Retinal arteriovenous oxygen saturation difference is inversely associated with cardiac output measured by echocardiography: a non-invasive pilot study Authors: Tomáš Křivohlavý, Martin Lada, Martin Malý, Martin Jakl, Kryštof Krása, Martin Šín Published: 2026 ABSTRACT Cardiac output is a key determinant of systemic perfusion and is commonly assessed using invasive or echocardiographic methods. Retinal oximetry provides a non-invasive estimate of retinal arteriovenous oxygen saturation difference and may reflect systemic haemodynamics. In this single-centre prospective observational pilot study, healthy volunteers and clinically stable cardiac patients underwent dual-wavelength retinal oximetry in both eyes and transthoracic echocardiography on the same day. Fourteen participants were included in the final analysis. Cardiac output showed an inverse association with the mean retinal arteriovenous oxygen saturation difference across both eyes (Pearson r = − 0.73, 95% CI − 0.91 to − 0.33, p = 0.003; Spearman ρ = −0.77, p = 0.001). These findings support the concept of systemic–ocular haemodynamic coupling. As this was a small exploratory study in a physiologically selected cohort, larger prospective studies are needed to assess reproducibility, external validity and potential clinical relevance. Introduction Retinal oximetry represents a relatively novel method in ophthalmology and has been studied in various physiological and pathophysiological scenarios. Using a fundus camera, multiple images are acquired in oxyhaemoglobin-sensitive (600 nm) and insensitive (570 nm) wavelengths, allowing estimation of oxygen saturation in the main retinal vessels including both arterioles and venules1. This method has been primarily studied in pathologies of the eye, namely in glaucoma2, ischaemic retinal diseases3, retinal vein occlusion4 and in children with inherited retinal diseases5 among others. Recently, there has been a growing interest in the use of retinal oximetry beyond ocular diseases. With respect to systemic circulation, retinal oximetry is sensitive to both local and systemic changes in oxyhaemoglobin saturation6. Owing to the close relationship between the retinal and central nervous system perfusion, multiple studies have focused on the possible early detection of neurological diseases and the changes in retinal oxygen saturation. In patients with moderate Alzheimer’s disease oxygen saturation in both retinal arterioles and venules was found to be significantly higher than in healthy subjects7. In patients with multiple sclerosis, venular oxygen saturation was increased compared to healthy individuals and the arteriovenous (AV) difference was lower than the control group, potentially caused by reduced oxygen uptake in patients with MS8. Together, these studies indicate that retinal oxygen saturation metrics can be influenced by both systemic oxygen delivery and local neurovascular/metabolic factors; therefore, their interpretation requires a clearly defined physiological context. The strong link between systemic and retinal circulation was further demonstrated in a 2017 study evaluating patients with chronic obstructive pulmonary disease (COPD) undergoing long-term oxygen therapy. These patients exhibited significantly lower oxyhaemoglobin saturation during ambient air breathing relative to healthy controls in both retinal arterioles and venules. Administration of their prescribed supplemental oxygen increased oxyhaemoglobin saturation in retinal arterioles but not in venules9. Cardiac output is defined as the volume of blood ejected by the heart per minute, expressed in litres per minute, and is calculated as stroke volume multiplied by heart rate. Its primary determinants are preload, afterload and myocardial contractility. Accurate assessment of cardiac output is essential for the management of cardiac patients, particularly those in critical care settings. Historically, cardiac output was first quantified by Adolf Fick in 1870 using a calculation he developed, now known as the Fick principle. The Fick principle links oxygen consumption to blood flow and the arterial–venous oxygen content difference, allowing cardiac output to be calculated as oxygen uptake divided by the AV oxygen content gradient10.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:CO=\frac{V{O}_{2}}{{C}_{a{O}_{2}}-{C}_{v{O}_{2}}}$$\end{document} CO = cardiac output (L/min). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:V{O}_{2}$$\end{document} = oxygen consumption (mL/min). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{a{O}_{2}}$$\end{document} = arterial oxygen content (mL O₂ per L blood). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{v{O}_{2}}$$\end{document} = venous oxygen content (mL O₂ per L blood). Non-invasive measurement of cardiac output using transthoracic echocardiography (TTE) was first validated in 1983 in a study conducted by Huntsman et al. who calculated stroke volume by measuring the left ventricular outflow tract area and multiplying it by the velocity time integral (VTI) of the LVOT flow. Stroke volume multiplied by heart rate then yields cardiac output. This method was validated against the gold-standard technique at the time, thermodilution11. Another important non-invasive approach, particularly useful when an adequate transthoracic window cannot be obtained, is transesophageal echocardiography (TEE). Using principles similar to those of TTE, transesophageal echo has also demonstrated a significant correlation with thermodilution in the assessment of cardiac output12. When comparing these methods, the general consensus today is that while in most cases there is not a significant difference between thermodilution (TD) and ultrasound techniques in measuring cardiac output, TD still remains the gold standard for accurate measurements, especially in the critically ill and those with significantly altered hemodynamic states, whereas echocardiography is better suited for monitoring trends rather than providing accurate absolute values13. From a physiological perspective, a lower cardiac output may be accompanied by reduced retinal perfusion, potentially increasing local oxygen extraction from the blood traversing the retinal circulation. This could theoretically result in a larger retinal arteriovenous oxygen saturation difference. However, retinal oxygen saturation is also influenced by local autoregulatory and metabolic mechanisms, and such a relationship should therefore be interpreted cautiously. In this study, we build primarily on a 2019 study by Van Keer et al., which investigated retinal oxygen saturation as a potential non-invasive tool for estimating mixed venous oxygen saturation and cardiac output. The results were promising as the authors identified a significant correlation between the inverse of cardiac output and the retinal arteriovenous oxygen saturation difference using Spearman’s rank correlation. Mixed venous oxygen saturation and cardiac output were obtained invasively via right heart catheterization14. One year later in 2020, Dinesen et al. published another study bridging ophthalmology and cardiology, demonstrating that coronary artery bypass surgery was independently associated with retinal vascular oxygen saturation, suggesting its potential use as a non-invasive marker for coronary artery disease15. In our research, we apply similar principles with key distinctions: cardiac output is measured non-invasively using transthoracic echocardiography and mixed-venous oxygen saturation is not assessed. Instead, we focus specifically on the inverse relationship between cardiac output and retinal arteriovenous oxygen saturation difference. Additionally, Pearson’s correlation coefficient was used to evaluate linear associations, while Spearman’s rank correlation coefficient was calculated as a robustness analysis given the small sample size. The aim of this study was to investigate whether retinal arteriovenous oxygen saturation difference is inversely associated with cardiac output measured non-invasively by transthoracic echocardiography. Methods Study population This was a single-centre prospective observational pilot study conducted at the Department of Internal Medicine and the Department of Ophthalmology of the Military University Hospital in Prague, Czech Republic. Participants were prospectively recruited from healthy volunteers and clinically stable inpatients. In both groups, eligibility required the absence of ocular or cardiovascular conditions specified in the exclusion criteria as likely to interfere with retinal oximetry or cardiac output assessment. Only participants meeting the prespecified eligibility criteria and with complete retinal oximetry and echocardiographic measurements were included in the final analysis (n = 14). Recruitment and data collection were conducted between September 2024 and January 2026. From an ophthalmologic perspective, we excluded individuals with diseases known to affect retinal oxygen saturation measurements including diabetic retinopathy, retinal vein occlusion, advanced glaucoma and uveitis16–19. Cardiovascular exclusion criteria targeted conditions that alter the hemodynamic parameters and could directly or indirectly influence cardiac output in unpredictable ways. Consequently, we excluded subjects with moderate or severe valvular disease, persistent arrhythmias, prosthetic heart valves and those with a history of cardiac surgery. We also excluded patients who had myocardial infarction within the past 3 months. Additionally, individuals with known carotid artery stenosis or with clinical findings raising concern for hemodynamically significant carotid disease were excluded, given the associated alterations in retinal oxygen saturation, particularly increased arterial saturation20. Assessment was based on medical history, carotid auscultation, and the absence of symptoms suggestive of carotid stenosis; no study-specific carotid imaging was performed. Exclusion criteria were intentionally extensive to minimize known sources of measurement artefact in retinal oximetry and major hemodynamic confounding. Consequently, the present cohort represents a physiologically “clean” proof-of-concept sample rather than a clinically representative population. The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Military University Hospital Prague / First Faculty of Medicine, Charles University (approval no. 108/19–59/2024, date 02.09.2024). Written informed consent was obtained from all participants prior to any study procedures. The sample size was not determined by formal power calculation, as the study was designed as an exploratory pilot investigation intended to assess feasibility and signal detection. Retinal oximetry Retinal oximetry was performed following pharmacological induction of mydriasis with 1% tropicamide. Adequate pupillary dilation was achieved after approximately 15–20 min. Fundus images were subsequently acquired using a fundus camera integrated with a retinal oximetry system (Oxymap T1). All images were evaluated by a single trained examiner using Oxymap Analyzer software. The minimum acceptable image quality threshold for inclusion in the analysis was set at a score of ≥ 5 on a 10-point scale. Vessel segments selected for oxygen saturation assessment were located within a peripapillary region defined by an inner and outer annulus corresponding to 1.5- and 3-times the optic disc diameter, respectively, with both centered on the optic nerve head. Within this region, major retinal arterioles and venules were manually identified in each of the four quadrants. Arterial and venous oxygen saturation values were then automatically computed by the software. The arteriovenous (A–V) oxygen saturation difference was calculated as the difference between arteriolar and venular saturation values. All measurements were obtained independently for each eye. The methodology used was similar to other works done in this field to ensure reproducibility for further studies21,22 (Fig. 1). Fig. 1Retinal oximetry workflow. (a) Native retinal oximetry image showing the major retinal arterioles and venules in a black-and-white fundus photograph. (b) Retinal oximetry image with a colour scale indicating oxygen saturation in the major retinal vessels; arterioles with higher oxygen saturation are shown in red and venules with lower oxygen saturation in green. (c) Retinal oximetry image showing the vessel segments selected for analysis within the predefined peripapillary measurement zone around the optic disc. Major retinal arterioles and venules were analysed within a standardized annular region. Images courtesy of the authors. Retinal oximetry workflow. (a) Native retinal oximetry image showing the major retinal arterioles and venules in a black-and-white fundus photograph. (b) Retinal oximetry image with a colour scale indicating oxygen saturation in the major retinal vessels; arterioles with higher oxygen saturation are shown in red and venules with lower oxygen saturation in green. (c) Retinal oximetry image showing the vessel segments selected for analysis within the predefined peripapillary measurement zone around the optic disc. Major retinal arterioles and venules were analysed within a standardized annular region. Images courtesy of the authors. Echocardiography Transthoracic echocardiography was performed on the same day as retinal imaging in all participants. The examinations followed the standard echocardiographic protocol established by the European Society of Cardiology (ESC) and the American Society of Echocardiography (ASE). In addition, stroke volume and cardiac output were measured, as recommended when these parameters are clinically relevant23. Stroke volume was determined using the left ventricular outflow tract (LVOT) continuity equation and subsequently multiplied by the heart rate to calculate cardiac output. The LVOT diameter was measured in the parasternal long-axis view (PLAX). The LVOT velocity-time integral (VTI) was obtained using the pulsed-wave (PW) Doppler in the LVOT from the apical five-chamber view in accordance with recommended methodology, using the formulas shown below24 (Fig. 2).\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{SV}=\left({\uppi\:}\hspace{0.17em}\frac{{d}_{\mathrm{LVOT}}^{2}}{4}\right)\hspace{0.17em}\mathrm{VT}{\mathrm{I}}_{\mathrm{LVOT}}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{CO}=\mathrm{SV}\cdot\:\mathrm{HR}$$\end{document} SV = stroke volume (mL). d2LVOT = LVOT diameter (cm). VTILVOT = LVOT velocity-time integral (cm). CO = cardiac output (L/min). HR = heart rate (beats per minute). Fig. 2Echocardiographic derivation of stroke volume and cardiac output. (a) Schematic illustration of stroke volume calculation using the continuity equation, showing the relationship between left ventricular outflow tract area and velocity-time integral in the derivation of stroke volume and cardiac output. Created in BioRender. (b) Transthoracic echocardiographic image illustrating measurement of the left ventricular outflow tract diameter in the parasternal long-axis view for calculation of left ventricular outflow tract cross-sectional area. (c) Transthoracic echocardiographic Doppler image showing acquisition of the left ventricular outflow tract velocity-time integral in the apical five-chamber view for calculation of stroke volume. Images courtesy of the authors. Echocardiographic derivation of stroke volume and cardiac output. (a) Schematic illustration of stroke volume calculation using the continuity equation, showing the relationship between left ventricular outflow tract area and velocity-time integral in the derivation of stroke volume and cardiac output. Created in BioRender. (b) Transthoracic echocardiographic image illustrating measurement of the left ventricular outflow tract diameter in the parasternal long-axis view for calculation of left ventricular outflow tract cross-sectional area. (c) Transthoracic echocardiographic Doppler image showing acquisition of the left ventricular outflow tract velocity-time integral in the apical five-chamber view for calculation of stroke volume. Images courtesy of the authors. Statistical analysis Statistical analyses were performed in 14 participants with complete retinal oximetry and echocardiographic data. The primary comparison of interest was the association between cardiac output measured by transthoracic echocardiography and retinal arteriovenous oxygen saturation difference, analysed for the mean of both eyes. Pearson’s correlation coefficient was used to assess linear association, and Spearman’s rank correlation coefficient was calculated as a robustness analysis given the small sample size and potential non-normality. Normality and linearity were assessed descriptively using the Shapiro–Wilk test and scatterplots. For Pearson correlations, 95% confidence intervals were calculated. Linear regression was performed for exploratory purposes and R² is reported. All tests were two-tailed, and p < 0.05 was considered statistically significant. The sample size was not determined by formal power calculation, as the study was designed as an exploratory pilot investigation intended to assess feasibility and signal detection. Analyses were performed using Microsoft Excel (Microsoft Corp., Redmond, WA, USA) and JASP (version 0.95.4.0; JASP Team, Amsterdam, The Netherlands). Results A total of fourteen participants were included in the analysis. The study population consisted of both healthy volunteers and clinically stable cardiology patients, 11 males and 3 females. The mean age was 43.8 ± 19.1 years, with ages ranging from 25 to 77 years. The demographic characteristics and hemodynamic parameters of the study population are summarised in Table 1. Blood pressure and peripheral oxygen saturation values were within clinically stable ranges across the study population. Table 1Baseline characteristics and hemodynamic parameters of the study population. Values are presented as mean ± standard deviation unless otherwise indicated. CO = cardiac output (L/min); SV = stroke volume (mL); HR = heart rate (beats/min); BMI = body mass index (kg/m²); BSA = body surface area (m²); SBP = systolic blood pressure (mmHg); DBP = diastolic blood pressure (mmHg); SpO₂ = peripheral oxygen saturation measured on room air (%).Participants (n)14Healthy volunteers (n)9Cardiac patients (n)5Sex (M/F)11 / 3Age (years)43.8 ± 19.1BMI (kg/m²)27.8 ± 6.4BSA (m²)2.01 ± 0.26SBP (mmHg)120 ± 16DBP (mmHg)75 ± 10SpO₂ on room air (%)≥ 96 (all participants)Cardiac output (L/min)5.78 ± 0.98Stroke volume (mL)83.2 ± 12.6Heart rate (beats/min)70.4 ± 12.6 Table 1Baseline characteristics and hemodynamic parameters of the study population. Values are presented as mean ± standard deviation unless otherwise indicated. CO = cardiac output (L/min); SV = stroke volume (mL); HR = heart rate (beats/min); BMI = body mass index (kg/m²); BSA = body surface area (m²); SBP = systolic blood pressure (mmHg); DBP = diastolic blood pressure (mmHg); SpO₂ = peripheral oxygen saturation measured on room air (%).Participants (n)14Healthy volunteers (n)9Cardiac patients (n)5Sex (M/F)11 / 3Age (years)43.8 ± 19.1BMI (kg/m²)27.8 ± 6.4BSA (m²)2.01 ± 0.26SBP (mmHg)120 ± 16DBP (mmHg)75 ± 10SpO₂ on room air (%)≥ 96 (all participants)Cardiac output (L/min)5.78 ± 0.98Stroke volume (mL)83.2 ± 12.6Heart rate (beats/min)70.4 ± 12.6 Baseline characteristics and hemodynamic parameters of the study population. Values are presented as mean ± standard deviation unless otherwise indicated. CO = cardiac output (L/min); SV = stroke volume (mL); HR = heart rate (beats/min); BMI = body mass index (kg/m²); BSA = body surface area (m²); SBP = systolic blood pressure (mmHg); DBP = diastolic blood pressure (mmHg); SpO₂ = peripheral oxygen saturation measured on room air (%). The mean retinal arteriovenous (AV) oxygen saturation difference across both eyes was 37.8 ± 8.0%, with values ranging from 28.0% to 52.5%. The mean AV oxygen saturation difference for the right eye was 38.9 ± 9.9%, with values ranging from 25.0% to 60%. The left eye’s AV oxygen saturation difference was 36.7 ± 7.7%, with values ranging from 26.0% to 48%. Arterial retinal oxygen saturation averaged 96.3 ± 5.1% in the right eye and 97.3 ± 6.0% in the left eye. Venous retinal oxygen saturation averaged 57.4 ± 12.6% in the right eye and 60.6 ± 10.1% in the left eye. Across both eyes, arterial retinal oxygen saturation averaged 96.8 ± 5.5%, and venous retinal oxygen saturation averaged 59.0 ± 11.3%. Retinal oximetry measurements for the right eye, left eye and the mean of both eyes are summarised in Table 2. Table 2Retinal oximetry measurements. Values are presented as mean ± standard deviation. sO₂ art. = oxygen saturation in retinal arterioles; sO₂ ven. = oxygen saturation in retinal venules; AV sat. diff = arteriovenous oxygen saturation difference.Retinal parametersRight eyeLeft eyeMean of both eyessO₂ art. (%)96.3 ± 5.197.3 ± 6.096.8 ± 5.5sO₂ ven. (%)57.4 ± 12.660.6 ± 10.159.0 ± 11.3AV sat. diff (%)38.9 ± 9.936.7 ± 7.737.8 ± 8.0 Table 2Retinal oximetry measurements. Values are presented as mean ± standard deviation. sO₂ art. = oxygen saturation in retinal arterioles; sO₂ ven. = oxygen saturation in retinal venules; AV sat. diff = arteriovenous oxygen saturation difference.Retinal parametersRight eyeLeft eyeMean of both eyessO₂ art. (%)96.3 ± 5.197.3 ± 6.096.8 ± 5.5sO₂ ven. (%)57.4 ± 12.660.6 ± 10.159.0 ± 11.3AV sat. diff (%)38.9 ± 9.936.7 ± 7.737.8 ± 8.0 Retinal oximetry measurements. Values are presented as mean ± standard deviation. sO₂ art. = oxygen saturation in retinal arterioles; sO₂ ven. = oxygen saturation in retinal venules; AV sat. diff = arteriovenous oxygen saturation difference. Cardiac output (CO) measured by echocardiography ranged from 4.34 to 7.48 L/min and the mean value was 5.78 ± 0.98 L/min. Mean stroke volume was 83.19 mL ± 12.55 mL, with values ranging from 68 mL to 106 mL. The average heart rate was 70.36 ± 12.61 beats per minute, values ranging from 49 to 88 beats per minute. Correlation analysis using Pearson’s correlation coefficient demonstrated an inverse association between retinal AV oxygen saturation difference and echocardiographically derived cardiac output. Using the mean AV difference across both eyes yielded r = − 0.73 (95% CI − 0.91 to − 0.33, p = 0.003). Linear regression (CO ~ mean AV difference) yielded R² = 0.53. The inverse association remained similar using Spearman’s rank correlation for the mean AV difference across both eyes (ρ = −0.77, p = 0.001) (Fig. 3). Fig. 3Association between cardiac output and retinal arteriovenous oxygen saturation difference. Scatterplot showing the inverse relationship between cardiac output and the mean retinal arteriovenous oxygen saturation difference across both eyes. Association between cardiac output and retinal arteriovenous oxygen saturation difference. Scatterplot showing the inverse relationship between cardiac output and the mean retinal arteriovenous oxygen saturation difference across both eyes. Discussion The observed inverse association should be interpreted as a physiological association rather than evidence that retinal AV oxygen saturation difference directly reflects systemic cardiac output. Retinal oxygen extraction is influenced by both systemic oxygen delivery and local autoregulatory mechanisms; the relationship is therefore biologically plausible but does not imply a direct causal pathway. This study should be viewed as hypothesis-generating. Although we observed a moderate-to-strong inverse association between retinal AV oxygen saturation difference and echocardiographically estimated cardiac output, the small sample size and limited clinical spectrum preclude conclusions regarding clinical utility, prediction of individual cardiac output values, or generalisability to broader patient populations. To reduce side-specific variability and avoid overinterpretation of eye-level differences in this small cohort, the primary analysis was based on values averaged across both eyes. The study cohort was small and heterogeneous, comprising both healthy volunteers and clinically stable cardiac patients. Although this reflects the exploratory proof-of-concept nature of the study, it limits the interpretability of subgroup differences and the generalizability of the findings. The mean BMI of the cohort was in the overweight range; however, BMI may incompletely reflect body composition or cardiovascular fitness in a small sample that included physically active male volunteers. Accordingly, the term “healthy volunteers” in this study denotes the absence of known relevant ocular and cardiovascular comorbidities based on prespecified exclusion criteria, rather than a strictly homogeneous low-risk population. There are certain potential limitations that we acknowledge. The participants included in this study were either healthy volunteers or clinically stable cardiology patients with normal values of cardiac output. Therefore, our findings may apply primarily in those patients, but the method could fail in those with a compromised cardiovascular system (e.g. severe systolic dysfunction, hypotension, cardiac tamponade). The sample size in this pilot study is too small to draw definite conclusions, but it shows a promising trend. Another potential pitfall in a clinical setting could be in examining patients that have an undiagnosed condition which interferes with retinal oximetry measurement – conditions that we purposefully excluded from our study (e.g. severe valvular disease, glaucoma, diabetic retinopathy etc.). The extensive exclusion criteria substantially limit generalizability; however, this was intentional. The aim of this pilot study was to examine whether a measurable relationship between retinal arteriovenous oxygen saturation difference and cardiac output could be detected under relatively controlled physiological conditions, before introducing additional sources of ocular or hemodynamic variability. Accordingly, populations in whom non-invasive cardiac output estimation might be most clinically relevant (e.g., patients with significant arrhythmias, valvular disease, or diabetes-related retinal disease) were not represented. Validation in these groups will be necessary to determine whether the observed association remains detectable under more complex clinical conditions. Multiple limitations exist in CO measurement using transthoracic echocardiography – some patients are difficult to image correctly, often due to obesity – higher BMI correlates positively with the incidence of non-diagnostic studies25, lung emphysema due to chronic obstructive pulmonary disease, intra-abdominal hypertension, mechanical ventilation and others26. Retinal oximetry may be less dependent on acoustic windows and could be feasible in some patients in whom transthoracic imaging is technically limited; however, this requires dedicated validation. A major source of error in TTE derived CO is the measurement of the LVOT area, calculated as π * (LVOT diameter2 ) / 4, meaning that even a small mistake in LVOT diameter results in gross mistakes in stroke volume and cardiac output24. It is well established that inter-observer variability affects those measurements and that the results also depend substantially on the examiner’s skill and experience. Retinal oximetry is unlikely to share the same source of measurement error (LVOT diameter), but it introduces its own potential sources of variability and therefore cannot be considered a direct substitute. Furthermore, there is also significant potential in automating the whole process of retinal oximetry with better technology and artificial intelligence as described in relevant literature27–29, making it even more appealing as an alternative tool to assess systemic hemodynamics if further studies validate this method. The observed range of cardiac output values may partly reflect true physiological interindividual variability, as cardiac output is influenced by heart rate, stroke volume, loading conditions and overall cardiovascular state. However, some contribution of echocardiographic measurement variability cannot be excluded. To reduce methodological variability, all cardiac output measurements were performed by a single examiner using a standardized protocol, and echocardiography and retinal oximetry were obtained on the same day in all participants to limit the influence of temporal physiological fluctuations. Retinal oximetry measurements may be influenced by image quality, vessel selection, and calibration characteristics of the device, which may introduce measurement variability. Because carotid imaging was not performed systematically, asymptomatic or subclinical carotid stenosis cannot be fully excluded. Conclusion Retinal arteriovenous oxygen saturation difference was inversely associated with echocardiographically estimated cardiac output in this exploratory pilot study. Although these findings do not establish clinical utility, they support further investigation of retinal oximetry as a non-invasive marker of systemic–ocular haemodynamic interactions in larger prospective studies.