Construction and verification of retinal vessel segmentation algorithm for color fundus image under BP neural network mo
- PDF / 987,534 Bytes
- 15 Pages / 439.37 x 666.142 pts Page_size
- 113 Downloads / 193 Views
Construction and verification of retinal vessel segmentation algorithm for color fundus image under BP neural network model Siyuan Tang1 · Feifei Yu2
© Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract To improve the accuracy of retinal vessel segmentation, a retinal vessel segmentation algorithm for color fundus images based on back-propagation (BP) neural network is proposed according to the characteristics of retinal blood vessels. Four kinds of green channel image enhancement results of adaptive histogram equalization, morphological processing, Gaussian matched filtering, and Hessian matrix filtering are used to form feature vectors. The BP neural network is input to segment blood vessels. Experiments on the color fundus image libraries DRIVE and STARE show that this algorithm can obtain complete retinal blood vessel segmentation as well as connected vessel stems and terminals. When segmenting most small blood vessels, the average accuracy on the DRIVE library reaches 0.9477, and the average accuracy on the STARE library reaches 0.9498, which has a good segmentation effect. Through verification, the algorithm is feasible and effective for blood vessel segmentation of color fundus images and can detect more capillaries. Keywords BP neural network model · Color fundus image · Retinal vessel segmentation algorithm · Gaussian matched filtering · Hessian matrix
* Feifei Yu [email protected] Siyuan Tang [email protected] 1
School of Baotou Medical College, Inner Mongolia University of Science and Technology, Baotou 014060, China
2
Department of Ophthalmology, Zhuji Affiliated Hospital of Shaoxing University, Zhuji, China
13
Vol.:(0123456789)
S. Tang, F. Yu
1 Introduction The color fundus image is an image taken by the fundus camera on the inner wall of the eyeball at different angles and has the characteristics of painless, noninvasive, convenient and fast, and low imaging cost. Through it, the retinal vascular diseases can be directly observed, such as rigid exudation, bleeding points, microaneurysms, and other diseases [1]. Retinal vessel segmentation is an important basis for color fundus image analysis and has important value in the diagnosis and large-scale screening of human eye lesions. Retinal vessel segmentation is a basic step to accurately display and quantify retinopathy and is particularly important for the diagnosis, adjuvant treatment, and surgical planning of retinal diseases. Therefore, it is necessary to reduce the workload of manual operators through accurate and efficient computer technology, thereby improving accuracy, efficiency, and reproducibility. Based on the characteristics of retinal vessels in the fundus image, retinal vessel segmentation methods can be roughly divided into algorithms based on vessel tracking, matched filtering, deformation model, morphological processing, and machine learning [2, 3]. Most of the segmentation methods based on machine learning are supervised algorithms. A classifier is trained by constructing image features
Data Loading...
