Cardiac Issues Viewed Through the Eyes: Ocular Clues, Retinal Imaging, and AI-Based Cardiovascular Screening

Abstract

Cardiovascular disease remains the leading cause of death worldwide; however many screening pathways rely on measurement in clinics, laboratory analysis, or advanced imaging, which is not always available to everyone. The eye provides a unique non-invasive gateway to the vascular system as its microvasculature is visible during routine examinations. Classical eye signs of hypertension (hypertensive retinopathy), stroke (retinal artery occlusion), anemia (conjunctival pallor), high cholesterol levels (corneal arcus and xanthelasma), infective endocarditis (Roth spots), and systemic inflammation can indicate vascular, embolic, metabolic, inflammatory, or medication-related disorders. Recent developments in technology, such as digital fundus photography, optical coherence tomography, optical coherence tomography angiography, wearable imaging devices, smart glasses, and artificial intelligence, allow for the potential to transform these signs into quantitative and predictive biomarkers. This paper summarizes the clinical and technological rationale behind the use of ocular signs in cardiovascular evaluation, lists eye-heart relationships in a clinical table, and presents an interdisciplinary approach to this field for future scoping review and translational research.

Keywords:Cardiovascular disease, Retina, Hypertensive retinopathy, Oculomics, OCTA, Artificial intelligence, Retinal vascular biomarkers, Cardiovascular screening

Introduction

Despite major advances in prevention and treatment, CVD remains the primary global cause of death, claiming more than 20 million lives per year [1]. The current practice of cardiovascular risk screening utilizes measurements of blood pressure, fasting lipid profile, blood glucose, body mass index/waist circumference, and cardiovascular risk score. If indicated in selected high-risk groups, coronary artery calcium scoring, carotid ultrasound, electrocar diography, echocardiography, or vascular imaging may be used to detect subclinical atherosclerotic disease or to define the risk more accurately [2]. These methods have been proven to be valuable; however often times these procedures and tests are inaccessible or difficult to implement in clinical practice. The uniqueness of the eye in cardiovascular medicine lies in its being the only accessible organ in which small blood vessels can be visualized non-invasively. The remodeling of retinal arterioles, venules, capillaries, and choroid, endothelial dysfunction, inflammation, thrombosis, metabolic injury, and senescence can all be detected using ophthalmic methods. All these features are characteristic. of coronary artery disease, cerebrovascular disease, hypertension, diabetic vascular disease, and systemic inflammatory conditions [3,4].

Historically, signs of systemic disease were discovered in the eye through bedside exam, fundoscopic exam, slit lamp exam, and identification of typical manifestations such as Roth spots, hypertensive retinopathy, retinal emboli, retinal vessel occlusion, corneal arcus, and xanthelasma. However, modern times have brought about new opportunities through the implementation of digital imaging, Optical Coherence Tomography (OCT), OCT angiography, and AI, which created an area known as oculomics or the application of data from ocular images to identify systemic disease state [5]. This paper suggests that the eye should not be considered as a substitute for conventional cardiovascular tests but as an additional tool for cardiovascular screening, triaging, and studies. The most significant clinical value is represented by the ability to find out whether findings in the eye should prompt cardiovascular investigation and development of imaging biomarkers.

Why the Eye Reflects the Cardiovascular System

The retina and optic nerve require a constant supply of oxygen and are very sensitive to vascular insults. Retinal blood vessels are tiny, clear structures against the background of the retina, making it possible for physicians and imaging devices to visualize vascular size, tortuosity, bleeding, exudation, emboli, ischemia, and capillary non-perfusion. Such findings can be related to eye disorders; however, they can also indicate vascular disease elsewhere in the body. There are multiple explanations for the link between eye and heart. Hypertension can result in arteriolar narrowing, arteriovenous nicking, cotton wool spots, flame shaped hemorrhages, hard exudates, and swelling of the optic disc. Atherosclerotic and embolic diseases lead to visible emboli in the retina and occlusion of retinal arteries. Diabetes and metabolic diseases lead to microaneurysms, bleeding, exudates, and macular edema. Vasculitis can affect the retina and lead to ischemia or uveitis while putting the heart at risk. Drug toxicity may also create parallel ocular and cardiac effects, as seen with hydroxychloroquine-related retinopathy and cardiopathy risk [6]. Due to the fact that many of these findings are not disease-specific, interpretation must remain clinical-context dependent. A Roth spot alone does not diagnose infective endocarditis; however, Roth spots in a patient with fever, murmur, sepsis, or embolic phenomena may support urgent evaluation [7,8]. Similarly, corneal arcus in an older adult may be age-related, whereas premature arcus in a younger person should prompt lipid evaluation [9].

Clinical Categories of Eye-Heart Associations

Clinical findings related to cardiovascular disease can be classified under three different groups. The first group comprises directly clinically significant findings. These findings are those that may give a direct indication of underlying systemic disease, such as retinal artery occlusion as an indication of embolic disease, conjunctival petechiae following thoracic trauma, and ectopia lentis as a finding in Marfan syndrome or homocystinuria. Such clinical findings can be important as they might be associated with an increased risk for stroke, aortic disease, cardiac tumors, thromboembolism, and significant vascular inflammation. Indirect vascular biomarkers do not provide the diagnosis of a specific heart disease, but they may show the total burden of the vasculature in the body. Indirect vascular biomarkers include retinal vessel caliber, venular widening, arteriolar narrowing, tortuosity, branching changes, diabetic retinopathy, and retinal vein occlusion. Such biomarkers may be indicative of hypertension, diabetes mellitus, vascular aging, endothelial dysfunction, hyperlipidemia, and inflammation [3,4]. Computationally derived predictive markers are the third category of eye-heart associations. These markers emerge when AI systems analyze fundus photography, OCT, or OCTA images. Computationally derived markers detect subtle changes in blood vessels, retinal structure, and textural changes that cannot be detected without AI systems. Predictive markers may be used to estimate cardiovascular risk, biological age, hypertension, stroke risk, and cardiometabolic burden [3,5]. Robust validation is required before widespread clinical decision-making parameters are recommended.

Technology Expanding the Eye-Heart Window

Fundus photography is a practical starting point for the development of ocular cardiovascular screening because it is quick, minimally invasive, and scalable. It captures retinal vessel caliber, emboli, hemorrhages, exudates, microaneurysms, changes at the optic disc, and many other abnormalities. In combination with cloud-based assessment or AI triaging, fundus photography could become an effective tool for screening in primary care, pharmacy practice, mobile units, community outreach, and tele-ophthalmology. Optical Coherence Tomography (OCT) offers cross-sectional imaging of retinal structure and optic nerve, and OCT angiography allows for noninvasive visualization of retinal and choroidal microcirculation. OCTA can measure vessel density, perfusion indices, capillary dropout, and microvascular rarefaction – all of which are important features in systemic microvascular dysfunction, hypertension, diabetes, atherosclerosis, and stroke [3]. AI combined with machine learning may have the ability to interpret ocular images into a cardiovascular risk predictor by discovering higher-dimensional characteristics of the ocular structures that can be routinely assessed clinically. However an AI-based solution needs to be critically assessed from the perspective of dataset diversity, calibration, explainability, fairness, generalizability, and clinical relevance. An algorithm that performs well with one individual or a certain demographic does not necessarily perform well in other demographics in terms of age, ethnicity, imaging device, and health system. Thus, AI ocular screening would best serve as an assistive technology, not a diagnostic shortcut without proper evaluation. Wearable imaging and smart glasses-based systems are promising engineering achievements that could revolutionize the field in the future. However, their effectiveness in clinical practice will depend on the quality of images they generate, users’ training, interoperability, privacy protection, regulatory requirements, and evidence base for better outcomes, not just data generation An algorithm that performs well with one individual or a certain demographic does not necessarily perform well in other demographics in terms of age, ethnicity, imaging device, and health system..

Integrated Clinical Table

Cardiac and cardiovascular diseases that may be suggested, risk stratified or even aided by ocular manifestations are summarized in Table 1. For better clinical readability, the following table is rearranged according to main diagnostic groups such as trauma, infection/inflammation, tumor/neoplasm, embolic disease, vascular markers, inherited/genetic, metabolic/infiltrative, and drug toxicity.

The table combines direct clinical signs, indirect vascular biomarkers, and computationally determined ocular signs pertinent to cardiovascular diseases. The numbering of references has been revised to correspond to the new classification order without altering the underlying evidence basis (Table 1).

Table 1:Cardiac and cardiovascular diseases viewed through ocular findings.

Proposed Scoping Review Framework

An adequate review of an ocular-cardiovascular assessment topic requires both clinical and engineering databases to be reviewed. Epidemiology, diagnosis, clinical validation, patient outcomes, and guideline-based information may be found using PubMed. Algorithm development, device architectures, image processing, wearable devices, smart glass technology, signal processing, sensor integration, and preliminary technological proofs-ofconcept may be found using IEEE Xplore. Relying on one ecosystem increases the chances of neglecting either clinical or technological relevance. A simple yet useful approach for the scoping review might consist of three interconnected questions: (1) which cardiovascular conditions or states carry ocular manifestations; (2) which ocular imaging technologies have been used to detect or quantify such manifestations; (3) how publication distribution is different between biomedical and engineering databases. The above design will provide us with an opportunity to reflect on clinical evidence, technical innovation, and translational gaps. Possible search concepts might be the following: cardiovascular condition, retinal imaging, OCTA, fundus imaging, oculomics, hypertensive retinopathy, retinal vascular biomarkers, artificial intelligence, machine learning, smart glasses, wearables, screening, risk prediction, validation. Results should be structured according to disease, imaging technology, study type, population, algorithm, validation, and clinical applicability.

Translational and Implementation Considerations

Ocular-cardiovascular screening would only be clinically applicable if the current research progresses past the demonstration of accuracy as a proof of concept. Implementation involves clear referral guidelines, integration with electronic health records, image acquisition training, quality control measures, confidentiality, regulation, and physician buy-in. There is no benefit to the screening test unless a subsequent action follows the test results, including blood pressure monitoring, cholesterol level measurement, diabetes investigation, stroke screening, echocardiography, consultation with a cardiologist, or even emergency care [31]. Equity is also an important factor in this situation. Training algorithms on limited sample sizes could produce worse results in populations that vary from the one used to train the model in terms of ethnicity, age, sex, socioeconomic status, imaging equipment, or prevalence of certain diseases. Consequently, external validation and performance in subgroups are crucial before implementation. While community screening for ocular pathology might promote equity by improving access to the test, it would only be effective if the process itself doesn’t increase health inequalities. Clinical language needs to be used carefully as well. Ocular imaging might help in cardiovascular risk stratification; however, presenting it as a diagnostic method of coronary artery disease, heart failure, arrhythmia, or systemic vasculitis is inappropriate.

Limitations of the Eye-Based Approach

The ocular manifestations themselves are usually non-specific. For example, retinal hemorrhages, cotton-wool spots, vessel tortuosity, and vascular occlusion may be caused by various factors, including hypertension, diabetes, anemia, inflammation, infection, thrombosis, injury, or a disease of the eye itself. It significantly restricts the diagnostic potential of the findings in the eyes. Imaging techniques are also limited in some technical aspects. Poor image quality may be due to cataracts, opacity of the media, inadequate fixation, small pupils, motion artifacts, or an inexperienced technician. OCTA results will depend on the device used, scanning protocol, segmentation, and processing software. AI technologies may be influenced by dataset biases, device shifts, and lack of transparency of the algorithms. In addition, one should avoid over testing. There has to be a reasonable approach to each screening tool, based on a certain signal, leading to further stepwise evaluation of a patient. If not handled appropriately, it may result in unnecessary stress, financial costs, and invasive procedures without any beneficial effect on health or prevention of disease.

Future Directions

The research to come should focus on prospective studies investigating whether ocular imaging affects cardiovascular health outcomes, not simply whether it can identify risk. Research should investigate the added value of artificial intelligence-enhanced ocular screening relative to existing risk calculators, cost-effectiveness, and whether incorporating ocular biomarkers alters management decisions. Diverse multi-center cohorts will be a requirement in order to obtain generalizable data. Standardization of imaging protocols and best practices for reporting will aid in improving reproducibility. Combining retinal imaging with blood pressure, lipid profiles, glucose testing, electrocardiograms, wearables, and medical histories leads to better risk prediction models than those that use ocular images alone. Engineering breakthroughs must be paired with clinical validity from the start. Technologies and algorithms need to be designed for practical use in the clinic with safety, interoperability, and actionability in mind. Hybrid devices will likely prove most effective in the future.

Conclusion

The eye is an important window for assessing cardiovascular health. Conventional ocular signs may indicate critical medical conditions; meanwhile, the eye itself, along with retinal imaging, OCTA, and AI, increases the potential for quantifying microvascular injury and risk assessment of cardiovascular disease.

In the interest of science, it would be more appropriate to say that the eye cannot take the place of cardiovascular tests, but it can provide us with useful clues, biomarkers, and triage signals. For a proper article on this issue, a combination of clinical ophthalmology, cardiology, vascular biology, biomedical engineering, and AI would be necessary.

Competing of Interest

None.

Acknowledgements

None.

References

1. Di Cesare M, Perel P, Taylor S, Kabudula C, Bixby H, et al. (2024) The Heart of the World. Glob Heart 19(1): 11.
2. Jonas DE, Reddy S, Middleton JC (2018) Screening for Cardiovascular Disease Risk with Electrocardiography: An Evidence Reviewer the U.S. Preventive Services Task Force. Rockville (MD): Agency for Healthcare Research and Quality (US) Evidence Synthesis No 163.
3. Ghenciu LA, Dima M, Stoicescu ER, Iacob R, Boru C, et al. (2024) Retinal Imaging-Based Oculomics: Artificial Intelligence as a Tool in the Diagnosis of Cardiovascular and Metabolic Diseases. Biomedicines 12(9): 2150.
4. Moss HE (2015) Retinal Vascular Changes are a Marker for Cerebral Vascular Diseases. Curr Neurol Neurosci Rep 15(7): 40.
5. Honavar SG (2022) Oculomics - The eyes talk a great deal. Indian J ophthalmology 70(3): 713.
6. Jordaan P, Dumotier B, Traebert M, Miller PE, Andre Ghetti, et al. (2021) Cardiotoxic Potential of Hydroxychloroquine, Chloroquine and Azithromycin in Adult Human Primary Cardiomyocytes. Toxicol Sci 180(2): 356-368.
7. Ruddy SM, Bergstrom R, Tivakaran VS (2026) Roth Spots. StatPearls. Treasure Island (FL): StatPearls Publishing Dec1.
8. Habib G, Lancellotti P, Antunes MJ (2015) 2015 ESC Guidelines for the management of infective endocarditis. Eur Heart J 36(44): 3075-3128.
9. Lertchavanakul A, Laksanaphuk P, Tomtitchong T (2002) Corneal arcus associated with dyslipidemia. J Med Assoc Thai 85 (Suppl)1: S231-S235.
10. Richards CE, Wallis DN (2005) Traumatic asphyxia. Injury 36(7): 829-830.
11. McCrindle BW, Rowley AH, Newburger JW (2017) Diagnosis, treatment, and long-term management of Kawasaki disease: a scientific statement for health professionals from the American Heart Association. Circulation 135(17): e927-e999.
12. Son MBF, Newburger JW (2018) Kawasaki disease. Pediatr Rev 39(2): 78-90.
13. Pasadhika S, Rosenbaum JT (2015) Ocular sarcoidosis. Clin Chest Med 36(4): 669-683.
14. Birnie DH, Nery PB, Ha AC, Beanlands RSB (2016) Cardiac sarcoidosis. J Am Coll Cardiol 68(4): 411-421.
15. Peter J, David S, Joseph G, Horo S, Danda D (2011) Ocular manifestations of Takayasu arteritis: a cross-sectional study. Retina 31(6): 1170-1178.
16. Hayreh SS (2021) Giant cell arteritis: its ophthalmic manifestations. Indian J ophthalmic 69(2): 227-235.
17. Inciardi RM, Lupi L, Zaccone G (2020) Cardiac involvement in a patient with coronavirus disease 2019. JAMA Cardiol 5(7): 819-824.
18. Sen M, Honavar SG, Sharma N, Sachdev MS (2021) COVID-19 and eye: a review of ophthalmic manifestations of COVID-19. Indian J Ophthalmology 69(3): 488-509.
19. Balasubramaniam S, Sreenivasan J, Dhawan A, Agarwal A (2022) A case of central retinal artery occlusion due to atrial myxoma with excellent visual recovery. Indian J ophthalmology 70(7): 2698.
20. Wong TY, Mitchell P (2004) Hypertensive retinopathy. N Engl J Med 351: 2310-2317.
21. Wong TY, McIntosh R (2005) Hypertensive retinopathy signs as risk indicators of cardiovascular morbidity and mortality. Br Med Bull 73-74: 57-70.
22. Milewicz DM, Braverman AC, De Backer J (2021) Marfan syndrome. Nat Rev Dis Primers 7: 64.
23. MacCarrick G, Black JH, Bowdin S (2014) Loeys-Dietz syndrome: a primer for diagnosis and management. Genet Med 16(8): 576-587.
24. Sodi A, Ioannidis AS, Pitz S (2006) Ophthalmological manifestations of Fabry disease. In: Mehta A, Beck M, Sunder-Plassmann G, editors. Fabry Disease: Perspectives from 5 Years of FOS. Oxford: Oxford Pharma Genesis Chapter 26.
25. Germain DP (2010) Fabry disease. Orphanet J Rare Dis 5: 30.
26. Morris AAM, Kozich V, Santra S (2017) Guidelines for the diagnosis and management of cystathionine beta-synthase deficiency. J Inherit Metab Dis 40: 49-74.
27. Christoffersen M, Frikke Schmidt R, Schnohr P, Jensen GB, Børge G Nordestgaard, et al. (2011) Xanthelasmata, arcus corneae, and ischaemic vascular disease and death in general population. BMJ 343: d5497.
28. Gertz MA (2020) Immunoglobulin light chain amyloidosis: 2020 update on diagnosis, prognosis, and treatment. Am J Hematol 95(7): 848-860.
29. Passman RS, Bennett CL, Purpura JM (2012) Amiodarone-associated optic neuropathy: a critical review. Am J Med 125(5): 447-453.
30. Piltz JR, Wertenbaker C, Lance SE, Slamovits T, H F Leeper et al. (1993) Digoxin toxicity. Recognizing the varied visual presentations. J Clin Neuroophthalmological 13(4): 275-280.
31. Wu S, Duan N, Wisdom JP, Kravitz RL, Owen RR, et al. (2015) Integrating Science and Engineering to Implement Evidence-Based Practices in Health Care Settings. Adm Policy Ment Health 42(5): 588-592.