retina (5,000 eyes) and retina affected by three different pathologies (choroid neovascularization, macular edema, multiple drusen, 15,000 eyes). The retinal pathology recognition system is based on a trained neural network VGG16 (developed by a visual geometry group of Oxford University).
Results. The main result of the research is the development of an algorithm, implemented on Python, for the diagnosis of retinal diseases from OCT images based on convolutional neural network AI tool. The sensitivity and selectiveness of the neural network model during the diagnosis of retinal diseases were 97 and 98%, respectively.
Conclusion. AI methods used in the retinal pathology automatic detection system developed at the Helmholtz National Medical Research Center of Eye Diseases as part of automated medical decision-making system have been shown to have high potential and efficiency. In the future, this service can be used to improve the effectiveness of early diagnosis and monitoring of retinal diseases in conditions of reduced availability of primary ophthalmological care in some of the territories of the Russian Federation, including that provided at the pre-doctoral stage.

1. Monroy GL, Won J, Spillman DR, Dsouza R, Boppart SA. Clinical translation of handheld optical coherence tomography: practical considerations and recent advancements. J Biomed Opt. 2017; 22 (12): 1–30. doi: 10.1117/1.JBO.22.12.121715

2. Das R, Spence G, Hogg RE, Stevenson M, Chakravarthy U. Disorganization of inner retina and outer retinal morphology in diabetic macular edema. JAMA Ophthalmol. 2018; 136 (2): 202–8. doi: 10.1001/jamaophthalmol.2017.6256

3. Schneider EW, Fowler SC. Optical coherence tomography angiography in the management of age-related macular degeneration. Curr Opin Ophthalmol. 2018; 29 (3): 217–25. doi: 10.1097/ICU.0000000000000469

4. Herranz-Cabarcos A, Vega-L pez Z, Salas-Fandos O, et al. Macular optical coherence tomography for screening of pathology prior to cataract surgery: An approach based on tele-evaluation. Eur J Ophthalmol. 2022; 32 (6): 3433–7. doi: 10.1177/11206721221080818

5. Age-related macular degeneration. Clinical recommendations. Available at: https://cr.minzdrav.gov.ru/recomend/114_2. (Accessed 10. Feb. 2023).

6. Vira J, Marchese A, Singh RB, Agarwal A. Swept-source optical coherence tomography imaging of the retinochoroid and beyond. Expert Rev Med Devices. 2020; 17 (5): 413–26. doi: 10.1080/17434440.2020.1755256

7. Fujimoto J, Swanson E. The development, commercialization, and impact of optical coherence tomography. Invest. Ophthalmol. Vis. Sci. 2016; 57 (9): OCT1–OCT13. doi: 10.1167/iovs.16-19963

8. Hu J, Chen Y, Yi Z. Automated segmentation of macular edema in OCT using deep neural networks. Med. Image Anal. 2019; 55: 216–27. doi: 10.1016/j.media.2019.05.002

9. Lu D, Heisler M, Sieun Lee S, et al. Deep-learning based multiclass retinal fluid segmentation and detection in optical coherence tomography images using a fully convolutional neural network. Med Image Anal. 2019; 54: 100–10. doi: 10.1016/j.media.2019.02.011

10. Soomro TA, Afifi A, Zheng L, Soomro S, et al. Deep learning models for retinal blood vessels segmentation: a review. IEEE Access. 2019; 7: 71696–717. doi: 10.1109/ACCESS.2019.2920616

11. Feng Z, Yang J, Yao L, et al. Deep retinal image segmentation: a FCN-based architecture with short and long skip connections for retinal image segmentation. ICONIP. Springer; 2017: 713–22. Available at: https://www.researchgate.net/publication/320580185_Deep_Retinal_Image_Segmentation_A_FCNBased_Architecture_with_Short_and_Long_Skip_Connections_for_Retinal_Image_Segmentation (Accessed 03.Nov. 2022).

12. Asgar R, Waldstein S, Schlanitz F, et al. U-Net with spatial pyramid pooling for drusen segmentation in optical coherence tomography. OMIA: Springer; 2019: 77–85 Available at: https://arxiv.org/pdf/1912.05404v1.pdf (Accessed 06. Jul. 2022).

13. Neroev V.V., Bragin A.A., Zaytseva O.V. Development of a prototype service for the diagnosis of diabetic retinopathy based on fundus photos using artificial intelligence methods. National Health Care (Russia). 2021; 2 (2): 64–72 (In Russ.). doi: 10.47093/2713- 069X.2021.2.2.64-72

14. Esfahani EN, Daneshmand PG, Rabbani H, Plonka G. Automatic classification of macular diseases from OCT images using CNN guided with edge convolutional layer. Annu Int. Conf. IEEE Eng. Med. Biol. Soc. 2022; 2022: 3858–61. doi: 10.1109/EMBC48229.2022.9871322

15. Kermany DS, Goldbaum M, Cai W, et al. Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell. 2018; 172 (5): 1122–31. doi: https://doi.org/10.1016/j.cell.2018.02.010. Available at: http://www.cell.com/cell/fulltext/S0092-8674(18)30154-5 (Accessed 20.01.2023).

16. Julli A., Pal S. Deep learning with Keras. Moscow: DMK Press; 2017 (In Russ.).

17. Nielsen MA. Neural networks and deep learning. Determination Press. 2018. Available at: https://static.latexstudio.net/article/2018/0912/neuralnetworksanddeeplearning.pdf (Accessed (03. Jan.2023).