• Nikita Mahangade
• Sonali Bhosale
• Dhananjay Shirke
• Sakshi Ghule

This Blood vessels are one of the most important indicators in retinal image analysis, which is becoming increasingly important for diagnosing eye diseases. A comprehensive review of deep learning-based retinal blood vessel segmentation and an automated and unsupervised method for segmenting retinal blood vessels from a fundus image are presented in this paper. In modern medicine, fundus images are one of the main ways to diagnose eye diseases. The patients' health status is reflected in the geometric characteristics of the retinal vessels, which also aid in the diagnosis of certain diseases like diabetic retinopathy, glaucoma and hypertension. Patients may avoid becoming completely blind if these conditions are correctly diagnosed and treated at the appropriate times. Due to their superior efficiency and accuracy over manual segmentation and other computer-aided diagnosis methods, deep learning algorithms have recently been rapidly applied to retinal vessel segmentation. M-GAN, a brand-new conditional generative adversarial network, is proposed in this paper for the purpose of balancing losses through stacked deep fully convolutional networks for precise retinal vessel segmentation. It consists of an M-discriminator with a deeper network for more effective adversarial model training and a newly designed M-generator with deep residual blocks for more robust segmentation. To clear up the retinal vessels in the input fundus image, we use automatic color equalization (ACE) to perform pre-processing. We compared the proposed M-GAN to other studies to verify the proposed method using the DRIVE, STARE, datasets. In order to conduct a comparison, we measured the accuracy, the intersection of union (IoU), the F1 score, and the Specificity, Sensitivity. Comparative results demonstrated that the proposed M-GAN performed better than other studies. In these paper different classifier is used, the classifier in the M-GAN algorithm helps to classify the diseases like diabetic retinopathy, glaucoma and hypertension. Researchers able to build more robust and advanced models with the help of this article.

Retinal blood vessel segmentation, fundus imaging, M-generative adversarial networks (M-GAN), convolutional networks, classification

"RETINAL FUNDUS IMAGING SEGMENTATION AND CLASSIFICATION USING MODIFIED GAN ALGORITHM", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 5, page no.398-408, May-2023, Available :http://www.jetir.org/papers/JETIRFX06070.pdf

"RETINAL FUNDUS IMAGING SEGMENTATION AND CLASSIFICATION USING MODIFIED GAN ALGORITHM", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 5, page no. pp398-408, May-2023, Available at : http://www.jetir.org/papers/JETIRFX06070.pdf