Hyperspectral Image Analysis for Oxygen Saturation Automated Localization of the Eye

Fig. 4.22. Reproducibility of saturation maps obtained in the same monkey at 10 mmHg from separate HSI recordings on the same day. The position of the ONH has been adjusted to the center of each image. Low-to-high saturation is indicated by the progression through blue–cyan–green–yellow–red. Nasal- to-temporal orientation is left-to-right and inferior-to-superior orientation is top to bottom. Reprinted with permission from Khoobehi, B., Kawano, H., Ning, J., Burgoyne, C.F., Rice, D.A., Khan, F., Thompson, H.W., and Beach, J.M. Oxygen saturation changes in the optic nerve head during acute intraocular pressure elevation in monkeys. In: Manns, F., Soderberg, P.G., and Ho, A. (eds.), Ophthalmic Technologies XIX, Proc of SPIE. 7163, 716320. ©2009 SPIE.

The oxygen saturation maps of ONH structures are shown in Fig. 4.23 for increasing IOP. At 30 mmHg, the veins and the ONH areas show slightly decreased saturation, while at 45 mmHg the ONH tissue saturation is substantially reduced. At 55 mmHg, a reduction in the artery is most marked and ONH tissue saturation is further reduced in the rim. The effect of IOP on the oxygen saturations visualized in the saturation maps is in agreement with the percentage saturations in Table 4.5. A linear relationship between numerical values of the saturation map and percentage saturation values obtained by our procedure was established previously.73 No differences in oxygen saturation levels were seen at 10 and 30 min after IOP level adjustment at any of the IOPs evaluated.

4.4.3. Discussion

Percentage oxygen saturation in retinal vessels has previously been reported using vessel oximetry in humans.74,75 Our values from the monkey retinal vessels at the normal IOP are 10–15 percentage points lower than those that had been previously reported. In the present study, deep anesthesia was needed to maintain a stable eye during HSI collection and IOP elevation. It is likely that under deep anesthesia, systemic saturation could have been lower because of less efficient respiration. As with vessels, the tissue values we report may also be lower than would otherwise be found in animals that were awake.

165

Bahram Khoobehi and James M. Beach

We previously reported the changes in the oxygen saturation distribution that resulted from experimentally raising IOP from a normal pressure (15 mmHg) to near the perfusion pressure (60 mmHg) for several minutes.33 Sustained high IOP produced dramatic reductions in blood saturation from the ONH tissue and overlying retinal arteries and veins, overriding the autoregulatory control of blood flow in all of these structures. Autoregulatory responses were observed, however, from the saturation rebound at the cup of the ONH. In the present study, we were interested in determining thresholds for the loss of autoregulation in these different structures resulting from the reduced oxygen saturation that occurred in response to graded increases in IOP.

The findings of this study in vessels were (1) elevation of the IOP to 30–45 mmHg did not result in a significant change in the retinal arterial saturation, whereas further elevation to 55 mmHg caused the saturation of arterial blood to decrease significantly; (2) the same stepwise increases in IOP caused approximately equal reductions in the retinal venous saturation at each step above normal IOP; and (3) between 10 and 45 mmHg where autoregulatory responses were effective, the overall decrease in saturation was larger in the veins than in the arteries, causing the retinal arterio-venous saturation difference (A-V difference) to increase with IOP. The additional oxygen extraction from the vein is consistent with a reduced flow resulting from dropping perfusion pressure. The retinal venous saturation is closely tied to the end-capillary blood saturation of the tissue microcirculation; hence, these responses are consistent with compensatory autoregulation of retinal tissue oxygen saturation at elevated IOP, up to 45 mmHg in our experimental model. A blood flow proportional to the perfusion pressure at high IOP was shown in the retinal and prelaminar ONH in monkeys76; in our study, the perfusion pressure dropped to single digits, which would indicate that the blood flow was significantly reduced.

The mechanism by which the retinal arterial saturation is decreased at reduced flow is not known. However, the reports of significant longitudinal gradients in the periarteriolar PO2 in small arteries and arterioles, with essentially no difference between the intraand extravascular tissue oxygen tensions, have been explained by diffusion of oxygen from these precapillary vessels.77 More recently, spectral imaging of the saturation distributions in rat-cremaster vessel networks demonstrated oxygen transfer from arteries to

166

Hyperspectral Image Analysis for Oxygen Saturation Automated Localization of the Eye

veins running in parallel, or at crossover points.78 Thus, there is an opportunity for oxygen exchange between the central RA and vein, which run together for several millimeters in close proximity, within the distance for oxygen exchange by diffusion, before their entrance from the ONH cup. If oxygen were removed by leakage at a fixed rate, lower volume flow would result in a decrease in the PO2 of the blood supply at the inner retinal arteries. As the dissolved oxygen concentration is reduced, arterial saturation must decrease according to the oxygen dissociation curve. Therefore, the stable relationship we observed between arterial saturation and IOP at 10 and 30 mmHg may reflect autoregulatory control, whereas the nonlinear relationship over the higher range of IOP may, as we speculate, reflect passive flow-dependent mechanisms that could involve oxygen exchange from central RA to retinal vein. There is presently no evidence for this, however.

A relationship between saturation and IOP, which is similar to that found for retinal arteries and veins, was observed in the ONH. However, the blood supply in the ONH is more complex. Our recordings in the ONH rim yielded results similar to those in retinal arteries, showing a stable saturation of the blood supply over the first step in IOP, and reduced saturation at higher IOP. This response is consistent with previous measurements of blood flow changes during acute elevations in IOP in humans.79 Over the lower range of elevated IOP, the absence of change in saturation at 10 and 30 min after raising the IOP suggests that autoregulation of flow in the ONH rim was present. Reduced oxygen saturation in the higher range of elevated IOP is consistent with reduced blood saturation in the blood supply to this tissue. Since this region receives its blood supply via peripapillary retinal arterioles, which are supplied through the central RA, those saturation responses that were observed in the RA could be also present at the ONH rim. The prelaminar blood supply has been shown to be masked by vessels in the NFL in early arterial phases of fluorescein angiograms in monkeys.80 Hence, it is likely that our observations of blood saturation in the ONH rim by reflectometry are confined to the NFL, which shares its blood supply with the peripapillary retina.

Within the ONH cup, particularly the temporal aspect, which is central to the spread of the NFL, we observed a steady series of smaller reductions in the blood saturation at each step above normal IOP, similar to the finding in the retinal vein. Since the NFL circulation does not mask the central

167

Bahram Khoobehi and James M. Beach

Fig. 4.23. ONH saturation maps for IOP values of 10 mmHg (top-left), 30 mmHg (top-right), 45 mmHg (bottom-left), and 55 mmHg (bottom-right). Low-to-high saturation is indicated by the progression through blue–cyan–green–yellow–red. Nasal-to-temporal orientation is left-to-right and inferior-to-superior orientation is top to bottom. Reprinted with permission from Khoobehi, B., Kawano, H., Ning, J., Burgoyne, C.F., Rice, D.A., Khan, F., Thompson, H.W., and Beach, J.M. Oxygen saturation changes in the optic nerve head during acute intraocular pressure elevation in monkeys. In: Manns, F., Soderberg, P.G., and Ho, A. (eds.), Ophthalmic Technologies XIX, Proc of SPIE. 7163, 716320. ©2009 SPIE.

cup, the deeper layers of the ONH, including the prelaminar and laminar regions, would be accessible to optical reflectometry at the central cup. The short posterior ciliary arteries and branches from the pial artery supply the blood flow to these layers of the optic nerve; little flow is derived from the central RA.3,4 Fluorescein fundus angiography has also demonstrated a blood supply to these regions from the peripapillary choroid.44 Thus, saturation changes in the temporal cup would not be expected to follow those seen in the rim. Although the relationship between cup saturation and IOP is similar to that of the retinal vein, it may be the result of redistribution of flow from the prelaminar region into the retrolaminar region behind the lamina cribrosa.81 Our recordings of the oxygen saturation levels of

168