Osprey Gannet optimization enabled CNN based Transfer learning for optic disc detection and cardiovascular risk prediction using retinal fundus images

The identification of retinal vascular features that have been demonstrated to predict cardiovascular (CV) risk has been made possible by recent developments in retinal imaging modality, and the detection of the optic disc (OD) is a key step in the development of automated screening systems for diabetic retinopathy (DR). However, there are some risk factors, which cannot be detected in both cardiovascular and OD. To bridge this gap, Transfer learning (TL) based Osprey Gannet optimization (OGO) utilizing retinal fundus image is established. Initially, the input fundus image is allowed into the preprocessing process utilizing a bilateral filter. Thereafter, the OD detection is done using the devised Osprey Gannet-active counter model, which is trained using OGO. Here, OGO is developed by the integration of Osprey Optimization (OO) and Gannet Optimization (GO). Simultaneously, the pre-processed image is passed to the blood vessel segmentation (BVS) and is performed by Res-UNet, trained by the Jaya Chronological Chef-Based Optimization Algorithm (Jaya-CCBOA). Hereafter, the feature extraction is carried out from the OD detection phase and BVS. On the other hand, the input image is also extracted using various feature extractors. Thereafter, feature selection is done using Tanimoto similarity. Finally, the detection of CV risk as normal and hypertensive is accomplished by the Convolutional neural network (CNN) based TL, trained by OGO. The measures employed in OGO_CNN based TL achieved maximum values of 92.1%, 91.5%, 93.1%, 87.9%, and 87.9%.