Retinal oximetry: new insights into ocular and systemic diseases

Abstract

Retinal oximetry, which refers to the measurement of the oxygen saturation of haemoglobin in retinal blood, has emerged as a promising tool for understanding ocular and systemic diseases over the past few decades. Advances in traditional dual-wavelength measurement techniques, as well as their integration with more advanced technologies, have driven significant progress in the field. Researchers have utilised commercially available devices to explore the applications of retinal oximetry in both healthy individuals and various disease states. To provide a comprehensive overview of the current status and development trends of retinal oximetry in ophthalmology, this review analyses relevant studies on retinal oximetry, including its principles, techniques, instrumentation, and applications in healthy individuals, ocular diseases, and systemic disorders. As a biomarker, retinal oximetry provides reliable insights into the retinal and systemic microcirculation in a noninvasive, rapid, and convenient manner. It has demonstrated potential value in the measurement of standardised data and in contributing to the understanding of disease mechanisms. However, its measurement accuracy is influenced by factors such as vessel diameter, retinal pigmentation, and optical media transparency. Future advancements, including wide-field imaging technology, nonmydriatic technology, integration with blood flow measurement techniques, and artificial intelligence, are expected to further improve the accuracy and clinical application potential of retinal oximetry.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00417-025-06831-8.

What is known

• Retinal oximetry is a noninvasive and reliable technique for assessing oxygen saturation in retinal blood vessels.

• Retinal oximetry has been applied in evaluating retinal oxygen saturation in healthy individuals and in various ocular and systemic disorders, such as diabetic retinopathy, glaucoma, cardiovascular diseases and kidney diseases.

What is new

• This review provides a comprehensive analysis of the current status and development trends on retinal oximetry.

• Future advancements in retinal oximetry are likely to be driven by the wide-field imaging, nonmydriatic technology, blood flow measurement techniques, and artificial intelligence, all of which will enhance its diagnostic and prognostic potential.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00417-025-06831-8.

Introduction

The eye, particularly retinal tissue, is one of the most sophisticated structures in the human body. It requires a significant amount of energy to perform its daily functions, which cannot be accomplished without a steady supply of oxygen [1]. Retinal oximetry serves as a potential indicator of ocular physiology, with its measurements offering insights into factors that may reflect the overall health status of individuals. Abnormal variations in blood oxygen levels not only reflect ocular diseases directly but also provide a window for observing and understanding the development of systemic diseases. Numerous ocular and systemic diseases, such as diabetic retinopathy, glaucoma and Alzheimer's disease, can lead to alterations in the retinal microcirculation, thereby impairing oxygen delivery to the retina [2]. Therefore, analysing retinal oxygen saturation is important for understanding the mechanisms of these diseases and may contribute to early monitoring and diagnosis. Research on retinal oximetry has made significant progress in recent decades, providing new insights into ocular and systemic diseases. Figure 1 shows an overview of the structure of this paper.

Fig. 1.

Overview of retinal oximetry: principles, techniques, and applications. (A) Picture of the measurement principles of retinal oximetry; details are shown in Fig. 2 (B) Picture of the instrumentation used for retinal oximetry; details are shown in Fig. 3 (C) Picture of the distribution map of retinal oximetry; details are shown in Fig. 3. Abbreviations: DR, diabetic retinopathy; RAO, retinal artery occlusion; RVO, retinal vein occlusion; AMD, age-related macular degeneration; RP, retinitis pigmentosa

Methods

The literature reviewed in this paper was sourced from major academic databases, including PubMed, Scopus, and Web of Science. A comprehensive search was conducted using keywords such as"retinal oximetry","oxygen saturation","ocular diseases"and"systemic diseases". The search covered studies published up to 2023, focusing on original research articles that provided valuable insights into the techniques, applications, and advancements of retinal oximetry. Articles were included if they were relevant to the scope of this review and met basic quality criteria, such as methodological rigor and relevance to the topic. Studies were excluded if they lacked sufficient methodological detail or were deemed outside the focus of this review. In total, approximately 77 articles were identified and reviewed to provide a balanced overview of the field. The details of these articles are shown in Table 1 and Supplementary Table 1–3.

Table 1.

Typical studies on the application of retinal oximetry in systemic diseases

No	Author (year)	Devices	Wavelengths (nm)	Groups	Subjects	Age	Sex (m/f)	Results, mean ± SD			OD vs. OS
								Artery (%)	Vein (%)	AVD (%)
1	Wenbo Zhang et al. (2021) [88]	Oxymap T1	570	Healthy	16	65.5 (41–76)	8/8	93.22 ± 5.98	56.57 ± 7.05	36.65 ± 7.33	—
		(Reykjavik, Iceland)	600	ICAS	16	64 (44–83)	12/4
				- stenotic side				100.14 ± 10.27 (p = 0.027*)	56.50 ± 10.79 (p = 0.983)	43.63 ± 7.71 (p = 0.013*)
				- contralateral side				96.55 ± 7.50 (p = 0.175)	57.42 ± 9.84 (p = 0.780)	39.39 ± 6.33 (p = 0.267)
2	Karel Van Keer et al. (2019) [90]	Oxymap T1	570	Right heart catheterisation	14	62.7 (21–77)	12/2	97.6 ± 5.1	62.7 ± 6.5	34.9 ± 4.3	randomly
			600
3	Anna Bryndis Einarsdottir et al. (2016) [89]	Oxymap T1	570	Healthy	18 Caucasians	64 ± 7	8/10	90.5 ± 3.1	49.7 ± 7.0	40.8 ± 6.6	mainly OD
			600	AD
				- mild AD	8 Caucasians	65 ± 9	3/5	93.1 ± 2.9 (p = 0.408)	50.6 ± 4.4 (p = 0.606)	42.5 ± 5.6 (p = 0.477)
				- moderate AD	10 Caucasians	72 ± 4	5/5	94.2 ± 5.4 (p = 0.028*)	51.9 ± 6.0 (p = 0.020*)	42.3 ± 8.8 (p = 0.970)
4	Stefan Palkovits et al. (2013) [93]	Imedos UG (Jena, Germany)	548	COPD	16	66.6 ± 5.5	13/3				the better visual acuity eye
			610	- with oxygen therapy, 1 st day				92.2 ± 4.3	67.6 ± 7.8	25 ± 5
				- without oxygen therapy, 1 st day				decline 2.1 ± 3.1 (p = 0.02*)	decline 2.5 ± 4.7 (p = 0.06)	25 ± 6 (p = 0.9)
5	Thorunn Scheving Eliasdottir et al. (2017) [94]	Oxymap T1	570	Healthy	11 Caucasians	69.6 ± 4.9	—	93.4 ± 4.3	55.2 ± 5.5	38.2 ± 4.0	OU
			600	COPD	11 Caucasians	70.4 ± 5.4	4/7	87.2 ± 4.9 (p = 0.02*)	45.0 ± 10.3 (p = 0.01*)	42.2 ± 8.0 (p = 0.17)
6	Ruibao Liu et al. (2021) [98]	Oxymap T1	570	Healthy	103 Chinese	43.82 ± 11.85	45/58	96.95 ± 3.12	62.15 ± 4.94	34.79 ± 4.60	OU
			600	CKD	83 Chinese	47.04 ± 12.99	43/40	99.80 ± 3.67 (p < 0.001**)	63.21 ± 5.54 (p = 0.052)	36.58 ± 5.07 (p < 0.001**)
7	Mozhgan Sharifizad et al. (2021) [99]	Imedos UG	548	Healthy nonpregnancy	33	27.8 ± 4.9	—	95 ± 2	64 ± 2	—	only OD
			610	Low-risk pregnancy	29	28.1 ± 5.1	—
				- TP2				94 ± 2	64 ± 2	—
				- TP3				94 ± 3	65 ± 2	—
				- PP				93 ± 3	64 ± 2	—
				High-risk pregnancy	25	29.2 ± 5.4	—
				- TP2				93 ± 2	68 ± 2	—
				- TP3				93 ± 2	69 ± 2	—
				- PP				93 ± 2 (p = 0.635)	68 ± 2 (p = 0.008**)	—

Development background and measurement principles of retinal oximetry

Retinal oximetry is a biomarker for measuring the oxygen saturation of haemoglobin in retinal vessels using a specially designed oximeter [3]. The techniques can be categorised into invasive and noninvasive methods. Invasive methods for measuring retinal oxygen saturation include the use of oxygen-sensitive electrodes, which directly measure oxygen partial pressure, and the injection of oxygen-sensitive dyes into the bloodstream, which allows for the indirect measurement of haemoglobin oxygen saturation. However, these methods are currently only used in animal experiments because of their invasive characteristics [4]. Noninvasive spectral imaging, with its convenient and noninvasive advantages, has become the main method for retinal oximetry at present.

Research on spectroscopic imaging of retinal oximetry dates back to 1959, when Hickam and Frayser invented the first noninvasive dual-wavelength spectrophotometric retinal oximeter in history and successfully obtained fundus images captured by two different wavelength light sources using a specially designed filter [5]. Over the past five decades, spectrophotometric-based retinal oximetry has undergone significant advances, driven by the rapid development of digital cameras and computer technology.

Most methods used for retinal oximetry spectroscopic imaging are based on the Lambert–Beer law, which describes the relationship between light absorption and the concentration of absorbing molecules. Oxyhaemoglobin (HbO₂) and deoxyhaemoglobin (Hb) exhibit distinct absorption properties across different wavelengths, except at isosbestic points where their absorption coefficients are equal. For example, at 570 nanometres (nm), the extinction coefficients of HbO₂ and Hb are approximately equal, meaning that the optical density (OD) of blood is insensitive to changes in oxygen saturation at this wavelength. In contrast, at 600 nm, the extinction coefficients of Hb and HbO₂ differ significantly, making the OD sensitive to oxygen saturation (Fig. 2) [6]. However, other pairs of wavelengths can also be used, depending on the specific instrument and its design.

Fig. 2.

Light absorptivity of Hb and HbO₂ at different wavelengths. This figure was created on the basis of data from Zijlstra et al. (2000) [6]

The optical density ratio (ODR), defined as the ratio of the OD at two distinct wavelengths, is approximately linearly related to the oxygen saturation of blood (SO₂) [7]:

$${\mathit{SO}}_2=a\times ODR+b$$ 1

Here, the ODR can be calculated as the OD₆₀₀/OD₅₇₀, representing the OD values measured at 600 nm and 570 nm for the same vessel location, respectively, and a and b are calibration constants. The OD is calculated as:

$$OD=\text{log}\left(I_0,/,I\right)$$ 2

In this equation, I₀ is the reflectance measured beside the vessel (incident light intensity), and I is the reflectance measured on the vessel (outgoing light intensity). These reflectance values are derived from the pixel greyscale values of the images.

By comparing the images captured at these two wavelengths, the oxygen saturation levels of the retinal arterioles and veins can be calculated separately. These findings provide valuable insights into retinal oxygenation dynamics.

Common techniques, instrumentation, and calibration in retinal oximetry

Retinal oximetry spectroscopic imaging techniques can be classified into three categories on the basis of the imaging devices used: fundus camera-based, optical coherence tomography (OCT)-based, and ophthalmoscope-based retinal oximetry [8]. The fundus camera-based technique is the most commonly used technique and has been extensively studied both nationally and internationally. The Oxymap oximeter (Reykjavik, Iceland) and the Imedos oximeter (Jena, Germany) are the most commonly used instruments equipped with this technique. The studies reviewed in this paper are also mainly based on these two types of equipment.

For example, the Oxymap T1 oximeter is mounted on a fundus camera equipped with two digital cameras and a custom-made optical adapter. The system illuminates the retina with white light, and the reflected light is filtered into two wavelengths (570 nm and 600 nm) for dual-wavelength imaging to calculate retinal oxygen saturation. The captured images are processed by oximetry software to derive final oxygen saturation values (Fig. 3A and B). Furthermore, a novel retinal functional imaging device named the MEFIAS 3200 has been developed in China (Fig. 3C and D) [9].

Fig. 3.

The instrumentation and techniques of retinal oximetry (Oxymap T1 and MEFIAS 3200 as illustrations). (A) Physical image of Oxymap T1. (B) Retinal oximetry distribution map of Oxymap T1. Figure 3A and B are from https://www.oxymap.com/. (C) Image of MEFIAS 3200. (D) Retinal oximetry distribution map of MEFIAS 3200. Figure 3C is from https://www.newvision-sz.com/

Notably, the values of the constants 'a' and 'b' must be determined in advance to calculate the retinal oxygen saturation of the subject according to Eq. (1) in Chapter 3. This value is calculated from the retinal oxygen saturation of healthy individuals using Eq. (1), and the retinal oxygen saturation of healthy individuals is derived from Schweitzer et al.'s study [10]. The ODR values of the images taken by an instrument are influenced by the characteristics of the instrument, external conditions such as lighting, and a variety of other complex factors. Therefore, calibrating the 'a' and 'b' values of each instrument before application is crucial, as this calibration directly impacts the accuracy and consistency of the calculated oximetry. The calibration methodology of the retinal oximetry spectroscopic imaging technique has been described in detail previously [7]. Subsequent studies have demonstrated the high reproducibility and reliability of the technique and its calibration strategy [11–14].

To increase measurement accuracy and minimise light scattering effects, scientists have developed multiwavelength and hyperspectral measurement methods [15–21]. In recent years, researchers have also expanded the scope of this technology towards deep learning [22], automated analysis [23], and integration with other structural or functional imaging techniques [9, 24]. Supplementary Table 1 summarises the key information from typical studies, including the previously mentioned retinal oximetry techniques based on OCT [25–28] and ophthalmoscopy [29–32].

Application of retinal oximetry in healthy individuals

Retinal oximetry provides valuable information about oxygen saturation in the retina, offering insights into the normal physiology of the visual system, and may offer a valuable reference for future studies on monitoring ocular health and disease progression.

In 1999, researchers measured retinal oxygen saturation levels in healthy individuals. They combined spectral imaging techniques with empirical modelling to propose a model for calculating oxygen saturation within a specific wavelength range and verified the model’s reproducibility, accuracy, and applicability. Furthermore, they were the first to investigate the differences in retinal arteriovenous oxygen saturation, as well as the relationships among oxygen saturation, vessel diameter, and retinal pigment, providing a reliable reference for subsequent studies [10]. Many later studies on retinal oximetry have relied on the results provided by Schweitzer et al.

Using the Oxymap T1 fundus oximeter, researchers reported that the mean arteriovenous difference in retinal oxygen saturation in healthy Caucasians was approximately 36.7%, with no statistically significant difference between the right and left eyes. This study analysed the relationship between retinal oxygen saturation and various factors, including age, sex, current smoking status, measured finger pulse oxygen saturation, and ocular perfusion pressure, and described variations in retinal oxygen saturation across different retinal quadrants, providing standardised data for retinal oxygen saturation in Caucasian populations [33]. Additionally, colour Doppler imaging (CDI) in combination with spectroscopic imaging revealed a weak negative correlation between the retinal arteriovenous oxygen difference and the ophthalmic artery resistance index (RI) in Caucasians. These findings suggest that the self-regulatory mechanism of the ophthalmic artery plays a role in maintaining the balance between oxygen supply and demand [34].

Some researchers first described retinal oximetry in healthy individuals from diverse ethnic backgrounds. They assessed the associations of age, sex, ethnicity, refraction, iris colour, history of controlled systemic hypertension, and smoking habits with retinal oxygen saturation. The study highlighted that age and ethnic characteristics may impact the measurements [35]. However, it should be noted that these differences might also be influenced by factors such as fundus pigmentation or apparent vessel diameters, which could lead to artefacts. Further research is needed to determine whether these observed variations are truly reflective of physiological differences or are artefactual in nature.

In China, a study of healthy young individuals aged 19–30 years revealed that their retinal oxygen saturation was comparable to that of Caucasians [36]. Additionally, investigations in the Chinese children and adolescents have revealed that retinal arterial and venous oxygen saturation levels are lower than in adults and increase gradually with age. Furthermore, individuals with higher degrees of myopia and longer axial lengths presented higher retinal oxygen saturation levels [37, 38]. These studies address a gap in the available data regarding these populations and contribute to a better understanding of how various factors, including age and ethnicity, influence retinal oxygen saturation. Future studies should appropriately control for these variables.

In addition to the factors mentioned above, several investigators have demonstrated that retinal oximetry is altered by other internal and external factors; internal factors, such as the quadrant in which the blood vessels are located [39] and diurnal rhythms [40]; and external factors, such as light [41], flash [42, 43] and altitude [44]. While Heitmar and Cubbidge investigated flash intensity as a confounder in retinal oximetry, Hammer et al. used flicker light as a physiological stimulus of neuronal activity and reported an increase in venous oxygen saturation in healthy subjects. For further details, please refer to Supplementary Table 2.

Application of retinal oximetry in ocular diseases

Diabetic retinopathy and retinal oximetry

Diabetic retinopathy (DR) is a common complication of diabetes that often involves abnormalities in retinal blood flow and the oxygen supply. These changes can result in impaired vision in diabetic patients. Studies have demonstrated that the mechanisms underlying DR-induced changes in retinal blood oxygenation are complex and multifactorial. Factors such as damage to the retinal vascular endothelium and changes in capillary morphology that result in less efficient oxygen delivery and uneven oxygen distribution may lead to retinal hypoxia [45].

To investigate the variations in retinal oxygen saturation in patients with distinct degrees of DR, researchers divided 41 patients with type 2 diabetes mellitus into four subgroups: mild nonproliferative DR (nPDR), moderate nPDR, severe nPDR, and proliferative DR (PDR). They reported that venous oxygen saturation tended to increase, whereas the arteriovenous difference decreased as disease severity progressed [46, 47]. Similar results were reported by others, although no significant difference was observed in the arteriovenous oxygen saturation difference in their study [45]. Researchers subsequently conducted a prospective cohort study to investigate changes in retinal oxygen saturation over time in patients with DR. The findings indicated that, over time, oxygen saturation increased more in the retinal veins than in the arterioles, leading to a decrease in the arteriovenous oxygen saturation difference, whereas there was no significant change in the grading of retinopathy. These results suggest that changes in retinal oxygen saturation may precede the progression of DR and could serve as an early marker of DR [48].

Some studies have further refined the subgroups of diabetic patients and performed more comprehensive analyses. For example, when diabetic patients without retinopathy were included, a progressive increase in arterial and venous oxygen saturation in the major branches of the retina was observed with worsening lesion severity [49]. Furthermore, when DR progresses to diabetic maculopathy, retinal autoregulation decreases, leading to a reduction in the arteriovenous oxygen saturation difference, as shown by a decrease in oxygen extraction [50].

As imaging technology has evolved, researchers have combined spectral imaging methods with other techniques to further expand the application of retinal oximetry in the study of DR. Hyperspectral computed tomographic imaging spectroscopy (HCTIS) technology, for instance, combines hyperspectral imaging and computed tomography to simultaneously acquire spectra from 450–700 nm with a spectral resolution of approximately 4 nm, providing more detailed and accurate information on blood oxygen levels. The blood oxygen levels of patients with PDR differed significantly from those in other subgroups, suggesting that HCTIS could serve as a tool to guide laser therapy [51]. Furthermore, spectral imaging combined with intravenous fluorescein angiography (IVFA) revealed a significant correlation between retinal oxygen saturation and the percentage of retinal ischaemia in patients with various stages of DR [52]. Another study combining spectral imaging with HD-OCT revealed that the foveal avascular zone (FAZ) area of diabetic patients was significantly enlarged by 27.18% compared with that of healthy individuals. These findings suggest that microvascular oxygen metabolic disorders have already occurred in diabetic patients before the development of significant retinopathy [53].

DR is a complication of diabetes mellitus; therefore, systemic indices such as blood glucose levels typically change before localised DR lesions appear. To study the relationship between retinal oximetry and DR, it is essential to obtain information on factors such as blood glucose and glycated haemoglobin to assess the systemic and retinal status of patients with DR comprehensively [54].

Glaucoma and retinal oximetry

Retinal oximetry is a subject of significant interest in glaucoma research. Glaucoma, one of the leading causes of blindness worldwide, is typically associated with elevated intraocular pressure (IOP), optic nerve degeneration, and visual field defects. However, the relationships among retinal oxygen saturation, blood flow status, and the development of glaucoma remain unclear. Factors such as elevated IOP, increased hypoxia-inducible factor 1a (HIF- 1a), and decreased blood flow may contribute to the structural and functional damage characteristic of glaucoma [55, 56]. Mechanical and vascular theories have been proposed to explain these changes [57].

With respect to retinal oxygen saturation in patients with normal-tension glaucoma (NTG) and high-tension primary open-angle glaucoma (POAG), it has been demonstrated that arterial oxygen saturation is reduced in patients with NTG. This reduction may be indicative of an abnormality in arterial blood supply. Additionally, a correlation between arterial oxygen saturation and the degree of glaucomatous damage has been identified [58]. Moreover, retinal oximetry is believed to correlate with areas of visual field damage [59]. Some researchers have proposed that the exacerbation of visual field defects in glaucoma is associated with increased venous oxygen saturation and reduced arteriovenous differences. This phenomenon may be attributed to pathological changes in glaucoma, such as tissue atrophy and loss of retinal ganglion cells, which result in decreased oxygen consumption, rather than changes in blood oxygen saturation, causing the pathological changes in glaucoma [56, 60–62]. Notably, some researchers reported no statistical correlations between retinal oxygen saturation and IOP, finger oximetry values, or perfusion pressure [60].

Furthermore, other studies have investigated the effects of medications on retinal oxygen saturation in patients with glaucoma. A randomised, crossover, double-blind trial demonstrated that both Brinzolamide and Dorzolamide, which are carbonic anhydrase inhibitors (CAIs), increased retinal oxygen saturation and improved ocular blood flow status in patients with POAG [63]. Compared with Timolol (a β-adrenergic receptor blocker) monotherapy, the combination of Dorzolamide and Timolol did not significantly decrease retinal oxygen saturation [64]. This may be because the increased retinal blood flow was sufficient to meet the oxygen demand of the tissue [65]. On the other hand, the combination of 0.5% Tropicamide and 5% Phenylephrine in patients with glaucoma had no effect on retinal oxygen saturation or vessel diameter compared with Tropicamide alone but significantly improved the quality of retinal images [66].

In addition to medications, surgery is an important treatment for glaucoma. For glaucoma filtration surgery, one study reported that there was no statistically significant increase in retinal arterial oxygen levels after surgery, whereas venous oxygen saturation remained unchanged [67]. In contrast, another study reported an increase in retinal venous oxygen saturation after surgery, which may be due to improved blood perfusion following the procedure [68].

In summary, retinal oximetry shows promise as a research tool for assessing the severity of glaucomatous lesions. It may serve as a potential biomarker to support early intervention strategies aimed at preserving visual acuity and improving quality of life in the future. However, most current studies are cross-sectional and do not fully elucidate the causal relationship and intrinsic mechanism between retinal oximetry and glaucoma, warranting further investigation.

RAO/RVO and retinal oximetry

Retinal artery occlusion (RAO) is an acute-onset ocular disease that can threaten vision. It occurs when artery occlusion affects perfusion and the oxygen supply to corresponding regions, resulting in tissue damage and vision loss or even blindness; the pathological mechanisms include embolism, vascular occlusion and vascular compression [69]. Retinal vein occlusion (RVO) is the second most common retinal vascular disease after DR and can also cause varying degrees of visual impairment. A recent review examined the clinical significance of retinal oximetry in RVO [70].

Some researchers have investigated retinal oxygen saturation in patients with RAO. They measured the oxygen saturation of the central and branch arteries before and after rheological therapy. The results showed that the oxygen saturation of the occluded branch arteries partially recovered after treatment, indicating the effectiveness of the therapy. Furthermore, venous oxygen saturation increased posttreatment, suggesting improved perfusion [47].

In the case of central retinal vein occlusion (CRVO) and branch retinal vein occlusion (BRVO), several studies have compared the oxygen saturation levels in the obstructed vessels of the affected eye, the unobstructed vessels of the same eye, and the vessels of the contralateral normal eye. In patients with CRVO, the venous oxygen saturation in the affected eye was lower than that in the contralateral normal eye. In addition, in patients with BRVO, the oxygen saturation of the small obstructed veins varies considerably among individuals. Some patients present with reduced venous oxygen saturation, whereas others present with normal or even elevated levels. These studies concluded that the most likely cause of reduced venous oxygen saturation is a decrease in retinal blood flow due to occlusion, which in turn limits oxygen delivery to the tissues. Although venous saturation levels may differ among individuals, this may be attributed to factors such as the severity of various diseases, the degree of vascular occlusion, tissue atrophy, recanalisation and collateral circulation, arteriovenous diffusion or vitreous oxygen transport [71, 72]. However, further studies are needed to determine the associations between different saturation values and clinical presentation and prognosis.

Other ocular diseases and retinal oximetry

Age-related macular degeneration (AMD) is a disease that requires intensive study. Some evidence links ischaemia and hypoxia, particularly reduced choroidal blood flow, to the development of exudative AMD. Key factors in this process include vascular endothelial growth factor (VEGF), hypoxia-inducible factor (HIF), vitreous wart accumulation, and vitreoretinal adhesions [73]. With age, retinal venous oxygen saturation increases and arteriovenous oxygen saturation differences decrease in patients with exudative AMD, whereas the opposite is true in healthy individuals [74].

Retinitis pigmentosa (RP) is an inherited disease that causes slow and progressive degeneration of photoreceptors. This leads to night blindness, constriction of the visual field, and ultimately, loss of vision. This disease is associated with alterations in retinal microcirculation, including increased retinal venous oxygen saturation and significant narrowing of the arteriovenous diameters. These changes are significantly correlated with the preserved visual field area [75–77] and exhibit an age-related pathogenesis [78]. The increase in retinal venous oxygen saturation may be due to reduced oxygen demand resulting from retinal atrophy in patients with RP. These findings suggest that reduced oxygen delivery and retinal blood flow may be a consequence of the disease rather than its primary cause [75].

Cataracts, a prevalent ophthalmic concern worldwide, may indirectly affect retinal oximetry results. Following cataract-related surgery, patients typically show a notable increase in mean oxygen saturation levels in retinal arterioles and veins, accompanied by a pronounced reduction in arteriovenous differences. Although the degree of change was found to correlate with the preoperative lens density grade [79, 80], it is important to note that these changes do not necessarily reflect improved fundus perfusion. The reduction in arteriovenous oxygen saturation difference may largely be attributed to changes in image quality, which is a significant confounding factor in retinal oximetry. The effect of cataract surgery on oximetry results is likely influenced by image quality improvements or potential treatment of other comorbidities, such as the epiretinal membrane (ERM) or idiopathic macular hole (IMH), rather than an actual improvement in perfusion.

Regarding the relationship between refractive error and retinal oximetry, most studies have focused on the effects of high myopia. In general, myopic patients have reduced retinal arterial oxygen saturation and narrower vessel diameters, with the degree of change significantly correlated with the spherical equivalent (SE) and axial length (AL). These changes involve mechanisms such as retinal thinning due to ocular elongation, retinal and choroidal degeneration and atrophy, and optical factors [81–83]. Importantly, these findings appear to contradict those of previous studies [37, 38]. It was hypothesised that the impact of age on oxygen saturation may be more pronounced during childhood growth and development. Additionally, a study showed that implantation of an intraocular collamer lens (ICL) did not have a lasting effect on retinal oxygen saturation in patients with high myopia, indicating the safety and effectiveness of the surgery [84].

Furthermore, several studies have explored the associations between retinal oximetry and amblyopia [85] and Graves'ophthalmopathy [86]. Additionally, the use of hyperspectral image mapping spectrometry (IMS), a technique with a higher spatial and spectral sampling rate, has been investigated for the creation of oxygen saturation maps in a variety of ocular diseases [87]. Please refer to Supplementary Table 3 for further information.

Application of retinal oximetry in systemic diseases

Retinal oximetry offers a unique opportunity to study the connection between ocular and systemic health. By analysing the oxygenation status of the retina, it is possible to gain insights into systemic conditions such as cardiovascular conditions [88] and neurological disorders [89]. As a noninvasive and reproducible measurement tool, retinal oximetry has shown potential value in exploring these relationships and advancing our understanding of systemic diseases.

Cardiovascular diseases and retinal oximetry

Cardiovascular diseases significantly impact the retinal microcirculation, reflecting systemic changes in blood flow and oxygenation. For example, studies on patients with unilateral internal carotid artery stenosis (ICAS) have demonstrated significantly greater retinal arterial oxygen saturation and arteriovenous differences in the eyes of the stenotic side than in the eyes of the contralateral side and healthy controls. Additionally, the arterial diameters were smaller than those in healthy subjects, indicating microcirculation disorders [88].

On the other hand, there was a significant correlation between retinal venous oxygen saturation and mixed venous oxygen saturation, as well as between the retinal arteriovenous difference and the inverse of cardiac output in patients following right heart catheterisation [90]. These findings reiterate the potential of retinal oximetry as a noninvasive, convenient, and more accurate reflection of systemic blood oxygenation.

Neurological diseases and retinal oximetry

Some neurological diseases, such as multiple sclerosis, Parkinson's disease (PD) and Alzheimer's disease (AD), have been shown to involve abnormalities in retinal nerve fibres or blood vessels. Some researchers have reported that retinal arterial and venous oxygen saturation are significantly greater in patients with moderate AD than in normal subjects. Additionally, changes in retinal oxygen saturation in patients with AD are not associated with ageing [89]. However, the biological basis of these findings remains unclear. Inflammatory responses, oxidative stress, and endothelial cell dysfunction may contribute to disturbances in retinal microcirculation, including alterations in retinal oximetry. These mechanisms are also present in ocular diseases such as DR and AMD, suggesting a potential link between these diseases and AD [91, 92].

Pulmonary diseases and retinal oximetry

Certain pulmonary diseases can impair gas exchange, leading to hypoxia and affecting the oxygen status in the retinal microcirculation. Studies have demonstrated that patients with chronic obstructive pulmonary disease (COPD) have significantly lower retinal arterial and venous oxygen saturation than healthy controls do, with similar arteriovenous differences. When patients with COPD underwent oxygen therapy, there was a significant increase in retinal arterial oxygen saturation. Furthermore, it is worth noting that retinal oximetry measurements tend to be lower than finger oximetry and arterial blood gas analyses are, although they are still correlated. This discrepancy may result from small retinal arterioles reducing oxygen saturation through diffusion, countercurrent exchange, and other mechanisms [93, 94]. Additionally, a detailed review described the structural and functional changes in the retinal microvasculature in patients with COPD [95].

Kidney diseases and retinal oximetry

The kidney and eye share a number of similarities in their embryonic development, structure, physiology, and certain pathogenic mechanisms. For example, both the glomerulus and the choroid have similarly structured microvascular networks, the inner retinal and glomerular filtration barriers have similar developmental pathways, and the renin‒angiotensin‒aldosterone cascade is present in both the kidney and eye [96]. The progression of chronic kidney disease (CKD) is often driven by intrarenal hypoxia and inadequate renal perfusion [97]. These pathological changes can also manifest in the retinal microcirculation, particularly in cases where CKD is caused by systemic conditions such as diabetes and hypertension. Importantly, however, CKD encompasses a broad spectrum of aetiologies, including infections and inflammatory conditions, which may not directly affect the retina.

Studies have shown that patients with CKD exhibit significantly increased retinal arterial oxygen saturation and arteriovenous differences, as well as a smaller retinal arterial diameter, than healthy subjects do. Additionally, a negative correlation was observed between retinal arterial oxygen saturation and the estimated glomerular filtration rate (eGFR), and a positive correlation was observed between retinal arterial diameter and the eGFR. While CKD can be challenging to detect in its early stages, changes in retinal oximetry may provide an accessible window into the condition, offering potential for earlier diagnosis and intervention [98].

High-risk pregnancy and retinal oximetry

Several studies have examined alterations in retinal microcirculation during pregnancy. When comparing retinal vascular parameters—such as flicker-induced vasodilation in retinal arteries (FLA) and veins (FLV), central retinal arterial and vein equivalent (CRAE, CRVE), and retinal arterial and venous oxygen saturation—among high-risk pregnant women, low-risk pregnant women and healthy nonpregnant women, significant alterations were observed in high-risk pregnant women, especially in the third trimester (TP3) [99]. These parameters could serve as biomarkers for pregnancy complications and aid in assessing the risk to maternal health. For further details, please refer to Table 1. Furthermore, Table 2 summarises the general trends in retinal oxygen saturation across various ocular and systemic diseases.

Table 2.

Retinal oxygen saturation trends in various diseases (excluding surgical and drug effects)

Diseases	SO2 trends*
	Artery	Vein	AVD
DR	↑	↑	↓
Glaucoma	↓	↑	↓
CRVO		↓
AMD		↑	↓
RP		↑
High myopia	↓
Amblyopic	↑	↑
ICAS	↑		↑
AD	↑	↑
COPD	↓	↓
CKD	↑		↑
High-risk pregnancy		↑

*Refers to general trends in the studies mentioned, with a few possible exceptions. ↑ = increase, ↓ = decrease

SO₂ oxygen saturation, AVD arteriovenous difference, DR diabetic retinopathy, CRVO central retinal vein occlusion, AMD age-related macular degeneration, RP retinitis pigmentosa, ICAS internal carotid artery stenosis, AD Alzheimer's disease, COPD chronic obstructive pulmonary disease, CKD chronic kidney disease

'Systemic diseases from the eye'has become a recent research focus in ophthalmology. The correlation between ocular manifestations and systemic status highlights the interconnectivity of human organs and systems. To fully understand this correlation, it is important to explore the micropathophysiological mechanisms underlying these phenomena, in addition to their clinical applications.

Discussion

Retinal oximetry has gained significant attention in recent years as a biomarker that provides critical information about retinal oxygen saturation, which, when combined with blood flow measurements, can help assess the balance between oxygen supply and demand. This noninvasive, rapid, and convenient method enables simultaneous measurement of oxygen saturation in retinal arterioles and veins.

Owing to its calibration and algorithmic optimisation, the repeatability and stability of retinal oximetry have the potential to exceed those of many diagnostic measurements currently used in clinical practice. This stability is further attributed to the inherent stability of chemical levels in the central nervous system, which are regulated to maintain constant levels of calcium, potassium, pH, and oxygen. Additionally, the application of sound physics and advanced optical technology contributes to the reliability of retinal oximetry [2].

Retinal oximetry has been applied in some studies of healthy individuals, ocular diseases, and some systemic diseases. (1) For healthy individuals, researchers have extensively studied the range of baseline values of retinal oximetry. These findings demonstrate that this value is influenced by a variety of factors, including age, the quadrant in which the vessels are located, diurnal rhythms, light exposure, and altitude, as previously described [10, 33–44]. These diverse data are suitable for meeting the developmental demands of the big data era. They can aid in constructing a standardised database that can differentiate between healthy and diseased status and provide references for understanding the mechanisms of disease. (2) In the case of ocular diseases, monitoring changes in retinal oxygen saturation may offer valuable insights into the severity of the disease and the effectiveness of therapeutic interventions, potentially facilitating long-term disease management. (3) Observing the development of systemic diseases from the state of microcirculation of the eye reflects the coordination and unity of various organs and systems of the human body. This research field holds significant potential but also presents challenges.

On the other hand, the potential for clinical application of retinal oximetry is currently limited by technological constraints. (1) The accuracy of retinal oximetry measurements is limited by many factors, including vessel diameter, retinal pigmentation, oximetry calibration methods and image processing algorithms [4]. (2) Retinal oximetry is based on optical technology and is therefore also affected by the transparency of the optical medium, especially in retinal oximetry studies in cataract patients [79, 80]. When comparing oximetry values, it is important to consider the effect of image quality. (3) Pupil dilation is required prior to photographing with Oxymap and Imedos oximeters. Research has demonstrated that light affects retinal oximetry [41], but it is currently unclear whether pupil dilation impacts the final measurements. In addition, pupil dilation requires some time for the eye drops to take effect. Although it does not significantly affect the time needed for the oximetry measurement itself, it may complicate data collection by increasing the overall procedure time or reducing patient cooperation.

Beyond technical limitations, the transition of retinal oximetry from research to clinical practice faces another critical challenge: establishing its prognostic value and diagnostic discrimination capacity. While numerous studies have demonstrated retinal oxygen saturation alterations in various diseases, the evidence remains insufficient to confirm whether these changes can reliably predict disease progression or therapeutic outcomes. Moreover, the ability to differentiate between healthy and diseased populations through retinal oximetry has not been consistently demonstrated across studies, highlighting the need for more robust validation of its diagnostic stratification potential. This represents a fundamental gap that must be addressed through large-scale longitudinal studies with standardized protocols, which should aim to establish clear cutoff values and correlate oximetry parameters with definitive clinical endpoints.

In studies related to retinal oximetry, there are prevalent limitations that need to be addressed. (1) Most studies have not explored parameters related to fundus blood flow. Retinal oxygen consumption should have a functional relationship with oxygen saturation and blood flow, and blood flow is related to vessel diameter. Therefore, it is important to pay attention to haemodynamic changes simultaneously [33]. (2) In the elderly population, various systemic or ocular diseases may be present, which could affect fundus images and interfere with the determination of retinal oxygen saturation. Therefore, it is important to adequately control for confounding factors in the study design. (3) Most of the studies had small sample sizes and cross-sectional designs. To verify the generalisability of the results, larger sample sizes and long-term longitudinal studies may be necessary.

Several important directions for the future development of this field can be foreseen. (1) The application of wide-field technology is necessary. Currently, retinal oximeters on the market are equipped with modules on narrow-field fundus cameras, which can observe only a small retinal area. If retinal oxygen saturation data can be collected on a wide-field fundus camera, it will have diagnostic value for some diseases in the peripheral retinal region. Furthermore, measuring blood flow at all levels of the retina and choroid could expand the potential applications of this technology. However, this would require additional technological advancements [8]. (2) The possibility of performing nonmydriatic technology. As previously noted, pupil dilation may impact the results and reduce subject cooperation. The use of undilated oximetry may provide a more convenient and patient-friendly alternative while ensuring accurate measurements. (3) Blood flow measurement technology can be combined to provide a more comprehensive understanding of retinal oxygen metabolism and microcirculation, as oxygen saturation alone is not sufficient. Combining retinal oximetry with blood flow measurement techniques, such as OCTA and Doppler flowmetry, can provide a more comprehensive and specific assessment of retinal microcirculation. This information can aid in understanding the intrinsic mechanisms of retinal diseases [100]. (4) Artificial intelligence (AI) technology can be applied. Researchers can utilise various advanced deep learning models to enhance oximetry algorithms and improve the accuracy of results. AI is also proficient in automating the repetitive processes involved in vessel segmentation and oximetry calculations. It has also been utilised in the field of fundus image recognition and analysis, which can significantly improve examination efficiency and assist doctors in making diagnoses.

In summary, retinal oximetry has emerged as a new perspective for understanding ocular and systemic diseases over the past few decades. Commercially available devices have provided a wealth of data in healthy individuals and in various disease states. Conventional dual-wavelength measurement techniques have been continually improved and combined with more advanced technologies to drive further advances in retinal oximetry. Further work is needed to demonstrate the correlation between retinal oximetry and disease progression. This will help to explore its potential for diagnostic, therapeutic, prognostic, and even predictive applications and to promote retinal oximetry as a standardised and commonly used clinical measurement.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Weixing Zhang and Haotian Lin made substantial contributions to the conception or design of the work; Weixing Zhang and Xueer Tu performed the literature search and data analysis; Weixing Zhang drafted the manuscript; All authors critically revised the work, approved the final version for publication, and agree to be accountable for all aspects of the work. Weixing Zhang and Xueer Tu contribute to this work equally.

Funding

This study is Funded by the Science and Technology Planning Project of Guangzhou City (2024B03J1233), the National Natural Science Foundation of China (92368205), the National Natural Science Foundation of China (82441003), the Science and Technology Planning Project of Guangzhou City (2024A04J4474), and the Basic scientific research projects of Sun Yat-sen University (23ykcxqt002).

Data availability

Not data are available.

Declarations

Ethical approval

This article does not contain any studies with human participants performed by any of the authors.

Informed consent

Informed consent was not required, as there were no human participants.

Conflict of interest

The authors have no financial or proprietary interest in the materials presented herein.