Title: Automated extraction of blood vessels in retinal images using rolling guidance filter

Authors: Deepak Kumar Maharana; Pranati Das

Addresses: Department of Electrical Engineering, IGIT Sarang, Dhenkanal, Odisha, 759146, India ' Department of Electrical Engineering, IGIT Sarang, Dhenkanal, Odisha, 759146, India

Abstract: This paper represents an unsupervised approach for the segmentation of the retinal vasculatures in an instinctive way. In this approach, first, the ophthalmoscope images are enhanced by contrast stretching along with logarithmic transformation and contrast limited adaptive histogram equalisation (CLAHE). Next, the edges of the enhanced images are preserved by using a rolling guidance filter, and, subsequently, the residual image is obtained by the subtraction of the rolling guidance image from the CLAHE image. After that, the optic disk is excluded to give a better segmentation result. Next, the ISODATA thresholding method is applied to obtain the binary image. Finally, morphological operation with noise removal and the boundary is excluded to obtain the final segmented image. The presented approach's performance is appraised using different mathematical measures. It achieves average false positive rate (FPR), true positive rate (TPR) and accuracy of 0.0152, 0.7037 and 0.9604 for DRIVE database and 0.0313, 0.776 and 0.9541 for STARE database.

Keywords: ophthalmoscope images; CLAHE; rolling guidance filter; RGF; false positive rate; FPR; true positive rate; TPR.

DOI: 10.1504/IJBET.2024.138614

International Journal of Biomedical Engineering and Technology, 2024 Vol.45 No.1, pp.45 - 64

Received: 12 May 2022
Accepted: 23 May 2023
Published online: 16 May 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article