Retinal Blood Vessel Segmentation by Using Matched Filtering and Fuzzy C-means Clustering with Integrated Level Set Method for Diabetic Retinopathy Assessment

Abstract

Background The condition of blood vessel network in the retina is an essential part of diagnosing various problems associated with eyes, such as diabetic retinopathy. Methods In this study, an automatic retinal vessel segmentation utilising fuzzy c-means clustering and level sets is proposed. Retinal images are contrast-enhanced utilising contrast limited adaptive histogram equalisation while the noise is reduced by using mathematical morphology followed by matched filtering steps that use Gabor and Frangi filters to enhance the blood vessel network prior to clustering. A genetic algorithm enhanced spatial fuzzy c-means method is then utilised for extracting an initial blood vessel network, with the segmentation further refined by using an integrated level set approach. Results The proposed method is validated by using publicly accessible digital retinal images for vessel extraction, structured analysis of the retina and Child Heart and Health Study in England (CHASE_DB1) datasets. These datasets are commonly used for benchmarking the accuracy of retinal vessel segmentation methods where it was shown to achieve a mean accuracy of 0.961, 0.951 and 0.939, respectively. Conclusion The proposed segmentation method was able to achieve comparable accuracy to other methods while being very close to the manual segmentation provided by the second observer in all datasets.

Artificial Intelligence in Ophthalmology: A Meta-Analysis of Deep Learning Models for Retinal Vessels Segmentation

“…Among various features in digital fundus images, retinal blood vessels provide useful information that is an important prerequisite for a number of clinical applications [8]. However, manual segmentation of retinal vessels by a trained human expert is time-consuming and highly variable [9,10]. The lack of human observers, infrastructure, and awareness are key challenges that need to be overcome.…”

Retinal vessel segmentation with constrained-based nonnegative matrix factorization and 3D modified Attention U-Net

“…Unsupervised algorithms are designed according to inherent features of the retinal vessels without relying on artificial labeled images. Some recent proposed unsupervised approaches can be roughly divided into matching filter methods [2], vascular tracing methods [3], level set methods [4], model-based method [5], hierarchical image matting model [6], etc. Generally, although unsupervised algorithms improve segmentation performance, thin vessels which affect the whole performance considerably is difficult to be detected [7].…”

“…To test the effectiveness of the proposed framework, we compared the output of our approach with several advanced algorithms U-Net [3] [1], AG-UNet [4] [34], IterNet [5] [26], DenseNet [6] [35], V-GAN [7] [36] on STARE and DRIVE. Their segmentation results are obtained by running publicly available codes.…”

An optimal solution based on image fusion for blood vessel diameter measurements in cerebral angiographies

“…where are parameters used to control sensitivity and λ 1 , λ 2, λ 3 represent the eigenvalues of the Hessian matrix. In the second step, a comparative study between blob structures on one side and tube and plate structures is made using the following three measures [16,17]:…”