1.

Introduction

The eye is a critical light-sensitive organ¹ divided into anterior and posterior segments.² The anterior segment includes the cornea, iris, pupil, and ciliary body. The posterior segment includes the vitreous body, retina, choroid, and optic nerve.³ The retinal pigment epithelium mostly contains melanin⁴ and plays a major role in energy and oxygen consumption.⁵ Retinal and choroidal circulation delivers oxygen to the inner retinal tissue and photoreceptors, respectively.⁶ Focal retinal hypoxia is thought to underlie the pathophysiology of many ocular diseases, such as diabetic retinopathy,^7,8 glaucoma,^9,10 and retinal venous occlusion.¹¹ Thus, the ability to image retinal oxygen gradients could be very valuable in treating and diagnosing retinal disease. Also, chorioretinal vasculature imaging has a significant role for diseases, such as neovascular age-related macular degeneration (AMD).¹²

There are four requirements for ocular imaging. First, motion artifacts and image distortion need to be compensated either through high-speed imaging or eye tracking software. Second, imaging should be noninvasive to decrease the risk of side effects. Third, light illumination should be below safety limits to prevent damaging the retina.¹³ Fourth, resolution is an important element of all imaging techniques. However, there are some applications where low-resolution imaging techniques are useful, such as magnetic resonance imaging¹⁴ and ultrasound.¹⁵ Current clinical ophthalmology imaging modalities such as optical coherence tomography (OCT),^16,17 confocal scanning laser ophthalmoscopy,^18,19 and fundus photography²⁰ provide anatomic data on the retina but do not measure oxygen gradients. Hyperspectral imaging^21,22 has shown potential for imaging oxygen gradients in the eye but still has not been systematically validated, and an artifact of spectral noise from pigmentation, ocular media, etc. can affect hemoglobin oxygen gradients.

More recently, photoacoustic imaging (PAI) has been described as a noninvasive and nonionizing imaging technique that combines optical absorption and ultrasound detection.^23–28 Photoacoustic signal is generated due to absorption and thermal expansion of endogenous or exogenous molecules using a nanosecond laser pulse followed by detection with a wideband ultrasound transducer.^29–31 This modality offers high penetration depth due to its weak ultrasound scattering in biological tissue. Oxyhemoglobin and deoxyhemoglobin (HbR and ${\mathrm{HbO}}_2$) as well as melanin are the main optical absorbers in tissue. PAI can measure oxygen saturation (${\mathrm{sO}}_2$) variations based on differences in HbR and ${\mathrm{HbO}}_2$ absorption.^32–35 In eye diseases, oxygenation is dysregulated due to increased intraocular pressure (IOP)—the choroidal blood flow decreases as the IOP increases.³⁶ In turn, oxygen availability decreases with decreasing $\mathrm{CR}-{\mathrm{sO}}_2$.³⁶ Although PAI is ideally suited to measure oxygen saturation, somewhat surprisingly, most studies in the field have focused only on anatomic imaging.

Examples of anatomic imaging include de La Zerda et al.³⁷ who utilized PAI in swine and rabbit eyes. Their system could image both the anterior and posterior segments of the ocular system. However, 90 min was required to image a $12\text{-}\mathrm{mm}\times 8\text{-}\mathrm{mm}$ area. Hu et al.³⁸ utilized optical resolution photoacoustic microscopy (OR-PAM) to do in vivo imaging of the anterior segment of a mouse eye; axial and lateral resolution of 15 and $5\mu \mathrm{m}$ were reported, but 120 min was required to cover only a $2\text{-}\mathrm{mm}\times 2\text{-}\mathrm{mm}$ area. Liu et al.³⁹ and Wu et al.⁴⁰ improved their OR-PAM system’s acquisition time to 20 min for a $\text{3-}\mathrm{mm}\times 3\text{-}\mathrm{mm}$-image and 6.5 min for a $2\text{-}\mathrm{mm}\times 2\text{-}\mathrm{mm}$-image, respectively. Jiao et al.,⁴¹ Song et al.,^42,43 and Liu et al.⁴⁴ proposed multimodal ocular imaging using PAI with an optical scanning method and decreased the acquisition time to 2.7 s for a $2\text{-}\mathrm{mm}\times 2\text{-}\mathrm{mm}$ image. Recently, Tian et al.⁴⁵ designed chorioretinal imaging of integrated PAM and OCT for rabbits. The acquisition time for a $3\text{-}\mathrm{mm}\times 3\text{-}\mathrm{mm}$ area was 65 s with PAM lateral and axial resolution values of 4.1 and $37\mu \mathrm{m}$, respectively.

Other groups have used microscopy techniques to image oxygen saturation in rodent eyes. Although these techniques offered very high spatial resolution, the penetration depth was limited to a few mm, and the technique required many minutes to map only a few mm² of tissue. An example of this includes Song et al.⁴³ who quantified retinal oxygen saturation in rats. They combined OCT and PAM to measure the retinal oxygenation.

However, to the best of our knowledge, only one study has yet evaluated oxygen saturation in a large animal eye using PAI. In that study, Hennen et al.⁴⁶ used acoustic resolution photoacoustic tomography (AR-PAT) to image the conjunctiva and other areas on the “anterior” of rabbit eyes with 2-Hz temporal resolution. However, a remaining fundamental limitation is oxygen saturation imaging of the “posterior” (chorioretina) of a large animal eye. This has not yet been reported. Indeed, as Hennen and coworkers⁴⁶ stated, such a technique would be useful for evaluating the role of oxidative damage, hypoxia, and ischemia in pathogenesis of various ocular diseases, such as glaucoma, diabetic retinopathy, AMD, and cataract.

Here, we describe high-speed photoacoustic ocular imaging (PAOI) to detect chorioretinal oxygen gradient in an in vivo model of hypoxia as well as an ischemia reperfusion model on albino live rabbits. This PAOI approach is faster and has a simpler optical design than acoustic resolution photoacoustic tomography and PAM techniques. PAOI could monitor $\mathrm{CR}-{\mathrm{sO}}_2$ changes in real time as demonstrated with oxygen tension experiments and ischemia-reperfusion models.

2.

Methods and Materials

3.

Results

Figure 1 shows the experimental setups in this study. Figure 1(a) shows the oxygen tension experiment, and Fig. 1(b) shows the ischemia-reperfusion setup.

Fig. 1

Illustration of experiment systems. (a) Oxygen tension experiments. Different percentages of oxygen [5%, 10%, 15%, 21% (air), 30%, and 100%] with the balance as nitrogen were applied to an anesthetized rabbit. A pulse oximeter and PAOI measured ${\mathrm{sO}}_2$ and $\mathrm{CR}-{\mathrm{sO}}_2$, respectively. (b) Ischemia-reperfusion model experiment. A 25-gauge needle was placed in anterior segment of the eye, and a syringe pump controlled the amount of PBS injected into the eye. The direction of light propagation explains the limited field-of-view on the retina surface.

3.1.

Oxygen Tension Experiment

We first validated whether PAOI could measure ${\mathrm{sO}}_2$ in the eye by comparing $\mathrm{PAOI}-\mathrm{CR}-{\mathrm{sO}}_2$ measurements and pulse oximetry measurements in animals breathing different percentages of oxygen. Figure 2(a) shows the B-mode photoacoustic/ultrasound image of a rabbit eye at 750 nm. The PAOI field of view is limited—only part of the chorioretina (40% of whole chorioretina) can be imaged using PAOI. This is a result of the light propagation [Fig. 1(b)] and can be improved by tilting the transducer.⁵⁰ The acquisition time for one B-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map is 0.8 frame per second. We see a very strong correlation ($R^2=98$, $n=5$) between PAOI and pulse oximetry as reference for ${\mathrm{sO}}_2$ measurement. This result validates the PAOI and oximetry methods.

Fig. 2

Evaluation of the PAOI response to oxygen tensions. (a) B-mode photoacoustic/ultrasound image at baseline condition (breathing normal air) including the cornea, conjunctiva, lens, and chorioretina. The image depth is 20 mm. (b) The B-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map when the rabbit is given 5% ${\mathrm{O}}_2$ and 95% ${\mathrm{N}}_2$, (c) 10% ${\mathrm{O}}_2$ and 90% ${\mathrm{N}}_2$, (d) 15% ${\mathrm{O}}_2$ and 85% ${\mathrm{N}}_2$, (e) normal air, (f) 30% ${\mathrm{O}}_2$ and 70% ${\mathrm{N}}_2$, and (g) 100% ${\mathrm{O}}_2$. For PAOI, the $\mathrm{CR}-{\mathrm{sO}}_2$ measurement frame rate is 0.8 Hz. (h) PAOI—$\mathrm{CR}-{\mathrm{sO}}_2$ measurements show linear dependence ($R^2=0.98$, $n=5$) with pulse oximeter measurements at different oxygen tensions. The error bars show the standard deviation of different five different eyes. All scale bars represent 5 mm.

Also, see in Fig. 3 when one animal transitioning from room air to 100% ${\mathrm{N}}_2$ and the accompanying change in $\mathrm{CR}-{\mathrm{sO}}_2$.

Fig. 3

Still frame from a video showing oxygen tension experiment for one animal transitioning from room air to 100% ${\mathrm{N}}_2$ and the accompanying change in $\mathrm{CR}-{\mathrm{sO}}_2$ (Video 1, mp4, 785 KB [URL: https://doi.org/10.1117/1.JBO.23.3.036005.1]).

Next, we further validated the kinetics of measurement. Figure 4(a) shows synchronization between pulse oximetry and PAOI. The temporal resolution of PAOI was 0.8 Hz, and the refresh rate on the pulse oximeter was 1 Hz. We also validated the reproducibility across 100 B-scans and created 20-mm MIP maps of $\mathrm{CR}-{\mathrm{sO}}_2$ during hyperoxia and normal conditions [Figs. 4(b) and 4(c)]. The coefficient of variation across 100 B-scans was 1.45% and 2.15% for normal and hyperoxia conditions, respectively. The $\mathrm{CR}-{\mathrm{sO}}_2$ values measured with PAOI are significantly lower than the ${\mathrm{sO}}_2$ measurements with a pulse oximeter.

Fig. 4

Validation of real time measurements of $\mathrm{CR}-{\mathrm{sO}}_2$ using PAOI. (a) Real-time monitoring of ${\mathrm{sO}}_2$ using a pulse oximeter and PAOI in the normal and hyperoxic conditions. PAOI has a frame rate of 0.8 Hz. (b) MIP $\mathrm{CR}-{\mathrm{sO}}_2$ map hyperoxia and (c) MIP $\mathrm{CR}-{\mathrm{sO}}_2$ map during normal breathing condition. The scan size was 20 mm, and the scale bars represent 5 mm.

3.2.

Ischemia Reperfusion Model

We next evaluated the ability of PAOI to detect $\mathrm{CR}-{\mathrm{sO}}_2$ gradients in an ischemia reperfusion model [Fig. 5(a)]. Figures 5(b)–5(e) show the B-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map after injection of 100, 200, 300, and $400\mu \mathrm{L}$ PBS, respectively. We first validated the model by injecting increasing volumes of PBS and noted a proportional increase in IOP as validated with a Tonopen [Fig. 5(f)]. As IOP increases, we see a corresponding decrease in $\mathrm{CR}-{\mathrm{sO}}_2$ levels [Fig. 5(g)]. The injection of $400\text{-}\mu \mathrm{L}$ PBS resulted in a 4.8-fold increase in IOP and a 1.6-fold decrease in $\mathrm{R}-{\mathrm{sO}}_2$ as measured by Tonopen and PAOI, respectively.

Fig. 5

Investigation of relationship between PBS injected volume, IOP, and $\mathrm{CR}-{\mathrm{sO}}_2$. (a) A b-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map under baseline condition (no PBS injection-no IOP increase). A 25-gauge needle was placed inside the anterior segment of eye. The red dotted rectangle shows the field of view of PAOI on depth of retina. (b) B-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map after injection of $100\text{-}\mu \mathrm{L}$ PBS, (c) $200\text{-}\mu \mathrm{L}$ PBS, (d) $300\text{-}\mu \mathrm{L}$ PBS, and (e) $400\text{-}\mu \mathrm{L}$ PBS, (f) linear correlation between IOP and volume of injected PBS in eye ($n=5$ eyes; $R^2=0.98$). IOP was measured using a Tonopen, and injected PBS volume was controlled by syringe pump. The IOP increased by increasing the PBS volume inside the eye. (g) Correlation between $\mathrm{CR}-{\mathrm{sO}}_2$ and PBS volume inside the eye. Negative correlation is observed between these two parameters ($n=5$ eyes; $R^2=0.99$). The scale bar is 5 mm. *shows the $p<0.01$. Error bars show a standard deviation among the five different eyes.

We evaluated whether PAOI could measure changes in chorioretinal oxygenation in an ischemia reperfusion model. IOP was elevated to occlude ocular blood flow using injection of PBS into the anterior chamber. After elevated IOP, we see a steady decrease in $\mathrm{CR}-{\mathrm{sO}}_2$ over 50 s that reached steady state [Fig. 6(e)]. After decreasing the IOP with subsequent reperfusion, the $\mathrm{CR}-{\mathrm{sO}}_2$ increased—close but not statistically identical to—the baseline values ($n=5$, $p=0.03$) [Figs. 6(f) and 6(g)].

Fig. 6

Real-time evaluation of ischemia-reperfusion model using PAOI. (a) B-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map at baseline condition (no PBS injection-no IOP increase); a 25-gauge needle is inside the anterior segment of eye. Red and yellow dotted rectangles show the retina and control areas, respectively. The $\mathrm{CR}-{\mathrm{sO}}_2$ is $68.76\pm 2.34$ at baseline (no IOP increase; $10.5\pm 0.33\mathrm{mm}\mathrm{Hg}$). (b) The $\mathrm{CR}-{\mathrm{sO}}_2$ map 5 s after injection of PBS in the eye. Here, the change in the retina map to blue indicates a decrease in $\mathrm{CR}-{\mathrm{sO}}_2$ after 5 s of IOP. (c) $\mathrm{CR}-{\mathrm{sO}}_2$ gradients after injection of PBS (15 s after start of injection). The IOP increased to $46.5\pm 2.54\mathrm{mm}\mathrm{Hg}$, and the $\mathrm{CR}-{\mathrm{sO}}_2$ decreased to $12.3\pm 3.54$. (d) $\mathrm{CR}-{\mathrm{sO}}_2$ map in reperfusion condition. The imaging plane is different from other three because we had to remove the transducer and measure the IOP with a Tonopen. $\mathrm{CR}-{\mathrm{sO}}_2$ recovered to $58.87\pm 3.23$ after the IOP returned to $16.55\pm 1.67\mathrm{mm}\mathrm{Hg}$. (e) Real-time monitoring of $\mathrm{CR}-{\mathrm{sO}}_2$ and ${\mathrm{sO}}_2$ using PAOI and pulse oximetry. The $\mathrm{\Delta }$ shows ${\mathrm{sO}}_2$ difference between baseline and ischemia. (f) Statistical analysis of $\mathrm{CR}-{\mathrm{sO}}_2$ for five eyes under different conditions including baseline, ischemia, and reperfusion using PAOI. The PAOI shows sixfold decrease in $\mathrm{CR}-{\mathrm{sO}}_2$ during ischemia ($n=5$). There was a very significant change between baseline and ischemia and a significant different in $\mathrm{CR}-{\mathrm{sO}}_2$ from baseline to reperfusion. (g) Statistical analysis of IOP for five eyes at baseline, ischemia, and reperfusion using Tonopen. There is negative correlation between IOP and $\mathrm{CR}-{\mathrm{sO}}_2$. Error bars shows the standard deviation of five different eyes. Here, * indicates $p<0.05$ and ** indicates $p<0.0001$.

Figure 6(a) is a B-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map that was imaged before the start of the ischemia experiment. Figures 6(b) and 6(c) show $\mathrm{CR}-{\mathrm{sO}}_2$ maps 5 and 15 s after the start of the injection of $600\text{-}\mu \mathrm{L}$ PBS in the anterior segment, respectively. Figure 6(c) shows an obvious decrease in blood oxygenation during reduced blood flow due to elevated pressure. This pressure was measured with a Tonopen both at baseline and after the injection of $600\text{-}\mu \mathrm{L}$ (ischemia). We then removed the needle from the eye and measured the IOP and $\mathrm{CR}-{\mathrm{sO}}_2$ 15 min later (reperfusion). Figure 6(d) shows the B-mode $\mathrm{CR}-{\mathrm{sO}}_2$ map in the reperfusion condition. The reperfusion map is not in the exact same plane as the previous condition, because we moved the PA/US transducer to measure the IOP and then repositioned it back onto the eye after the IOP measurement.

The PAOI showed a nearly sixfold decrease in $\mathrm{CR}-{\mathrm{sO}}_2$ postinjection of $600\mu \mathrm{L}$ ($n=5$, $p=0.00001$). The ${\mathrm{sO}}_2$ measurement for control areas (conjunctive) confirms that the ischemia only affects the ${\mathrm{sO}}_2$ in the chorioretina and not other parts of the eye. The IOP did not fully return to baseline after reperfusion. It remained elevated relative to baseline [$n=5$, $p=0.03$; Figs. 6(f) and 6(g)]. The $\mathrm{CR}-{\mathrm{sO}}_2$ and IOP are negatively correlated [Figs. 6(f) and 6(g)]. The $\mathrm{CR}-{\mathrm{sO}}_2$ decreased with increased IOP ($n=5$, $p=0.00002$). Figure 6(e) shows real-time monitoring of both the pulse oximeter and the PAOI during the ischemia-reperfusion experiment. As expected, the ischemia-reperfusion model did not affect ${\mathrm{sO}}_2$ measurements via the pulse oximeter on the animal’s paw.

Figure 7 presents the sequence of increase in IOP and subsequent decrease in $\mathrm{CR}-{\mathrm{sO}}_2$.

Fig. 7

Still frame from a video showing real-time evaluation of ischemia-reperfusion model using PAOI (Video 2, mp4, 1 MB [URL: https://doi.org/10.1117/1.JBO.23.3.036005.2]).

4.

Discussion

We demonstrate high speed, simple, and noninvasive PAI to provide spectral, spatial, and temporal data with sufficient resolution to visualize chorioretinal oxygen gradients. This is the first measurement of chorioretina oxygen gradients using an ischemia-reperfusion model and PAI. Rabbits with pigmented eyes (Dutch belted rabbit) were also evaluated using our approach. This strain has pigmented eyes, and we attempted to measure the chorioretinal oxygen gradients using PAOI. The retinal pigmented epithelium is a pigmented layer that is firmly attached to the underlying chorioretinal vessels. This layer contains melanin and absorbs light to generate strong photoacoustic signal. Therefore, we cannot measure the photoacoustic signals from the chorioretina in the pigmented rabbit eye. We used a commercial PAI system (Vevo LAZR, VisualSonics Inc.) and showed that this system is reliable and reproducible for measurement of $\mathrm{CR}-{\mathrm{sO}}_2$ with the frame rate of 0.8 Hz, which is significantly higher than those in the literature. Using this tool, we observed $\mathrm{CR}-{\mathrm{sO}}_2$ gradients both by changing oxygen tension as well as through an ischemia-reperfusion model. Our paper is the first to demonstrate the ability of PAOI to monitor oxygen level on the posterior segment of the eye over such a wide range of oxygen levels. Ketamine–xylazine can reduce oxygen saturation on room air.^51,52 In addition, we were using the pulse oximeter on the rabbit’s paw, which had much more hair than a human finger. This could result in an artificially low oxygenation saturation reading. However, during our initial calibration using oxygen/nitrogen mixtures (Fig. 2), the oxygen saturation reported using PAOI was consistently lower than that reported by the pulse. This difference has been reported in hyperspectral computed tomography system²² and AR-PAM⁴⁶ as well. One possible explanation for the difference is that while the peripheral pulse oximeter measures arterial ${\mathrm{sO}}_2$, the PAOI quantifies the $\mathrm{CR}-{\mathrm{sO}}_2$ using the average of both arterial and venous vessels.⁴⁶ Also, the large number of photons are absorbed and attenuated in the anterior segment of eye, such as iris and lens and only limited number of photons is delivered through the eye. Oxygen saturation measurement using PAOI depends on light fluence. The reason of having different numbers from pulse oximeter can be the change of fluence spectrum in depth.⁵³ Nevertheless, the PAOI $\mathrm{CR}-{\mathrm{sO}}_2$ measurements strongly correlated with systemic measurements of oxygen saturation (pulse oximeter).

One limitation of this approach is balancing spatial resolution and penetration depth. Rabbits’ retinal vessels are 20-mm deep. Therefore, we used a 15-MHz transducer with 580 and $290\mu \mathrm{m}$ as lateral and axial resolution, respectively. Although this could not discriminate individual vessels as in photoacoustics microscopy,^43,54 we could achieve much deeper penetration. An ideal technique would have good depth penetration and 50-micron resolution to determine if specific sections of the retina are hypoxic. This current PAOI approach cannot discriminate between the retina and the choroid, and the area of study includes both features.

Previous work²² applied ischemia-reperfusion model on a rabbit’s eye and measured the $\mathrm{CR}-{\mathrm{sO}}_2$ using hyperspectral computed tomography system. This work used an elevated reservoir of balanced salt solution (BSS) to increase the IOP, but there was no precise control on the amount of injected BSS and IOP in the eye. Here, we used a syringe pump to more precisely control the injected volume (Fig. 5). With this model, the $\mathrm{CR}-{\mathrm{sO}}_2$ decreased with increasing volume of injected PBS due to interrupted blood flow. In $\mathrm{CR}-{\mathrm{sO}}_2$ map, we found that 10 s after starting the PBS injection, blood flow to the chorioretinal rapidly dropped (Fig. 6). The $\mathrm{CR}-{\mathrm{sO}}_2$ during reperfusion was lower than baseline ($n=5$, $p=0.03$) likely because of compensatory mechanisms of repair in the eye.

Another limitation is the use of a Tonopen designed for human on rabbit subjects. Mermoud et al.⁵⁵ demonstrated that the Tonopen significantly underestimated the pressure from 5 to 80 mmHg. However, they showed high correlation between the transducer pressure and Tonopen as well. This means that the IOP numbers reported with the Tonopen on rabbits might be lower than the true IOP. The IOP numbers are likely 10 to 25 mmHg, because the Tonopen was used after the needle was removed from the eye after injection. Nevertheless, the good correlation seen here between injected volume of saline, Tonopen, and PAOI suggests that this model is indeed creating ischemia. Figure 6 shows changes in IOP (from baseline to $46.5\pm 2.54\mathrm{mm}\mathrm{Hg}$) with just one step injection of $600\mu \mathrm{L}$ PBS in the eye.

The maximum permissible exposure (MPE) from the American National Standard Institute⁵⁶ for 700- to 1050-nm laser light with 10-s illumination is $5.0{\mathrm{C}}_{\mathrm{A}}{\mathrm{C}}_{\mathrm{E}}\times {10}^{-7}\mathrm{J}/{\mathrm{cm}}^2$. Here, there is a 20-Hz repetition rate and a 5-ns pulse width. This results in $\sim 1\mu \mathrm{s}$ of actual laser pulse illumination time. With 700- to 1050-nm laser light and based on illumination specifications, $C_A={10}^{2(\lambda -0.700)}$ and $C_E=\alpha /({\alpha }_{\min }{\alpha }_{\max })$. Here, $\lambda$ and $\alpha$ are the radiation wavelength and the full angle retina exposure, respectively.⁵⁷ For $\lambda =750\mathrm{nm}$ and $\alpha =1100\text{miliradians}$, the MPE is $\sim 5.1\mathrm{mJ}/{\mathrm{cm}}^2$. Our approach used $\sim 4.5{\mathrm{mJ}/\mathrm{cm}}^2$. Hennen et al. used $5{\mathrm{mJ}/\mathrm{cm}}^2$ and showed no damage via histology one after imaging session. To decrease this risk, light-emitting diodes (LED) with much lower incident fluence could also be used.^58,59 We already characterized the LED-based PAI system in terms of penetration depth. Temporal resolution up to 30 Hz and signal-to-noise ratio of higher than 3 can be achieved for embedded pencil lead inside the chicken breast. We also evaluated application of LED system for monitoring chorioretina vessels on ex vivo rabbit eyes with frame rate of 15 Hz.⁵⁹

5.

Conclusion

In this paper, we describe a high-speed PAOI technique to monitor chorioretina oxygen gradients in vivo. This the first report of PAI applied to $\mathrm{CR}-{\mathrm{sO}}_2$ measurements of ischemia-reperfusion in a large animal eye. To the best of our knowledge, a frame rate of 0.8 Hz for measuring $\mathrm{CR}-{\mathrm{sO}}_2$ is the highest frame rate yet reported, allowing for dynamic measurements of chorioretinal oxygenation in real time. We demonstrate a very strong correlation between PAOI $\mathrm{CR}-{\mathrm{sO}}_2$ and pulse oximetry with varying levels of inspired oxygen, and we demonstrate a dramatic decrease in oxygen gradients during ocular ischemia. These experiments demonstrate that PAOI can serve as a dynamic measurement oxygen gradients in the eye and may be useful in the future for applications in ocular disease.