Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{171896,
        author = {Varshitha S and Yashaswini H U and Kshema R Gowda and Impana S M and Madhu C K},
        title = {Segmentation of Retinal Layers in OCT for Identification of Neural Disorders by Machine Learning Models},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {8},
        pages = {1411-1417},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=171896},
        abstract = {The segmentation of the retinal layers in OCT images is a very important aspect in diagnosing neural disorders, such as glaucoma, Alzheimer's disease, and multiple sclerosis. Early detection of these disorders affects the treatment outcomes and quality of life of patients. In principle, traditional clinical observation-based, manual analysis-of-OCT-based diagnostic methods have always been associated with time and being subjective, causing inconsistencies. However, this research introduces a computer vision-based framework for the machine learning-based automation of retinal layer segmentation on OCT images as an early screening tool for neural disorders. To improve the image quality, an advanced preprocessing combination of noise reduction, image normalization, and contrast enhancement Optical Coherence Tomography Angiography is an advanced, non-invasive imaging technology. It has allowed the ability to visualize, with high resolution, the retinal blood vessels at the level of capillaries. The automated process of segmenting vessels within OCTA images has always been challenging because of problems such as poor visibility of capillaries and the intricate structure of retinal vessels. The ROSE dataset was prepared using 229 OCTA images annotated at centerline and pixel level. A new vessel segmentation network called OCTA-Net is proposed for OCTA images. In OCTA-Net, the two-step segmentation approach is used: coarse segmentation module to generate an initial vessel confidence map and a refined segmentation module that fine-tunes the vessel boundaries for accurate detection of both thick and thin vessels. The performance of the vessel segmentation task is tested aggressively against state-of-the-art vessel segmentation techniques on the ROSE dataset. Further analysis reveals that the fractal dimension of the segmented retinal vasculature present differences in healthy and AD patients, thus showing the possibility of analysis of retinal vasculature for neurodegenerative diseases.},
        keywords = {},
        month = {January},
        }