Fractional chef based optimization algorithm trained deep learning for cardiovascular risk prediction using retinal fundus images

• Published in: Biomedical signal processing and control 2024-08, Vol.94, p.106269, Article 106269
• Main Authors: Rajesh Kumar, T., Enireddy, Vamsidhar, Kalai Selvi, K, Shahid, Mohammad, Vijendra Babu, D, Sudha, I
• Format: Article
• Language: English
• Subjects: Blood cell segmentation; Cardiovascular disease; Chef based Optimization; Deep Learning; Spatial Attention U-Net
• ISSN: 1746-8094
• DOI: 10.1016/j.bspc.2024.106269

•The highlights of the article are given below for your kind perusal. Kindly, consider and forward my article for further processes.•This research work is focused on DL enabled CVD prediction and classification.•Retinal fundus images are gathered from the dataset.•Grey scale conversion is carried out in pre-processing stage.•Pre-processed images are fed to optic disc detection by FCBOA enabled Psi-net. The modernhealthcaresystem has been linked to quicker growth as well as the capacity to transform themedical data for predicting thesignificant healthcare policy thatfacilitate timely preventive health services. Cardiovascular disease (CVD) is the key source of disability and death of developing country. In the modern global climate, identifying the Cardiovascular (CV) based on initial signs is extremely difficult. The CV riskprediction using DL method is employed in this work. Here,retinal fundus image preprocessing is the initial process, in which grey color conversion technique is utilized. Following this,optic disc is detected with the aid of Psi-Net and the proposed Fractional Chef based Optimization (FCBOA) is used in training process of Psi-Net. Afterwards, blood vessel segmentation is accomplished using the FCBOA enabled Spatial Attention U-Net (SA-UNet). Moreover, output image of segmentation and optic disc detection are fedto feature extraction. Furthermore, the texture features are extracted from input image. Moreover, these two classes of feature extracted images are applied to the CV risk prediction system, where the FCBOS-based SpinalNet (FCBOA-SpinalNet) is utilized for categorizing the image as normal or hypertensive type. The CV risk prediction is evaluated with regards to three metrics includes accuracy, sensitivity, and specificity, which offer thefinest values of 0.913, 0.917, and 0.918.