Performance Analysis of Diabetic Retinopathy and Hemorrhages Segmentation System Using Enhanced Visual Geometry Group (EVGG) Architecture

In this article, the retinal images are classified into either healthy or DR using the proposed EVGG deep learning architecture. This proposed method consists of Gabor transform module, data augmentation module, classification module along with hemorrhages segmentation module. The Gabor transform is used to transform the spatial pixel coordinates into multi resolution pixel coordinates. Then, the data augmentation methods are applied on the Gabor image to increase the image counts. The EVGG architecture is proposed in this work to classify the retinal image into either healthy or DR. Finally, the hemorrhages are segmented in DR retinal images. The proposed DR image classification system is applied and tested on the retinal images on DIARETDB1 and HRF datasets. The performance of the proposed hemorrhages segmentation algorithm is evaluated in terms of sensitivity, specificity and accuracy. The experimental results of the proposed method stated in this article are significantly compared with other similar works.