University of Shiga Prefecture, Hikone-city, Japan.
Adv Exp Med Biol. 2020;1213:107-120. doi: 10.1007/978-3-030-33128-3_7.
At medical checkups or mass screenings, the fundus examination is effective for early detection of systemic hypertension, arteriosclerosis, diabetic retinopathy, etc. In most cases, ophthalmologists and physicians grade retinal images by the condition of the blood vessels, lesions. However, human observation does not provide quantitative results, thus blood vessel analysis is an important process in determining hypertension and arteriosclerosis, quantitatively. This chapter describes the latest automated blood vessel extraction using the deep convolution neural network (DCNN). Diabetic retinopathy is a common cardiovascular disease and a major factor in blindness. Therefore, early detection of diabetic retinopathy is very important to preventing blindness. A microaneurysm is an initial sign of diabetic retinopathy, and much research has been conducted for microaneurysm detection. This chapter also describes diabetic retinopathy detection and automated microaneurysm detection using the DCNN.
在体检或大规模筛查中，眼底检查对于早期发现系统性高血压、动脉硬化、糖尿病性视网膜病变等疾病非常有效。在大多数情况下，眼科医生和内科医生通过血管状况、病变程度来对视网膜图像进行分级。然而，人的观察并不能提供定量结果，因此血管分析是确定高血压和动脉硬化的一个重要过程，这是定量的。本章描述了使用深度卷积神经网络（DCNN）进行最新的自动血管提取。糖尿病性视网膜病变是一种常见的心血管疾病，也是失明的一个主要因素。因此，早期发现糖尿病性视网膜病变对于预防失明非常重要。微动脉瘤是糖尿病性视网膜病变的初始迹象，已经有很多研究致力于微动脉瘤的检测。本章还描述了使用 DCNN 进行糖尿病性视网膜病变检测和自动微动脉瘤检测。
