基于改进小波边缘检测算法的视网膜血管宽度测量

Improved wavelet edge detection algorithm-based measurement of retinal vascular width

背景:眼底荧光血管造影可反映视网膜血管结构、血流动力学改变、血管病理生理变化及其相关结构的病理改变,广泛应用于视网膜、脉络膜及视神经疾病的鉴别诊断。目的:通过分析视网膜血管的特点和小波边缘检测算法的优缺点,对小波边缘检测算法进行改进,并利用该算法对视网膜血管进行边缘检测与宽度测量。方法:首先由样条函数构造样条小波,由此得到小波滤波系数。根据改进的小波边缘检测方法得到眼底荧光图像的血管初始边缘,利用边缘细化算法对初始边缘进行细化,基于分形技术对细化后的边缘进行连接,由噪声去除算法消除边缘图像中的噪声点,得到连续的单象素点血管边缘。通过两条边缘之间垂直线上的象素点数,得到血管的实际内径。结果与结论:文章使用的方法较好的解决了传统方法边缘细节和噪声太多的问题,具有更好的边缘连续性和更少的过检测点。将其应用于视网膜血管宽度的测量之中,诊断结果与实际情况较为接近,可以为眼科医生的临床诊断提供较大帮助。

BACKGROUND：Fundus fluoresce in angiography can reflect the structure of retinal blood vessels,hemodynamic changes,disease management and related physiological changes and the structure of pathological changes,and has been widely used in the retina,choroid and optic nerve disease in the differential diagnosis. OBJECTIVE：Through the analysis of the characteristics of retinal vessels and advantages and disadvantages of wavelet edge detection algorithm,to improve wavelet edge detection algorithm and to detect edge and measure width of retinal vessels. METHODS：By the spline function structure transect wavelet,the wavelet filter ratio was obtained. Using the improved wavelet edge detection method,blood vessels initial edge of fundus fluorescence images was obtained. The edges of the initial margin refinement algorithm were detailed,and connected based on the fractal technology after the refinement edge. The noise reduction algorithm was used to eliminate the noise in the edge image,and pixel by a row of single-vessel edge was obtained. Using edge by two points between the vertical line of pixels,the actual diameter of blood vessels was obtained. RESULTS AND CONCLUSION：This method well solved conventional routes edge detail and noise,with better continuity and less edge detection points. This method applied in the retina blood vessel width survey obtained similar diagnosis result to actual situation,which may help ophthalmologist＇s clinical diagnosis.