RESUMEN
Objective.This study aims to automate the segmentation of retinal arterioles and venules (A/V) from digital fundus images (DFI), as changes in the spatial distribution of retinal microvasculature are indicative of cardiovascular diseases, positioning the eyes as windows to cardiovascular health.Approach.We utilized active learning to create a new DFI dataset with 240 crowd-sourced manual A/V segmentations performed by 15 medical students and reviewed by an ophthalmologist. We then developed LUNet, a novel deep learning architecture optimized for high-resolution A/V segmentation. The LUNet model features a double dilated convolutional block to widen the receptive field and reduce parameter count, alongside a high-resolution tail to refine segmentation details. A custom loss function was designed to prioritize the continuity of blood vessel segmentation.Main Results.LUNet significantly outperformed three benchmark A/V segmentation algorithms both on a local test set and on four external test sets that simulated variations in ethnicity, comorbidities and annotators.Significance.The release of the new datasets and the LUNet model (www.aimlab-technion.com/lirot-ai) provides a valuable resource for the advancement of retinal microvasculature analysis. The improvements in A/V segmentation accuracy highlight LUNet's potential as a robust tool for diagnosing and understanding cardiovascular diseases through retinal imaging.
Asunto(s)
Aprendizaje Profundo , Fondo de Ojo , Procesamiento de Imagen Asistido por Computador , Humanos , Vénulas/diagnóstico por imagen , Vénulas/anatomía & histología , Procesamiento de Imagen Asistido por Computador/métodos , Arteriolas/diagnóstico por imagen , Arteriolas/anatomía & histología , Vasos Retinianos/diagnóstico por imagenRESUMEN
Eye development and function rely on precise establishment, regression and maintenance of its many sub-vasculatures. These crucial vascular properties have been extensively investigated in eye development and disease utilizing genetic and experimental mouse models. However, due to technical limitations, individual studies have often restricted their focus to one specific sub-vasculature. Here, we apply a workflow that allows for visualization of complete vasculatures of mouse eyes of various developmental stages. Through tissue depigmentation, immunostaining, clearing and light-sheet fluorescence microscopy (LSFM) entire vasculatures of the retina, vitreous (hyaloids) and uvea were simultaneously imaged at high resolution. In silico dissection provided detailed information on their 3D architecture and interconnections. By this method we describe successive remodeling of the postnatal iris vasculature, involving sprouting and pruning, following its disconnection from the embryonic feeding hyaloid vasculature. In addition, we demonstrate examples of conventional and LSFM-mediated analysis of choroidal neovascularization after laser-induced wounding, showing added value of the presented workflow in analysis of modelled eye disease. These advancements in visualization and analysis of the respective eye vasculatures in development and complex eye disease open for novel observations of their functional interplay at a whole-organ level.