Abstract
Relevance. Nearly 200 million people worldwide suffer from agerelated macular degeneration (AMD), 10% of which is neovascular, the cause of severe vision loss for most patients. Vascular endothelial growth factor inhibitors (anti-VEGF therapy) make it possible to achieve regression of the neovascularization process and preserve vision. However, today it is a rather expensive method of treatment, which is accompanied by various complications. The neovascular form of agerelated macular degeneration is the most common cause of such a complication as rupture of the pigment epithelium. Predictors of this anatomical outcome, as well as predictors of functional outcome or final visual acuity, can be assessed using optical coherence tomography (OCT). To automatize the processes of identifying morphological structures in OCT images deep learning methods are used. Purpose. The aim of this work was to create an algorithm for the automated detection of the antiVEGF therapy outcome biomarkers in patients with n-AMD and PED on OCT images. Material and methods. We used a set of retrospective data in the form of 251 annotated OCT images obtained during the initial examination of patients who were treated with n-AMD using anti-VEGF therapy from 2014 to 2021 to develop a segmentation algorithm. The architecture of the neural network was a convolutional neural network UNET. To evaluate the effectiveness of the proposed model, the Dice coefficient (DSC) was used. Results. The segmentation accuracy showed high values for the determination of all biomarkers – from 0.97 to 0.99. For retinal pigment epithelium detachment, DSC shows a good value of 0.8. However, for the pigment epithelium and subretinal fluid, DSC values are 0.4, and for other biomarkers from 0.3 to 0.15. Conclusion. The obtained results of segmentation of OCT images showed a high accuracy of pixel determination (accuracy). The Dice coefficient showed good values for segmentation of retinal pigment epithelium detachment. Further research will focus on increasing the neural network training and validation dataset and improving segmentation accuracy for other biomarkers. Keywords: age-related macular degeneration, OCT, artificial intelligence, machine learning, biomarkers, anti-VEGF therapy
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.