糖尿病视网膜病变自动检测研究

Automatic Detection of Diabetic Retinopathy

针对国内糖尿病视网膜病变自动检测中硬性渗出检测速度和精度较低的问题,提出在眼底视网膜图像的绿色通道提取硬性渗出候选区域,对硬性渗出与软性渗出的特征进行描述,设计合适的分类器识别硬性渗出,通过硬性渗出的量化评价指标和糖尿病视网膜病变标准数据库对比,验证硬性渗出识别的有效性。通过该项目研究,将提高硬性渗出的诊断效率和精确度,从而使大批量糖尿病的普查及跟踪治疗成为可能。

Aiming at the low speed and accuracy of hard exudation detection in the automatic detection of diabetic retinopathy in China,a method of extracting hard exudation candidate region in the green channel of retinal image is proposed.Describe the characteristics of hard exudation and soft exudation,design a suitable classifier to identify hard exudation.Through the comparison of the quantitative evaluation index of hard exudation and the standard database of diabetic retinopathy,verify the effectiveness of identification of hard seepage.Through the research of this project,the diagnostic efficiency and accuracy of hard exudation will be improved,so that a large number of diabetes screening and follow-up treatment will be possible.