Intelligent Retinal Image Analysis for Predicting Cardiovascular Disease Risk Using Deep Neural Networks

Abstract

Cardio-vascular disease is the leading cause of mortality in world. The early detection of the cardio-vascular disease risk. Now days there is a technology is get introduced in the medical field called as the Retinopathy. This is used for the detecting the retinal diseases in past but by adding advancement into this we can used for Diabetes Risk prediction and know by doing advancement into this we build it for prediction of Cardiovascular Diseases Risk Prediction Using retinopathy. Retinopathy is the technique of capturing the image of the retina of the eye. Retina is the light sensitive tissues lining the back of the eye, responsible for converting light into the signals that the brain detects and process as images. There are multiple diseases present which may cause the defect in the retina of the eye. By this system we can detect several diseases which are the symptom of the cardiovascular health, or which may be affected by cardiovascular disease. Building the machine learning model for detecting the predicting the output using the various deep learning model.

Introduction

Cardiovascular disease (CVD) is the leading global cause of death, with cases rising sharply from 271 million in 1990 to 523 million in 2019. Risk factors such as hypertension, obesity, and smoking contribute significantly to this increase. Early detection and risk assessment are crucial to prevent severe outcomes.

This research explores using retinal fundus images combined with deep learning techniques to predict cardiovascular risk non-invasively. The retina’s microvascular changes often reflect systemic conditions like hypertension and heart disease, making retinal imaging a promising diagnostic tool. Convolutional Neural Networks (CNNs) and other machine learning models are applied to large datasets of retinal images and clinical data to develop accurate predictive models.

Literature highlights include:

• Use of retinal images to derive CVD risk scores (Reti-WHO), showing potential but needing further validation and refinement.

• Deep learning models achieving high accuracy (up to 98.9%) by combining retinal images with patient health data.

• CNN models integrated with MobileNet architecture for efficient and accurate classification.

• Recurrent Neural Networks (RNNs) applied to analyze sequential retinal data for heart attack risk.

• Projects demonstrating the feasibility of non-invasive retinal-based heart attack risk prediction.

Methodology involves:

• Collecting and preprocessing retinal images (resizing, normalization, noise removal, contrast enhancement, data augmentation).

• Extracting features via handcrafted methods (vessel segmentation, tortuosity, bifurcation) and deep learning.

• Using machine learning classifiers (SVM, Random Forest, XGBoost) and deep learning models (CNN) for risk prediction.

• Evaluating performance through accuracy, precision, recall, F1-score, and ROC-AUC metrics.

• The CNN-based model showed strong predictive ability and clinical applicability.

Future directions emphasize:

• Integration with electronic health records (EHR) for streamlined clinical use.

• Real-time applications optimized for edge devices to enable immediate diagnosis.

• Incorporating AI-driven assistants or chatbots for patient and doctor support.

• Developing federated learning models to enhance security, scalability, and robustness by leveraging diverse data sources.

Conclusion

The scope of this research of using the deep learning specially CNN, to predict the risk of the cardiovascular disease by analysing the retinal images. Retinal microcirculatory vessels such as microaneurysms, haemorrhages, and vascular abnormalities serve as important indicator of underlying heart-related conditions. The developed model offers external, efficient, and scalable approach for early disease detection, which could significantly improve preventive healthcare and reduce the burden of cardiovascular disease. In future integration with the federated learning can further enhance the capabilities of the model by adding the collaborative training with the institutions while preserving patient data privacy. This would lead to a more robust, accurate, and ethical complaints system suitable for real –world clinical development.

References

[1] Abdollahi, M., Jafarizadeh, A., Ghafouri?Asbagh, A., Sobhi, N., Pourmoghtader, K., Pedrammehr, S & Acharya, U. R. (2023). Artificial intelligence in assessing cardiovascular diseases and risk factors via retinal fundus images: A review of the last decade. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, e1560. [2] Vineetha, T., Reddy, D. R., Mahendra, K., & Lakshmi, B. D. (2024, May). Prediction of Cardiovascular Diseases with Retinal Images Using Deep Learning. In 2024 International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI) (pp. 1-9). IEEE. [3] Shaikh, S., Wagh, I., Zaveri, V., Mujawar, M. B., & Surve, D. (2023, December). Heart Disease Prediction Using Eye Retinal Images. In 2023 6th International Conference on Advances in Science and Technology (ICAST) (pp. 417-422). IEEE. [4] Prakash, Y. P., Siri, B., & Jabez, J. (2024, March). Identifying the Abnormalities in Retinal Images Towards the Prediction of Cardiovascular Disease Using Deep Learning. In 2024 International Conference on Wireless Communications Signal Processing and Networking (WiSPNET) (pp. 1-6). IEEE. [5] Zhang, W., Tian, Z., Song, F., Xu, P., Shi, D., & He, M. (2023). Enhancing stability in cardiovascular disease risk prediction: A deep learning approach leveraging retinal images. Informatics in Medicine Unlocked, 42, 101366. [6] Vineetha, T., Reddy, D. R., Mahendra, K., & Lakshmi, B. D. (2024, May). Prediction of Cardiovascular Diseases with Retinal Images Using Deep Learning. In 2024 International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI) (pp. 1-9). IEEE. [7] Vineetha, T., Reddy, D. R., Mahendra, K., & Lakshmi, B. D. (2024, May). Prediction of Cardiovascular Diseases with Retinal Images Using Deep Learning. In 2024 International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI) (pp. 1-9). IEEE.

Copyright

Copyright © 2025 Sushant Chavan, Siddhi Chavan, Maitrayee Divekar, Aditi Dhamale, Akshada Patil. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.