眼底图像中黄斑中心与视盘自动检测新方法

A Novel Method of Macula Fovea and Optic Disk Automatic Detection for Retinal Images

在眼底图像自动分析中,视盘与黄斑的定位是糖尿病性视网膜病变计算机辅助诊断或筛查的先决条件。该文提出一种应用方向局部对比度滤波结合局部血管密度的方法,直接先行检测黄斑中心再行定位视盘,不同于现有的先行检测视盘或血管再行定位黄斑的一般方法,有效地提高黄斑定位正确率,能更好地应用于糖尿病性黄斑水肿的自动评估。实验选取了网络公开的HEI-MED数据集中169幅黄斑水肿眼底图像,黄斑和视盘的定位正确率同步达到98.2%,算法简单且无监督,优于现有的方法,具有良好的临床应用前景。

In automatic retinal image analysis, the location of the optic disk and macula fovea is the prerequisite for a system of computer aided diagnosis or automatic screening of diabetic retinopathy. This paper proposes a novel method to detect macular firstly before optic disk location or vessels detection using directional local contrast filter and local vessel density feature, which improves effectively the accuracy of the detection of the macular and optic disk, and is different from existing algorithms that detected optic disk firstly before macula fovea location. 169 color images with diabetic macular edema from HEI-MED public database are used to evaluate the performance of the developed method. From the experimental results, the accuracy rate both of macular and optic disk detection is found to be 98.2%. The proposed method is simple and unsupervised, and has high practical value.