1.

Introduction

Much of the laser bioeffects database used to establish the ocular laser exposure guidelines has been obtained from in vivo measurements in nonhuman primate (NHP) eyes. Theoretically, the NHP eye is capable of focusing the collimated light of a laser beam to a spot on the retina of $4\text{to}5\mu \mathrm{m}$ in diameter. However, during laser retinal injury threshold studies, the minimum lesion size observed is much larger than this. Computer modeling¹ predicts that the energy required to produce retinal injury will decrease with decreasing retinal irradiance area (i.e., beam spot area) for beam spot diameters down to values approaching the assumed minimal visible lesion size, typically taken to be $20\text{to}30\mu \mathrm{m}$ . Experiments with bovine retinal explants, in which the anterior portions of the eye, including the neural retina, are removed so that the retinal pigmented epithelium (RPE) layer can be directly irradiated, show that indeed the energy required to damage the RPE does decrease with the diameter of the irradiated area for beam spot diameters of $20\mu \mathrm{m}$ and larger.^{2, 3}

The NHP data does not follow the trend predicted by the models and observed in the explant data. Figure 1 shows the retinal beam spot size dependence of the ${\mathrm{ED}}_{50}$ obtained from in vivo NHP measurements. The ${\mathrm{ED}}_{50}$ is defined as that dose having a 50% probability of producing a minimum visible lesion (MVL) on the retina. Note that in Fig. 1 and throughout this paper, the dose is given as the total intraocular energy (TIE), that energy incident at the cornea within the projected area of the ocular pupil, because this quantity can be directly measured for in vivo exposures. Figure 1 includes data from a number of investigations covering a range of laser wavelengths and exposure durations.^{4, 5, 6, 7, 8} The ${\mathrm{ED}}_{50}$ (TIE) in an intact NHP eye varies only weakly with spot size for retinal beam spot diameters smaller than $\sim 100\mu \mathrm{m}$ . Equivalently, the threshold retinal radiant exposure $H_R$ varies inversely as the square of the diameter for small retinal beam spots. For short-duration exposures, the threshold $H_R$ is constant, independent of the diameter of the irradiated area for retinal spot sizes greater than $\sim 100\mu \mathrm{m}$ . Note that this “flattening” of the ${\mathrm{ED}}_{50}$ curve is consistent even though the primary damage mechanism changes from a thermal process for the longer duration exposures to a photomechanical or photoacoustical process for the shorter duration exposures.^{9, 10, 11, 12}

Fig. 1

${\mathrm{ED}}_{50}$ versus retinal beam spot diameter for several laser wavelengths and exposure durations. Circles: $514\mathrm{nm}$ , $0.1\mathrm{s}$ (Ref. 4). Squares: $590\mathrm{nm}$ , $3\mu \mathrm{s}$ (Refs. 5, 6). Triangles: $532\mathrm{nm}$ , $7\mathrm{ns}$ (Refs. 5, 6). Inverted triangles: $633\mathrm{nm}$ , $0.125\mathrm{s}\text{to}1\mathrm{s}$ (Ref. 7). Diamonds: $1060\mathrm{nm}$ , $150\mathrm{fs}$ (Ref. 8).

The difference in spot size dependence is illustrated in Fig. 2 , in which ${\mathrm{ED}}_{50}$ values obtained from in vivo NHP measurements and in vitro retinal explant measurements are directly compared. One possible explanation for the discrepancy in the spot size dependence of the retinal damage threshold between the in vivo and in vitro models is that uncompensated refractive error and higher-order aberrations of the dilated eye of an anesthetized NHP significantly affect the quality of focus. Researchers attempt to minimize the impact of these aberrations by selecting subjects having a refractive error within 0.5 or 0.25 diopters from emmetropia. In addition, the diameter of the laser beam entering the eye is kept small, typically 3 or $4\mathrm{mm}$ in diameter, so that only the central portions of the focusing elements of the eye are used. However, poorer focus will lead to a larger beam spot size at the retina. Even in the absence of higher-order aberrations, an uncertainty of $\pm 0.25\mathrm{D}$ in refractive error will lead to an uncertainty in the retinal beam spot size that will tend to increase the value of the ${\mathrm{ED}}_{50}$ extracted from the data.¹³ In the bovine explant measurements, there are no preretinal focusing elements to distort the beam, offering precise control of the beam spot size.

Fig. 2

Comparison of ${\mathrm{ED}}_{50}$ versus retinal beam spot diameter measured in vivo (NHP, open symbols) and in vitro (bovine explant, closed symbols). For the in vivo data, the ${\mathrm{ED}}_{50}$ is given as the total intraocular energy (TIE). For the in vitro data, the ${\mathrm{ED}}_{50}$ is given as the total incident energy. The dashed lines are the ${\mathrm{ED}}_{50}$ for the in vivo data extrapolated from large spot size to minimum spot size assuming a trend parallel to the in vitro data. Open squares: $514\mathrm{nm}$ , $0.1\mathrm{s}$ (Ref. 4). Open circles: $532\mathrm{nm}$ , $7\mathrm{ns}$ (Refs. 5, 6). Solid squares: $532\mathrm{nm}$ , $0.1\mathrm{s}$ (Ref. 2). Solid circles: $532\mathrm{nm}$ , $100\mathrm{ps}$ (Ref. 3).

To test this hypothesis, a method to compensate for the aberrations of the NHP eye is required. Adaptive optics (AO), utilizing a deformable mirror in a closed loop with a wavefront sensor, has been used to correct for ocular aberrations to improve the resolution of retinal images^{14, 15, 16, 17} and to investigate visual function and supernormal vision.^{14, 18} We have incorporated an adaptive optics system into a standard retinal exposure setup used to measure laser retinal injury thresholds. The adaptive optics system compensates for aberrations of the eye by predistorting the wavefront of the laser beam incident to the eye. The distorted beam is focused onto a spot on the retina that more closely approaches the diffraction limit imposed by the size of the pupil.

The ${\mathrm{ED}}_{50}$ for laser-induced retinal injury was measured for $0.1\text{-}\mathrm{s}$ and $3.5\text{-}\mathrm{ns}$ exposures at $\lambda =532\mathrm{nm}$ . Measurements were made with and without wavefront correction and compared. Furthermore, measurements were made with a small-diameter $(\sim 3\mathrm{mm})$ beam and a large-diameter beam $(\sim 6\mathrm{mm})$ incident at the eye. Ocular aberrations will lead to a poorer focus of the large-diameter beam than of the small-diameter beam.¹⁹ Therefore, the large-diameter beam should exhibit a larger difference between the ${\mathrm{ED}}_{50}$ measured with and without wavefront correction.

2.

Method

Measurement of the ${\mathrm{ED}}_{50}$ for $0.1\text{-}\mathrm{s}$ exposures was performed first. Improvements to the experimental setup and procedure were made based on experience gained from this experiment before performing the $3.5\text{-}\mathrm{ns}$ exposure duration measurements.

2.2.

Exposure Setup

2.2.1.

Uncorrected $0.1\text{-}\mathrm{s}$ exposures

The measurement of the ${\mathrm{ED}}_{50}$ for $0.1\text{-}\mathrm{s}$ exposures for the uncorrected beam (no wavefront correction) was carried out using a standard retinal laser exposure setup, illustrated in Fig. 3 .

Fig. 3

Diagram of the setup for performing the $0.1\text{-}\mathrm{s}$ duration uncorrected exposures. T: beam expanding telescope. A: aperture. NDF: neutral density filters. M: movable mirror.

A $532\text{-}\mathrm{nm}$ continuous-wave (CW) beam was obtained from a Spectra-Physics Millenia Pro frequency-doubled Nd:YAG laser (Spectra-Physics, Mountain View, Calfornia). After passing through a beam-expanding telescope, the central portion of the beam was selected using a circular aperture. An electronic shutter was used to produce pulses of $0.1\text{-}\mathrm{s}$ duration.

Exposures were made with a highly collimated “small” and “large” diameter beam. A $3\text{-}\mathrm{mm}\text{-}\mathrm{diam}$ aperture was used to produce the small-diameter beam. This beam had a uniform, or “top hat” intensity profile, with a divergence of $0.26\mathrm{mrad}$ . The large-diameter beam, produced with a $7\text{-}\mathrm{mm}\text{-}\mathrm{diam}$ aperture, also had a uniform intensity profile, but a divergence of $0.12\mathrm{mrad}$ . Since the fully dilated pupil of the cynomolgous monkey eye was measured to be $\sim 8\mathrm{mm}$ in diameter, the exposure beams were not clipped by the iris in passing through to the interior of the eye.

The beam was directed into the subject eye by a front surface mirror mounted on a translation stage just in front of a fundus camera. The mirror could be moved aside to allow viewing of the retina and then be accurately repositioned for the exposure. The setup was carefully aligned so that the beam entering the eye was collinear with the optical axis of the fundus camera. Retinal target sites could then be designated using the fundus camera crosshair.

Neutral density filters (NDF) were used to determine the beam energy for an exposure. Beam calibration and selection of the appropriate NDF for the desired pulse energies was performed by placing a detector at the eye position prior to placing an animal in the setup. The calibration was verified by repeating the calibration procedure after removing the animal from the setup following the planned series of exposures.

2.2.2.

Wavefront correction

The setup used to perform the wavefront-corrected exposures is illustrated in Fig. 4 . Wavefront correction was performed using a Clarifi-3D Closed Loop Adaptive Optical System purchased from AgilOptics (AgilOptics, Inc., Albuquerque, New Mexico). A 37-actuator deformable mirror is controlled by proprietary software running on a personal computer. Feedback is provided by a Hartmann wavefront analyzer (HWA), which consists of a rectangular array of pinhole apertures mounted on a video camera. In the exposure setup, the deformable mirror and the pinhole plate of the HWA are located at positions conjugate to the pupil plane of the subject eye.

Fig. 4

Setup for performing the $3.5\text{-}\mathrm{ns}$ duration wavefront-corrected exposures. The setup for performing the $0.1\text{-}\mathrm{s}$ exposures was similar. M1 to M13: mirrors. L1 to L4: achromatic doublet lenses. P: pellicle beamsplitter. DBS1: dichroic beamsplitter; reflects at $635\mathrm{nm}$ and transmits at $532\mathrm{nm}$ . DBS2: dichroic beamsplitter; reflects at $532\mathrm{nm}$ and transmits at $635\mathrm{nm}$ . BS: beamsplitter. T: beam expanding telescope. A: apertures. NDF: neutral density filters. The $635\text{-}\mathrm{nm}$ reference beam from the diode laser (thin dark line) is directed into the eye. Light from the reference beam reflected off the retina (light gray) exits the eye and is expanded by the first telescope (L1 to L2) to fill the active area of the deformable mirror. Defocus is introduced into this telescope using the optometer combination (M3 to M6). The diameter of the beam is reduced by the second telescope (L3 to L4) to match the entrance of the wavefront sensor. Data from the wavefront sensor is processed by the computer, which controls the shape of the deformable mirror. The $532\text{-}\mathrm{nm}$ exposure beam (dark gray) is introduced into the system by the dichroic beamsplitter (DBS2). The lens L5 is used to compensate for the chromatic difference of refraction of the subject eye for the wavelengths of the reference and exposure beams. The aperture A2 prevents large-angle diffraction from the beam-diameter-defining aperture A1 from entering the setup.

A solid-state diode laser produced the $635\text{-}\mathrm{nm}$ wavelength beam used as the reference beam for the adaptive optics system. This beam was approximately $1\mathrm{mm}$ in diameter at the 1/e irradiance point, with a divergence of $0.75\mathrm{mrad}$ . To prevent reflections off the cornea from reaching the wavefront analyzer, the reference beam is introduced into the eye near the edge of the pupil (i.e., off the optic axis of the subject eye), but still parallel to the axis of the fundus camera.

In order to produce a strong enough signal for the HWA, the reference beam had to be relatively intense. The reference beam used to perform wavefront correction for the $3.5\text{-}\mathrm{ns}$ exposures typically introduced a total intraocular power (TIP) of $\sim 350\mu \mathrm{W}$ . The maximum permissible exposure (MPE)²⁰ for a long-duration cw exposure at $\lambda =635\mathrm{nm}$ corresponds to a TIP of $385\mu \mathrm{W}$ .

Light from the reference beam diffusely reflects off the retina and exits the eye in a collimated beam limited by the diameter of the ocular pupil. The beam is expanded by a telescope to fill the active area of the deformable mirror. In Fig. 4, this telescope consists of the optical elements from lens L1 to lens L2. Per manufacturer recommendation, the adaptive optics system was set up with some curvature on the surface of the deformable mirror, allowing the system to make both positive and negative corrections of aberrations. This required that some defocus be deliberately introduced into the telescope. The optometer, consisting of the four mirrors labeled M3, M4, M5, and M6 in Fig. 4, was used for this purpose. Mirrors M4 and M5 were mounted on a translation stage. The optical length of the telescope, and therefore the focus, was altered by adjusting the position of these two mirrors.

After reflecting off the deformable mirror, the diameter of the reference beam is reduced to match the entrance aperture of the wavefront sensor by a second telescope, consisting of lenses L3 and L4 in Fig. 4. A beam-splitter directed most of the reference beam light into the HWA. A small portion of the reference beam passes through the beamsplitter and is imaged by a $75\text{-}\mathrm{mm}$ lens onto the detector array of a video camera that was therefore conjugate to the retina of the eye. This provided a means for visually monitoring the performance of the adaptive optics system.

The Spectra-Physics laser produced a CW beam. The $3.5\text{-}\mathrm{ns}$ , $\lambda =532\mathrm{nm}$ pulses were produced using a Continuum Minilite II Q-switched Nd:YAG laser (Continuum, Santa Clara, California). The choice of laser for a given exposure was determined by the position of mirror M10. After passing through a beam-expanding telescope (T), the central portion of the beam was selected using a circular aperture (A1). Aperture diameters were selected to produce a $3\text{-}\mathrm{mm}$ -diam beam at the cornea for the small-diameter exposures, and a $6\text{-}\mathrm{mm}$ -diam beam at the cornea for the large-diameter exposures. Neutral density filters (NDF) were used to determine the pulse energy for the exposure. The shutter determined the exposure duration for the CW laser or selected a single Q-switched pulse from the pulsed laser. After passing through the optics of the exposure system, the $3\text{-}\mathrm{mm}$ -diam beam had a divergence of $0.9\mathrm{mrad}$ , and the $6\text{-}\mathrm{mm}$ -diam beam had a divergence of $1.0\mathrm{mrad}$ .

The lens L5 in Fig. 4 was used to correct for the chromatic difference of refraction of the rhesus monkey eye between the wavelength of the reference beam $\left(635\mathrm{nm}\right)$ for the AO system and the wavelength of the exposure beam $\left(532\mathrm{nm}\right)$ . Choice of focusing power and position of this lens is discussed in Section 2.3. The $0.1\text{-}\mathrm{s}$ wavefront corrected exposures did not include correction for the chromatic difference of refraction of the subject eye.

The exposure beam was introduced into the system using the dichroic beamsplitter labeled DBS2. This beamsplitter was highly reflective at $\lambda =532\mathrm{nm}$ , the wavelength of the exposure beam, but transparent at $\lambda =635\mathrm{nm}$ , the wavelength of the reference beam. The beamsplitter therefore did not interfere with the operation of the adaptive optics system, and wavefront correction could continue during the duration of an exposure.

The pellicle (P) that directed the reference laser into the eye also deflected a constant proportion of the $532\text{-}\mathrm{nm}$ exposure beam into a reference detector (labeled Detector in Fig. 4). The ratio of the energy at the eye position to the energy at the reference detector was determined and used for dosimetry of the Q-switched exposures. The dichroic beamsplitter DBS1 protected the reference detector from the $635\text{-}\mathrm{nm}$ reference beam.

2.2.3.

Uncorrected $3.5\text{-}\mathrm{ns}$ exposures

The uncorrected $3.5\text{-}\mathrm{ns}$ exposures were performed through the wavefront correction setup illustrated in Fig. 4. There was no separate setup to perform the uncorrected exposures, as was the case for the $0.1\text{-}\mathrm{s}$ exposures. In order to perform the $3.5\text{-}\mathrm{ns}$ uncorrected exposures, the deformable mirror was flattened (using the control software), and the optometer (mirrors M3 to M6) was adjusted to produce a collimated beam at the position of the subject eye.

2.3.

Compensation of Chromatic Aberration

The reference beam used to measure the aberrations of the subject monkey’s eye had a wavelength of $635\mathrm{nm}$ . The adaptive optics system works to provide the best focus (i.e., smallest beam spot diameter) at the retina for light having the same wavelength. The measurement of astigmatism and higher-order aberrations has been shown to be only weakly dependent on the wavelength of the reference beam used.²¹ However, the ocular medium exhibits significant chromatic dispersion, and the effective focal length of the eye is significantly wavelength dependent.

The effective focal length of the rhesus monkey eye is generally taken to be $13.5\mathrm{mm}$ . The adaptive optics system works to bring light of $\lambda =635\mathrm{nm}$ to a focus at this effective distance. The chromatic difference of refraction of the human eye between light of wavelengths $\lambda =532\mathrm{nm}$ and $\lambda =635\mathrm{nm}$ is $-0.52$ dipoters.²² If the rhesus monkey eye is assumed to have the same chromatic dispersion as the human eye, then, when the adaptive optics system is working, light of the wavelength $\lambda =532\mathrm{nm}$ will be focused at an effective depth of $13.4\mathrm{mm}$ . The green exposure beam was therefore focused about $100\mu \mathrm{m}$ anterior to the retina.

The focus of the green exposure beam $(\lambda =532\mathrm{nm})$ can be moved to the same plane as the focus of the reference beam $(\lambda =635\mathrm{nm})$ by introducing a small divergence into the exposure beam. The lens L5 in Fig. 4 was used to do this. Lens L5 was placed $61\mathrm{cm}$ before the deformable mirror in the optical path of the exposure beam. This location was largely determined by the available space in the setup. Simple geometric ray tracing indicated that a $-0.11$ diopter lens at this location would move the focus of the exposure beam to the location of the focus of the reference beam at the retina.

The ray tracing prediction was verified by measuring the ${\mathrm{ED}}_{50}$ for a $3.5\text{-}\mathrm{ns}$ exposure with no lens (0 diopter), a $-0.12$ diopter lens, and a $-0.25$ diopter lens at the position of L5 in Fig. 4. A $-0.11$ diopter lens was not available; lenses were taken from a standard optometrist’s trial lens set. A $3\text{-}\mathrm{mm}$ -diam beam was used, and the adaptive optics system was run to provide wavefront correction during these measurements. A plot of the ${\mathrm{ED}}_{50}$ values obtained is shown in Fig. 5 . Note that data collection continued after this verification process was complete, and therefore Fig. 5 is based on a subset of the final data set used to determine the ${\mathrm{ED}}_{50}$ reported here. The lowest ${\mathrm{ED}}_{50}$ value was indeed obtained using the $-0.12$ diopter lens. This lens was therefore used for all subsequent $3.5\text{-}\mathrm{ns}$ exposures.

Fig. 5

Verification of the lens used to correct for chromatic difference of refraction between the wavelength of the AO reference beam and the exposure beam for the $3.5\text{-}\mathrm{ns}$ exposures. The plot shows the ${\mathrm{ED}}_{50}$ for a $3.5\text{-}\mathrm{ns}$ exposure, expressed as the total intraocular energy, obtained by using a lens of the indicated focusing power at the location L5 of Fig. 4. The adaptive optics system was run to provide wavefront correction for these measurements. For comparison, the ${\mathrm{ED}}_{50}$ obtained without using wavefront correction is included. Error bars indicate the 95% confidence limits. Note that this figure is based on a subset of the final data set, as data collection continued after the verification process was completed.

Note that the chromatic aberration compensation was done only for the $3.5\text{-}\mathrm{ns}$ exposure ${\mathrm{ED}}_{50}$ measurement. The $0.1\text{-}\mathrm{s}$ exposure measurements did not include this correction.

3.

Results

3.1.

Threshold for $0.1\text{-}\mathrm{ns}$ Exposures

Table 1 lists the ${\mathrm{ED}}_{50}$ values obtained for the $0.1\mathrm{s}$ duration exposures for the six exposure conditions (small/large beam, wavefront corrected/uncorrected, macular/paramacular regions of the retina). ${\mathrm{ED}}_{50}$ values are given for $1\mathrm{h}$ and $24\mathrm{h}$ post-exposure evaluation. Also listed are the 95% confidence limits and the ${\mathrm{ED}}_{84}∕{\mathrm{ED}}_{50}$ ratios obtained from the probit analysis. In all cases, a smaller ${\mathrm{ED}}_{50}$ was measured when wavefront correction was used. The rightmost column of Table 1 shows the percent decrease in ${\mathrm{ED}}_{50}$ measured with a wavefront-corrected beam relative to the value measured with the uncorrected beam.

Table 1

ED50 for producing a discernible retinal alteration in the eye of a cynomolgous monkey (Macaca fascicularis) from a 0.1-s duration exposure to laser irradiation of wavelength λ=532nm , measured with and without wavefront correction. Values are expressed as the total intraocular energy (TIE). The rightmost column is the fractional change in the value of the ED50 obtained using wavefront correction relative to the value obtained without wavefront correction.

Beamdiameter	Retinallocation	Observationtime	Uncorrected			Wavefront corrected			Relativechangein ED50
			ED50 (mJ)	95%confidencelimits	ED84∕ED50	ED50 (mJ)	95%confidencelimits	ED84∕ED50
Small	Macula	$1\mathrm{h}$	1.72	1.25–2.25	1.17	1.56	1.27–1.86	1.23	$-0.09$
		$24\mathrm{h}$	1.35	0.97–1.71	1.36	1.05	0.68–1.27	1.29	$-0.22$
	Paramacula	$1\mathrm{h}$	2.09	1.81–2.45	1.22	1.97	1.76–2.24	1.19	$-0.06$
		$24\mathrm{h}$	1.42	1.00–1.70	1.37	1.14	1.00–1.31	1.21	$-0.20$
Large	Macula	$1\mathrm{h}$	2.07	1.69–2.74	1.18	1.92	1.50–2.52	1.31	$-0.07$
		$24\mathrm{h}$	1.96	1.50–2.67	1.28	1.06	0.71–1.40	1.50	$-0.46$
	Paramacula	$1\mathrm{h}$	2.52	2.18–3.07	1.20	2.16	1.89–2.51	1.24	$-0.14$
		$24\mathrm{h}$	2.29	1.96–2.70	1.23	1.47	1.21–1.67	1.20	$-0.36$

For the $3\text{-}\mathrm{mm}$ -diam uncorrected beam, we determined the ${\mathrm{ED}}_{50}$ for a $0.1\text{-}\mathrm{s}$ exposure at $532\text{-}\mathrm{nm}$ wavelength to be $1.25\text{-}\mathrm{mJ}$ TIE for a macular exposure and $1.42\text{-}\mathrm{mJ}$ TIE for a paramacular exposure. These values are for the $24\text{-}\mathrm{h}$ observation time. No previous measurement of the ${\mathrm{ED}}_{50}$ for a $0.1\text{-}\mathrm{s}$ exposure at $532\mathrm{nm}$ has been reported. The ${\mathrm{ED}}_{50}$ for a $0.1\text{-}\mathrm{s}$ exposure at $514\mathrm{nm}$ in the rhesus monkey (Macaca mulatta) was recently reported to be $1.05\text{-}\mathrm{mJ}$ TIE for a macular exposure and $1.23\text{-}\mathrm{mJ}$ TIE for a paramacular exposure.³ Given the difference in laser wavelength and subject species, we concluded that the measured ${\mathrm{ED}}_{50}$ values reported here are consistent with the earlier data.

The ${\mathrm{ED}}_{50}$ values obtained with and without wavefront correction for the $0.1\text{-}\mathrm{s}$ exposures are plotted versus the evaluation time in Fig. 6 . For comparison, the ${\mathrm{ED}}_{50}$ measured in the rhesus monkey⁴ at $514\mathrm{nm}$ is included in the plot of the uncorrected beam results. The impact of using wavefront correction in the exposure system is displayed graphically in Fig. 7 .

Fig. 6

${\mathrm{ED}}_{50}$ for producing discernible retinal alteration from a $0.1\text{-}\mathrm{s}$ exposure to a $532\text{-}\mathrm{nm}$ wavelength laser, expressed as the total intraocular energy (TIE). The ${\mathrm{ED}}_{50}$ is plotted versus the evaluation time ( $1\mathrm{h}$ or $24\mathrm{h}$ ). The retinal location (macula/paramacula) and wavefront correction status (on/off) are indicated. For comparison, the ${\mathrm{ED}}_{50}$ for a $0.1\text{-}\mathrm{s}$ exposure at $514\mathrm{nm}$ , taken from Ref. 3, is included. Error bars indicate the 95% confidence limits. Points are offset for clarity. Solid circles/solid lines: large-diameter beam, $532\mathrm{nm}$ . Open circles/dashed lines: small-diameter beam, $532\mathrm{nm}$ . Solid triangles/dotted lines: $514\mathrm{nm}$ (Ref. 3).

Fig. 7

Impact of wavefront correction on measured ${\mathrm{ED}}_{50}$ values for a $0.1\text{-}\mathrm{s}$ exposure at $532\mathrm{nm}$ . The region of the retina (macula/paramacula) and the evaluation time $(1\mathrm{h}∕24\mathrm{h})$ are indicated. Error bars indicate the 95% confidence limits. Points are offset for clarity. Solid circles/solid line: large-diameter beam. Open circles/dashed line: small-diameter beam.

3.2.

Threshold for $3.5\text{-}\mathrm{ns}$ Exposures

Table 2 lists the ${\mathrm{ED}}_{50}$ values determined for the $3.5\text{-}\mathrm{ns}$ -duration exposures for the four exposure conditions ( $3\text{-}\mathrm{mm}∕6\text{-}\mathrm{mm}$ -diam beam, wavefront corrected/uncorrected). All exposures were placed in the macula. ${\mathrm{ED}}_{50}$ values are given for $1\mathrm{h}$ and $24\mathrm{h}$ post-exposure evaluation. The wavefront-corrected measurements include compensation for the chromatic difference of refraction between the wavelength of the AO reference beam $\left(635\mathrm{nm}\right)$ and the wavelength of the exposure beam $\left(532\mathrm{nm}\right)$ . The rightmost column of Table 2 shows the decrease in the ${\mathrm{ED}}_{50}$ obtained using wavefront correction relative to the value obtained with the uncorrected beam.

Table 2

ED50 for producing a discernible retinal alteration in the macula of a rhesus monkey (Macaca mulatta) from a 3.5-ns pulse at λ=532nm , measured with and without wavefront correction. Values are expressed as the total intraocular energy (TIE).

Beamdiameter	Observationtime	Uncorrected			Wavefront corrected			Relativechange in ED50
		ED50 (μJ)	95%confidencelimits	ED84∕ED50	ED50 (μJ)	95%confidencelimits	ED84∕ED50
$3\mathrm{mm}$	$1\mathrm{h}$	1.05	0.82–1.39	1.81	0.77	0.61–0.99	1.72	$-0.37$
	$24\mathrm{h}$	0.81	0.64–0.97	1.50	0.51	0.38–0.65	1.76	$-0.37$
$6\mathrm{mm}$	$1\mathrm{h}$	2.01	1.42–2.67	2.14	1.33	1.02–2.14	1.81	$-0.34$
	$24\mathrm{h}$	1.55	1.09–1.98	1.85	0.57	0.29–0.79	2.22	$-0.63$

For the $3\text{-}\mathrm{mm}$ -diam uncorrected beam, we determined the ${\mathrm{ED}}_{50}$ for a $3.5\text{-}\mathrm{ns}$ exposure at $532\text{-}\mathrm{nm}$ wavelength to be $1.05\text{-}\mu \mathrm{J}$ TIE at $1\mathrm{h}$ post-exposure and $0.81\text{-}\mu \mathrm{J}$ TIE at $24\mathrm{h}$ . Zuclich ⁶ reported the ${\mathrm{ED}}_{50}$ for a $5\text{-}\mathrm{ns}$ exposure at $532\mathrm{nm}$ to be $1.31\text{-}\mu \mathrm{J}$ TIE at $1\mathrm{h}$ post-exposure and $0.64\text{-}\mu \mathrm{J}$ TIE at $24\mathrm{h}$ for a macular exposure.

The ${\mathrm{ED}}_{50}$ values obtained with and without wavefront correction for the $3.5\text{-}\mathrm{ns}$ exposures are plotted versus the evaluation time in Fig. 8 . For comparison, the ${\mathrm{ED}}_{50}$ values reported by Zuclich ⁶ are included in the plot of the uncorrected beam results. Figure 9 shows the impact of using wavefront correction along with chromatic aberration compensation.

Fig. 8

${\mathrm{ED}}_{50}$ for producing discernible retinal alteration from a $3.5\text{-}\mathrm{ns}$ exposure to a $532\text{-}\mathrm{nm}$ wavelength laser, expressed as the total intraocular energy (TIE). The ${\mathrm{ED}}_{50}$ is plotted versus the evaluation time ( $1\mathrm{h}$ or $24\mathrm{h}$ ) for each beam diameter. The left plot shows the ${\mathrm{ED}}_{50}$ obtained using the uncorrected beam. For comparison, the ${\mathrm{ED}}_{50}$ for a $5\text{-}\mathrm{ns}$ exposure at $532\mathrm{nm}$ , taken from Ref. 5, is included. The right plot shows the ${\mathrm{ED}}_{50}$ obtained when wavefront correction was used, and correction was made for the chromatic difference of refraction between the wavelengths of the adaptive optics reference beam $\left(635\mathrm{nm}\right)$ and the exposure beam $\left(532\mathrm{nm}\right)$ . Error bars indicate the 95% confidence limits. Points are offset for clarity. Solid circles/solid lines: $6\text{-}\mathrm{mm}$ -diam. beam. Open circles/dashed lines: $3\text{-}\mathrm{mm}$ -diam. beam. Solid triangles/dotted line: $5\mathrm{ns}$ (Ref. 5).

Fig. 9

Impact of wavefront correction on measured ${\mathrm{ED}}_{50}$ values for a $3.5\text{-}\mathrm{ns}$ exposure at $532\mathrm{nm}$ . Correction was made for the chromatic difference of refraction between the wavelengths of the adaptive optics reference beam $\left(635\mathrm{nm}\right)$ and the exposure beam $\left(532\mathrm{nm}\right)$ when wavefront correction was turned on. The evaluation time $(1\mathrm{h}∕24\mathrm{h})$ is indicated. Error bars indicate the 95% confidence limits. Points are offset for clarity. Solid circles/solid lines: $6\text{-}\mathrm{mm}$ -diam. beam. Open circles/dashed lines: $3\text{-}\mathrm{mm}$ -diam. beam.

4.

Discussion

Although we have no direct measure of the resulting spot size, the ${\mathrm{ED}}_{50}$ results provide evidence that the AO system was indeed compensating to a significant extent for the refractive error and higher-order aberrations of the subject NHP eyes. For all exposure conditions (retinal location, incident beam diameter, exposure duration), the ${\mathrm{ED}}_{50}$ measured while using the AO system was lower than the ${\mathrm{ED}}_{50}$ obtained without using wavefront correction. The reduction in the ${\mathrm{ED}}_{50}$ is more pronounced for the large-diameter beam than for the small-diameter beam. In particular, the ${\mathrm{ED}}_{50}$ obtained for the two beam diameters converge to similar values when wavefront compensation is used (Figs. 7 and 9).

The large-diameter beam “samples” more of the ocular aberrations and is expected to result in a more distorted retinal beam spot.¹⁹ Effectively correcting for these aberrations should therefore have a more significant impact on the quality of focus of the larger diameter beam. The diffraction-limited spot size is $3\mu \mathrm{m}$ for the $6\text{-}\mathrm{mm}$ beam and $6\mu \mathrm{m}$ for the $3\text{-}\mathrm{mm}$ beam. If diffraction-limited spots were achieved when the AO system was used, the peak retinal irradiance would be higher for the larger-diameter beam, and the ${\mathrm{ED}}_{50}$ obtained using the large-diameter beam would be lower than the ${\mathrm{ED}}_{50}$ obtained from the small-diameter beam. However, the divergence of the beams at the eye after passing through the optical path of the exposure system limits the smallest achievable retinal spot size to no less than about $10\mu \mathrm{m}$ for both beams, even after compensating for all ocular aberrations. The retinal beam spots produced by the large- and small-diameter beams were thus about the same size when the AO system was used to provide wavefront correction.

The ${\mathrm{ED}}_{50}$ for laser-induced retinal damage from exposure to a $532\text{-}\mathrm{nm}$ wavelength laser has been measured using small- and large-diameter beams incident at the cornea. The large-diameter beam measurements demonstrate that the adaptive optics system did effectively compensate for refractive error and other ocular aberrations of the NHP eye. However, the ${\mathrm{ED}}_{50}$ is usually measured using a small-diameter beam, in order to minimize the impact of the ocular aberrations.

The ${\mathrm{ED}}_{50}$ at the $24\text{-}\mathrm{h}$ observation point for the $3\text{-}\mathrm{mm}$ beam was measured to be $0.51\text{-}\mu \mathrm{J}$ TIE for a $532\text{-}\mathrm{nm}$ , $3.5\text{-}\mathrm{ns}$ exposure when wavefront correction was used (Table 2). This is 37% smaller than the value of $0.81\mu \mathrm{J}$ obtained without any compensation for ocular aberrations. In Fig. 2, the dashed lines show an extrapolation of the in vivo NHP ${\mathrm{ED}}_{50}$ data assuming that the ${\mathrm{ED}}_{50}$ continues to decrease with decreasing retinal spot size, as it does for the in vitro retinal explant data. For both sets of data ( $514\text{-}\mathrm{nm}$ , $0.1\text{-}\mathrm{s}$ exposures and $532\text{-}\mathrm{nm}$ , $7\text{-}\mathrm{ns}$ exposures), the extrapolated minimum spot ${\mathrm{ED}}_{50}$ is an order of magnitude less than the measured value. If the sole explanation for the flattening of the threshold versus spot size curves (Fig. 2) for small spot sizes is that ocular aberrations lead to a large retinal beam spot size, then correcting for those aberrations would be expected to lead to a reduction in the measured ${\mathrm{ED}}_{50}$ of $\sim 90\%$ from the value obtained without correction. This suggests that the ocular aberrations of the anesthetized NHP eye not only might spread the energy of the laser pulse to an area larger than a diffraction-limited spot size, but also might impact the ability to see the small lesions during an ophthalmic examination. Scatter in the ocular media must also be considered.

Typically, the ${\mathrm{ED}}_{50}$ inferred from the $24\text{-}\mathrm{h}$ post-exposure evaluation is smaller than the value extracted from the $1\text{-}\mathrm{h}$ post-exposure data. This is generally attributed to the difficulty in detecting the smallest damaged areas until metabolic processes related to the injury response mechanism expand the altered area of the retina beyond the initial injury site to an area large enough to be detected via ophthalmoscopic observation. The data presented here follow this trend. For all exposure conditions (retinal location, exposure duration, spot size, AO on/off), the ${\mathrm{ED}}_{50}$ obtained from the $24\text{-}\mathrm{h}$ observations is less than the ${\mathrm{ED}}_{50}$ obtained from the $1\text{-}\mathrm{h}$ observations. Furthermore, the relative decrease in the ${\mathrm{ED}}_{50}$ measured with wavefront correction, compared to the value measured without wavefront correction, is less for the $1\text{-}\mathrm{h}$ data than for the $24\text{-}\mathrm{h}$ data. This further supports the conclusion that ocular aberrations not only lead to a poorly focused beam, therefore producing a larger than diffraction-limited retinal beam spot, but also negatively impact the ability of an examiner to detect the smallest lesions.

For the $0.1\text{-}\mathrm{s}$ exposures, the ${\mathrm{ED}}_{50}$ at the $24\text{-}\mathrm{h}$ observation point for the $3\text{-}\mathrm{mm}$ -diam beam was measured to be $1.05\text{-}\mathrm{mJ}$ TIE when wavefront correction was used, a reduction of 22% from the value of $1.35\text{-}\mathrm{mJ}$ TIE obtained without wavefront correction. However, the ${\mathrm{ED}}_{50}$ measurements for the $0.1\text{-}\mathrm{s}$ exposure do not include compensation for the chromatic difference of refraction of the NHP eye between the wavelength of the AO reference beam $\left(635\mathrm{nm}\right)$ and the wavelength of the exposure beam $\left(532\mathrm{nm}\right)$ . This compensation was included in the $3.5\text{-}\mathrm{ns}$ exposure measurements (Fig. 5). For a $3\text{-}\mathrm{mm}$ -diam beam, it was found that the $1\text{-}\mathrm{h}$ ${\mathrm{ED}}_{50}$ dropped from $1.12\mu \mathrm{J}\text{to}0.69\mu \mathrm{J}$ when compensation for chromatic aberration was included in addition to wavefront correction. The $24\text{-}\mathrm{h}$ ${\mathrm{ED}}_{50}$ was lowered from $0.87\mu \mathrm{J}\text{to}0.53\mu \mathrm{J}$ when compensation for chromatic aberration was included. In both cases, the decrease in ${\mathrm{ED}}_{50}$ is about 38% when compensating for the chromatic difference of refraction. The damage mechanisms are different for the ns-duration and the $100\text{-}\mathrm{ms}$ -duration exposures. Thermal diffusion is not a factor for the $3.5\text{-}\mathrm{ns}$ exposures but is a significant factor for the $100\text{-}\mathrm{ms}$ duration exposures and may lessen the impact of further improving the focus of the beam using chromatic compensation in addition to wavefront correction for the $100\text{-}\mathrm{ms}$ exposures. However, we can use the $3.5\text{-}\mathrm{ns}$ exposure results to place an upper limit on the effect of chromatic compensation on the $100\text{-}\mathrm{ms}$ data. Assuming the same impact on the $0.1\text{-}\mathrm{s}$ exposure measurements, we can estimate that use of chromatic compensation would at most reduce the ${\mathrm{ED}}_{50}$ for a macular exposure from $1.05\mathrm{mJ}$ at the $24\text{-}\mathrm{h}$ endpoint as listed in Table 1 to $0.65\text{-}\mathrm{mJ}$ TIE. This is a relative reduction of 52% from the value obtained without wavefront correction, $1.35\mathrm{mJ}$ .

Although the $0.1\text{-}\mathrm{s}$ exposure measurements do not include compensation for the chromatic difference of refraction, the ${\mathrm{ED}}_{50}$ measured with wavefront correction is smaller than the ${\mathrm{ED}}_{50}$ measured without such correction. Wavefront correction may not be providing the smallest achievable retinal spot size in this case but is nonetheless improving the concentration of energy at the retina.

This research was motivated by a concern that the MPE as provided by current laser safety guidelines^{20, 24, 25, 26} might not provide an adequate margin of safety for an actively focusing alert eye. We have attempted to more closely approximate such an eye by compensating for the ocular aberrations in the commonly used NHP model. While we did not find an order of magnitude reduction in the ${\mathrm{ED}}_{50}$ as suggested by the trend of the in vitro data, we did find a threshold value for $3.5\text{-}\mathrm{ns}$ duration, $532\text{-}\mathrm{nm}$ exposures that is lower than any previously reported. The value of the ${\mathrm{ED}}_{50}$ as determined for a MVL detectable $24\mathrm{h}$ after a macular exposure, $0.51\mu \mathrm{J}$ , is a factor of only 2.6 higher than the MPE for a $3.5\text{-}\mathrm{ns}$ , visible wavelength exposure ( $0.2\text{-}\mu \mathrm{J}$ TIE).

5.

Conclusion

An adaptive optics system was incorporated into a laser retinal exposure setup in order to correct for the refractive error and higher-order aberrations of a nonhuman primate eye in order to more closely approximate the exposure conditions experienced by an actively focusing eye in an alert individual. Using this system, the ${\mathrm{ED}}_{50}$ for a $100\text{-}\mathrm{ms}$ , $532\text{-}\mathrm{nm}$ small spot size macular exposure was measured to be $1.05\mathrm{mJ}$ TIE, a reduction of 22% from the value of $1.35\text{-}\mathrm{mJ}$ TIE measured without aberration correction.

Using wavefront correction, the ${\mathrm{ED}}_{50}$ for a $3.5\text{-}\mathrm{ns}$ , $532\text{-}\mathrm{nm}$ small spot size macular exposure was measured to be $0.51\text{-}\mu \mathrm{J}$ TIE, the lowest ${\mathrm{ED}}_{50}$ measured for a ns-duration exposure. This value is a factor of only 2.6 higher than the MPE for a $3.5\text{-}\mathrm{ns}$ , visible wavelength exposure. When no correction for ocular aberrations was used, the ${\mathrm{ED}}_{50}$ was measured to be $0.81\mu \mathrm{J}$ . Use of the adaptive optics system to correct for aberrations reduced the ${\mathrm{ED}}_{50}$ by 37% from this value. The trend of in vitro measurements using bovine retinal explants projects to an ${\mathrm{ED}}_{50}$ one order of magnitude (90%) less than the value obtained from in vivo measurements with no aberration correction (Fig. 2). Distortion of the incident beam by ocular aberrations cannot fully explain the discrepancy between the in vivo measurements and the in vitro results. Other factors such as a limited ability to visually detect threshold retinal lesions in vivo and small-angle forward scatter in the preretinal ocular media must also impact the retinal damage threshold measurement.