Department of Computer Science, University of A Coruña, 15071, A Coruña, Spain.
Instituto Oftalmológico Victoria de Rojas, A Coruña, Spain.
Med Biol Eng Comput. 2017 Dec;55(12):2209-2225. doi: 10.1007/s11517-017-1660-8. Epub 2017 Jun 17.
Retinal vessel tree extraction is a crucial step for analyzing the microcirculation, a frequently needed process in the study of relevant diseases. To date, this has normally been done by using 2D image capture paradigms, offering a restricted visualization of the real layout of the retinal vasculature. In this work, we propose a new approach that automatically segments and reconstructs the 3D retinal vessel tree by combining near-infrared reflectance retinography information with Optical Coherence Tomography (OCT) sections. Our proposal identifies the vessels, estimates their calibers, and obtains the depth at all the positions of the entire vessel tree, thereby enabling the reconstruction of the 3D layout of the complete arteriovenous tree for subsequent analysis. The method was tested using 991 OCT images combined with their corresponding near-infrared reflectance retinography. The different stages of the methodology were validated using the opinion of an expert as a reference. The tests offered accurate results, showing coherent reconstructions of the 3D vasculature that can be analyzed in the diagnosis of relevant diseases affecting the retinal microcirculation, such as hypertension or diabetes, among others.
视网膜血管树提取是分析微循环的关键步骤，这是相关疾病研究中经常需要的过程。迄今为止，这通常是通过使用 2D 图像采集范式来完成的，这种方法只能有限地可视化视网膜血管的真实布局。在这项工作中，我们提出了一种新的方法，通过结合近红外反射视网膜摄影信息和光学相干断层扫描 (OCT) 切片，自动分割和重建 3D 视网膜血管树。我们的方法可以识别血管、估计其口径，并获得整个血管树所有位置的深度，从而能够重建完整动静脉树的 3D 布局，以便进行后续分析。该方法使用 991 个 OCT 图像和相应的近红外反射视网膜摄影进行了测试。使用专家的意见作为参考，对该方法的不同阶段进行了验证。测试结果准确，显示出可以在诊断高血压或糖尿病等影响视网膜微循环的相关疾病时进行分析的 3D 血管结构的连贯重建。
