• Rahul Sawhney

Automatic segmentation of blood vessels in fundus images is critical because several systemic illnesses and ocular problems may cause apparent pathologic alterations. The human eye is a very sophisticated sensory organ. Its value extends beyond the visual system since it permits non-invasive examination of anatomical features like blood vessels. The light entering a healthy eye is directed onto the retina, where chemical processes translate the light's energy into electrical messages. The maintenance of normal vision depends on the ocular vascular system's ability to adjust to the metabolic demands of the eye. The retina is one of the most metabolically active parts of the body because of the large amounts of oxygen and nutrients it requires to function. Since there are only two possible labels for each pixel, we may call this a binary classification problem (vessel or non-vessel). We use a graphic processing unit (GPU) implementation of deep max-pooling convolutional neural networks for artery and vein segmentation. We run our tests using the open-source DRIVE dataset, and the findings show that the deep learning methodology is quite successful. An AUC of 0.9753 is achieved with our approach, and its average accuracy is 0.9324.

Blood vessel segmentation, retinal imaging, deep neural networks, GPU

"Another Deep Learning Strategy for Vein Segmentation in Retinal pictures in view of Convolutional kernels", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 4, page no.b551-b564, April-2023, Available :http://www.jetir.org/papers/JETIR2304167.pdf

"Another Deep Learning Strategy for Vein Segmentation in Retinal pictures in view of Convolutional kernels", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 4, page no. ppb551-b564, April-2023, Available at : http://www.jetir.org/papers/JETIR2304167.pdf