Enhancement Method for Augmenting Blood Vessels of Retinal Images
pdf

Keywords

CLAHE
Diabetic Retinopathy
DRIVE
PSNR

How to Cite

Enhancement Method for Augmenting Blood Vessels of Retinal Images. (2024). African Journal of Biomedical Research, 27(4S), 5778-5794. https://doi.org/10.53555/mbmcqc16

Abstract

Diabetes can cause changes in the blood vessels of the retina, causing them to swell and leak fluid. Diabetic Retinopathy (DR) is a common impediment of diabetes and is the primary cause of preventable blindness. Structural and functional alterations take place in different retinal cells comprising neurons, retinal endothelial cells, and retinal pigment epithelium before clinical symptoms of DR. The appearance of microaneurysms and diabetic macular edema leads to vision loss. The research paper focuses on enhancing the retinal images to detect the occurrence of early symptoms so that the treatment advances in an accurate direction. The research paper demonstrates the enhancement of retinal images obtained from the DRIVE dataset using CLAHE and Morphological methods. The proposed method made use of a Fisher Information matrix and Kalman filter for reducing noise and blurring regions of an image. The vesselness of the retinal images obtained from the DRIVE dataset has been conducted on five scales (0.3, 0.5, 1.0, 1.5, and 2.0) to extract the blood vessels from the retinal images. 
The vesselness is conducted in both RGB and grayscale modes. The obtained ten vesseled images (five colored and five grayscale) are compared with the original input image. 
The colored vesseled images are compared with the colored input image and the grayscale vesseled images are compared with the grayscale input image. The vesseled image with the highest value of PSNR and lowest value of MSE and RMSE is selected for further processing. 
The proposed method makes use of a Histogram Equalization (HE), linear filter, and spatial filter for the removal of noise, and Gamma Transformation for enhancing the retinal images. The percentage of enhancement achieved by the proposed method as compared to CLAHE for MSE, RMSE, and PSNR are 98.23%, 86.81%, and 83.15% respectively, and against morphological operation is 94.93% for MSE, 77.67% for RMSE, and 41.68% for PSNR.

pdf
Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright (c) 2024 Vandana, Vijay Laxmi, Shalu Gupta (Author)