基于图像处理的青光眼分类研究

Research on Glaucoma Classification Based on Image Processing

青光眼致盲率高,检测难度大,视盘及视杯检测对青光眼早期诊断至为关键,为提高检测效率,提出一种改进的青光眼分类方法。方法通过掩膜闭合操作分割视盘,再将超像素分割与阈值相结合分割视杯,从中提取杯盘比特征来.对青光眼进行分类。详细介绍视盘区域的确定过程,以及如何通过阈值和椭圆拟合得到的视杯候选区域。在实验中通过REFUGE数据库进行测试,获得最终分类结果,识别准确率可达83.64%。该分类方法具有较高精度,在同类研究当中具有一定的竞争优势。

The blindness rate of glaucoma is high, and the detection is difficult. The detection of optic disc and cup is the key to the early diagnosis of glaucoma. In order to improve the detection efficiency,an improved classification method of glaucoma is proposed. In the method, the optic disc is segmented by mask closing operation, and then the superpixel segmentation and threshold segmentation are combined to segment the optic cup, from which the feature of cup-to-disc ratio is extracted to classify glaucoma. The determination process of optic disc area and how to get the candidate area of optic cup by threshold and ellipse fitting are introduced in detail. In the experiment, the final classification result is obtained through the test of REFUGE database, and the recognition accuracy rate can reach 83.64%. The classification method has high accuracy and has certain competitive advantages in similar research.