Neural Networks Application for Accurate Retina Vessel Segmentation from OCT Fundus Reconstruction

Abstract

The use of neural networks for retinal vessel segmentation has gained significant attention in recent years. Most of the research related to the segmentation of retinal blood vessels is based on fundus images. In this study, we examine five neural network architectures to accurately segment vessels in fundus images reconstructed from 3D OCT scan data. OCT-based fundus reconstructions are of much lower quality compared to color fundus photographs due to noise and lower and disproportionate resolutions. The fundus image reconstruction process was performed based on the segmentation of the retinal layers in B-scans. Three reconstruction variants were proposed, which were then used in the process of detecting blood vessels using neural networks. We evaluated performance using a custom dataset of 24 3D OCT scans (with manual annotations performed by an ophthalmologist) using 6-fold cross-validation and demonstrated segmentation accuracy up to 98%. Our results indicate that the use of neural networks is a promising approach to segmenting the retinal vessel from a properly reconstructed fundus.

Advancing Ocular Imaging: A Hybrid Attention Mechanism-Based U-Net Model for Precise Segmentation of Sub-Retinal Layers in OCT Images

“…Recent advancements in network-based methodologies for optical coherence tomography angiography (OCTA) segmentation address challenges in retinal vascular structure delineation, particularly under low-light conditions, and offer the potential for improved disease diagnosis, such as branch vein occlusion (BVO) [ 29 ]. One study [ 30 ] explored the application of five neural network architectures to accurately segment retinal vessels in fundus images reconstructed from 3D OCT scan data, achieving up to 98% segmentation accuracy, thus demonstrating the promise of neural networks in this domain. Viedma et al [ 31 ] evaluates Mask R-CNN for retinal OCT image segmentation, showcasing its comparable performance to U-Net with lower boundary errors and faster inference times, offering a promising alternative for efficient automatic analysis in research and clinical applications.…”

Electronic anti-theft protection for vehicles of people with special needs

“…Researchers have begun to deal with various aspects of its development and are finding more and more innovative ways to use it. It is intensively developed in the field of medicine, where biometric systems combined with vision algorithms allow for the automatic identification of pathological lesions in the retina of the human eye (Marciniak et al, 2023), X-ray images of the lungs (Natashia, 2020), the recognition of cancer lesions in computed tomography images (Ma et al, 2020) and recognizing driver fatigue (Poliak et al, 2023). Application solutions dedicated to mobile phones are also available, allowing for deep learning of characteristic vein patterns from a 2D image obtained using a smartphone (Garcia-Martin and Sanchez-Reillo, 2021).…”