What adaptive optics can do for the eye |
147 |
|
|
What adaptive optics can do for the eye
David R. Williams
Center for Visual Science, University of Rochester, Rochester, NY, USA
Methods to correct the optics of the human eye are at least 700 years old. Spectacles have been used to correct defocus as early as the 13th century1,2 and to correct astigmatism since the 19th century.3 Higher-order aberrations, such as coma and spherical aberration, have traditionally been ignored, partly because they are not usually as significant as defocus and astigmatism in blurring the retinal image and partly because they have been difficult to measure and even more difficult to correct.
Today technological developments make it possible to measure all of the monochromatic aberrations of the eye and to greatly reduce their impact on retinal image quality. One key development was the introduction of a rapid and accurate method to measure the eye’s wave aberration. Though a number of investigators beginning with Smirnov4 had measured some of the eye’s higher-order aberrations, Junzhong Liang’s wavefront sensor5 eventually enabled the widespread use of the wave aberration in both clinical and research settings. Liang, working as a graduate student under the direction of Joseph Bille at the University of Heidelberg, demonstrated that the Shack-Hartmann method, which had a long history in assessing the quality of optical instruments, could also be successfully applied to the eye. Subsequently, Liang joined my laboratory where we developed a highresolution wavefront sensor that provided a more complete description of the eye’s wave aberration, measuring up to 10 radial orders.6 These measurements showed that higher-order aberrations can be significant sources of retinal image blur in some eyes when the pupil is large. Liang, Don Miller, and I then showed that we could correct the wave aberration with adaptive optics.7
The idea that adaptive optics could be used to correct an arbitrary pattern of aberrations was already well-known. The astronomer, Horace Babcock, had suggested in 1953 that adaptive optics could remove the blurring effects of turbulence in the atmosphere on telescopic images of stars.8 The U.S. Defense Department later invested heavily in the development of adaptive optics technology to improve the effectiveness of laser weapons as part of its Star Wars Program.
Figure 1 shows the world’s largest telescope, the twin-domed Keck telescope. Like many modern ground-based telescopes, it is equipped with adaptive optics.
Address for correspondence: David R. Williams, PhD, Director, Center for Visual Science, William G. Allyn Professor of Medical Optics, University of Rochester, Box 270270, Rochester, NY 146270270, USA.
Wavefront and Emerging Refractive Technologies, pp. 147–157
Proceedings of the 51st Annual Symposium of the New Orleans Academy of Ophthalmology, New Orleans, LA, USA, February 22-24, 2002
edited by Jill B. Koury
© 2003 Kugler Publications, The Hague, The Netherlands
148 |
D.R. Williams |
|
|
Fig. 1. The twin 10-meter Keck telescope on Mauna Kea in Hawaii. Adaptive optics technology is used to overcome the blurring effects of atmospheric turbulence, allowing resolution with ground-based imaging that in some cases exceeds that of the Hubble Space Telescope. (Courtesy of the Keck Observatory.)
Fig. 2. These images illustrate the power of adaptive optics in astronomy. Taken with the Keck telescope, they show Neptune as observed in the near-infrared with and without adaptive optics. Adaptive optics technology allows ground-based telescopes to monitor Neptune’s evolving weather systems. The image on the right reveals high-altitude clouds, which appear bright against the darker disk. (Courtesy of Bruce Macintosh.)
Due to the benefit of adaptive optics, this telescope can achieve retinal images that are in some cases superior to those obtained with the Hubble Space Telescope. This benefit in resolution is illustrated by pictures of Neptune taken by Claire Max and her colleagues at Lawrence Livermore National Laboratories9 (Fig. 2).
What adaptive optics can do for the eye |
149 |
|
|
In 1989, Dreher et al. made the first attempt to use a deformable mirror to improve retinal images in a scanning laser ophthalmoscope.10 They were able to correct the astigmatism in one subject’s eye based on a conventional spectacle prescription. The development of wavefront sensing technology as well as improved deformable mirrors allowed Liang, Miller, and me in 1997 to demonstrate the first adaptive optics system that could measure and correct not just astigmatism, but the eye’s entire wave aberration. Moreover, this system was closed loop in the sense that the wavefront sensor measured the aberrations in the system repeatedly, updating the mirror each time to optimize image quality. We showed that adaptive optics could improve vision as well as images of the retina. The successful outcome of our initial experiments at Rochester stimulated interest in replacing the deformable mirror in an adaptive optics system with other correcting methods such as refractive surgery, contact lenses, and intraocular lenses so that some of the benefits of adaptive optics could be obtained by patients in every-day vision. The interest in higher-order aberration measurement and correction has become especially keen in the area of refractive surgery because the surgery itself introduces higher-order aberrations, especially spherical aberration, and these aberrations may be responsible for some patients complaining of poor night vision.
How adaptive optics works
The adaptive optics technology used in ophthalmic applications is very similar to that used in astronomy, so much so that astronomers and vision scientists have teamed together to form the National Science Foundation Science and Technology
Fig. 3. In an adaptive optics system designed to image the retina at high resolution, the deformable mirror unbends the distorted wavefront emerging from an aberrated eye, forming the planar wave front required for diffraction-limited imaging. Adaptive optics can also work in reverse. That is, the same mirror shape that corrects the light leaving the eye would also correct light entering the eye, allowing the eye to have an unaberrated view of the world.
150 |
D.R. Williams |
|
|
Fig. 4. Schematic of the Rochester 2nd Generation Adaptive Optics System for the eye. A superluminescent laser diode forms an infrared spot of light on the retina, which is used by the wavefront sensor to measure the eye’s aberrations. These data are transformed by computer into command signals to control the 97 actuators on the deformable mirror. The system measures and corrects the eye’s wave aberration 30 times a second. Once adaptive correction has been achieved, visual performance can be measured by stimuli presented via the display. This same system is also capable of high-resolution retinal imaging (not shown).
Center for Adaptive Optics (http://cfao.ucolick.org/). Figure 3 shows the principle of adaptive optics for forming a sharp image of the retina, unblurred by the eye’s aberrations. If the eye were perfect, light from a single point on the retina would leave the eye as a plane wave. In all real eyes, however, light leaves the eye as a distorted wavefront as shown in the figure. If we measure this wavefront with a wavefront sensor (not shown), we can then instruct a deformable mirror to adopt a shape that exactly compensates for the eye’s wave aberration, converting the wavefront into the plane wave desirable for sharp imaging. This will correct not only defocus and astigmatism, but also all other higher-order aberrations of the eye as well.
At the University of Rochester, we have constructed a second generation adaptive optics system that can automatically measure and correct the eye’s wave aberration 30 times a second11 (Fig. 4). This rate is fast enough to capture most of the changes that occur in the wave aberration over time, such as those caused by microfluctuations in accommodation. An infrared superluminescent diode is imaged on the retina. A Shack-Hartmann wavefront sensor, equipped with 221 lenselets, uses the light scattered back out of the eye from this source to measure the wave aberration across 6.8 mm of the dilated pupil. The deformable mirror we are currently using is made by Xinetics, Inc., (Devens, MA) and consists of an aluminized glass plate which has 97 actuators attached to the back surface in a square array. These actuators change the mirror shape by pushing and pulling on the mirror, displacing it locally by less than ± 2 m. Once the wavefront correction
What adaptive optics can do for the eye |
151 |
|
|
is complete, generally taking less than half a second, we can conduct experiments on visual performance or take high-resolution pictures of the retina. Figure 4 shows the arrangement for measuring visual performance. The subject views a display through the deformable mirror, which provides him or her with a nearly aberra- tion-free view if the mirror is used to correct vision, or an aberrated view if the mirror is used to introduce particular aberrations, as we will describe later.
Improving retinal images
The improvement that can be obtained by imaging the retina with adaptive optics is striking (Fig. 5). Without adaptive optics, structure in the image is difficult to discern, but with adaptive optics it is possible to see individual photoreceptors near the foveal center that are only 5 m in diameter. The dark, vertical feature in the middle of the image is a single capillary about 7 m in diameter that lies above the photoreceptors and is therefore out of focus. To increase image quality beyond what is achievable with a single picture, we can take many pictures and register them as shown on the right.
We have known for more than 200 years that there are three kinds of color
Fig. 5. Three images of the same cone mosaic, showing the benefit of adaptive optics.
signals generated in the eye, and we more recently discovered that these signals originate in three kinds of cone photoreceptors, each with its own photopigment tuned to a particular part of the visual spectrum. We have not, however, had a complete picture of the relative number of these different cone classes or of their arrangement in the cone mosaic. By combining adaptive optics with a method to measure the spectral absorption characteristics of the photopigment in each cone, Austin Roorda and I showed that it is possible to image all three cone types in the living eye.12 Examples of images from two subjects are shown in Figure 6. Interestingly, though one subject has more than three times as many long wavelength sensitive cones as the other, color vision for both subjects is nearly identical, leading to the view that the neural visual system compensates for differences in the relative number of cones so that color vision is not disrupted by variations in the numbers of cones across the retina.
Future work in retinal imaging with adaptive optics will focus on whether this
152 |
D.R. Williams |
|
|
Fig. 6. Pseudocolor images showing the topography of all three types of cones in the living eyes of two subjects. Adaptive optics and retinal densitometry were combined to determine the photopigment in each cone.
Fig. 7. Austin Roorda’s group at the University of Houston has incorporated adaptive optics into a scanning laser ophthalmoscope. The three images show different sections of the retina. The first shows the nerve fiber layer and a blood vessel that runs across the retinal surface. The next panel reveals a second blood vessel and capillaries that run beneath the nerve fibers, and the final panel shows the cone photoreceptors that lie about 300 µm below the surface of the retina. Videos obtained with this instrument of white blood cells flowing through the capillaries are available at: www.opt.uh.edu/research/aroorda/aoslo.htm.
technology can aid in the diagnosis and treatment of retinal disease. For example, the ability to image photoreceptors noninvasively in the eye may help us understand and treat retinal degeneration. Alternatively, the ability to image ganglion cells could accelerate the development of new therapies for glaucoma. The clinical utility of adaptive optics will depend on combining adaptive optics with other technologies, such as confocal microscopy and optical coherence tomography (OCT), that increase the information we can extract from the living retina. Austin Roorda and colleagues at the University of Houston have recently demonstrated an adaptive optics system coupled to a scanning laser ophthalmoscope.13 This device combines the high transverse resolution provided by adaptive optics with the optical sectioning capabilities of confocal microscopy (Fig. 7).
What adaptive optics can do for the eye |
153 |
|
|
The increased contrast and resolution provided by such new imaging technologies will allow retinal disease to be tracked at the level of single cells. Currently, studies of retinal disease at this spatial scale rely on microscopic examination of excised retinas. Today, tracking the course of retinal disease or the development of the normal retina at a microscopic spatial scale requires piecing together evidence from large numbers of excised retinas, whereas microscopic imaging of the living retina could reveal the whole sequence in individual eyes.
Improving vision
In addition to improving retinal imaging, adaptive optics can improve vision. Fundamental studies of the limits of vision with adaptive optics have practical significance in that such studies place an upper boundary on how much improvement in vision is realistic with customized correction of vision with, for example, laser refractive surgery or contact lenses. Liang, Miller, and I showed that adaptive optics can increase contrast sensitivity at high frequencies by as much as sixfold in monochromatic light. Subsequently, Geun-Young Yoon and I also showed that in broadband light, which is more characteristic of normal vision, these increases are much more modest, about a factor of two in the subjects we measured.14
Visual acuity improvements can also be demonstrated though they are small. In the normal eye, a host of factors including chromatic aberration and neural factors conspire to limit the visual benefit of correcting all the monochromatic aberrations.15 On average, the benefit is equivalent to about one-third of a diopter with a 6-mm pupil.16 Some eyes already have such exquisite optics that there would be no benefit at all. But for eyes with substantial amounts of higher-order aberrations, such as those following conventional refractive surgery, their correction remains an important goal and customized correction methods will offer substantial visual benefit to this patient population. My laboratory has been working closely with Bausch and Lomb and Dr. Scott MacRae on the developments of customized refractive surgery.
Phoropters equipped with adaptive optics
Another application in which I anticipate that adaptive optics will appear is the phoropter. For conventional refraction of the eye, current phoropters require a large number of sphere and cylinder lenses and the capability to rotate each into place. In principle, a single deformable mirror could perform the same function and correct higher-order aberrations as well as defocus and astigmatism. Moreover, if also equipped with a wavefront sensor, the phoropter could automatically measure and correct the eye’s wave aberration, much as the current Rochester adaptive optics system does. Obviating the need for the lengthy testing that occurs with the standard subjective refraction methods, adaptive optics could converge on the best correction automatically in a small fraction of a second. The patient would almost instantly be presented with the clearest possible view through the AO system, which the practitioner could confirm, if necessary, with a suitable acuity test. Lawrence Livermore National Laboratories, in collaboration with the
154 |
D.R. Williams |
|
|
University of Rochester, has recently constructed such a phoropter and deployed it at the University of California, Davis.l7
A current limitation of these devices is the high cost of deformable mirrors. For example, the current mirror in the Rochester instrument costs approximately $100,000. A number of companies, however, are working on low-cost alternatives, including Hamamatsu Corporation (Bridgewater, NJ), which has developed spatial light modulators that can be used for wavefront control, and Boston Micromachines (Watertown, MA), which has developed a MEMS (Micro-Electro-Mechanical Systems) mirror that Nathan Doble has successfully tested in the Rochester adaptive optics system.l8 Mirrors based on MEMS technology (similar to that used to fabricate integrated circuits) could be as inexpensive as $1000 per mirror in production, which would make adaptive optics affordable in clinical instruments.
Using adaptive optics to blur the retinal image
Despite the recent proliferation of wavefront sensors in clinical settings, we actually know much less than we need to about the role that aberrations play in visual performance. These wavefront sensors generally decompose the wave aberration into individual Zernike modes. They also provide a physical measure of the severity of each patient’s wave aberration in the form of the RMS wavefront error (the square root of the sum of the squares of the deviation of the actual wavefront from the ideal wavefront). However, Figure 8 illustrates that RMS wavefront error is a poor metric for estimating subjective impact of the wave aberration. Shown are the retinal images of the letter E for three hypothetical eyes: one suffering only from defocus, one suffering from spherical aberration, and one suffering from the sum
Fig. 8. The subjective blur produced by aberrations is not always predicted by the RMS wavefront error. Shown on the left is the letter E blurred with defocus. The middle image shows the same stimulus blurred by spherical aberration. If these two aberrations are added together, as shown on the right, they tend to cancel subjectively, leaving a better image even though this image is blurred with a wave aberration having the greatest RMS. Other pairs of aberrations also cancel in this way, such as coma/secondary coma and astigmatism/secondary astigmatism.
What adaptive optics can do for the eye |
155 |
|
|
Fig. 9. The letter E blurred with wave aberrations corresponding to each of the Zernike modes in orders 2 to 5 at the same RMS. The subjective blur for letters along the edge of the pyramid is less than that for letters in the center of the pyramid. This illustrates that aberrations vary in their potency, and that aberrations in the center of the pyramid are especially troublesome.
of defocus and spherical aberration in the same amounts as present in the first two eyes. Strikingly, though the RMS wavefront error is highest in the third eye, the image quality is obviously best. Several laboratories are now working on methods to predict how much subjective blur to expect from any given wave aberration. The goal is to develop a metric implemented in each wavefront sensor that will tell the clinician how good or bad a particular wave aberration actually is. The development of such a metric depends on psychophysical measurements of the blur produced by different wave aberrations.
Ray Applegate, OD, PhD, is using one approach to this problem in which he is simulating on a computer the retinal images of Snellen targets blurred with individual Zernike modes.19 He has shown that the aberrations along the edges of the Zernike Pyramid are less potent at blurring the retinal image than aberrations in the center of the pyramid (Fig. 9). The intuition behind this result is that the modes along the edge of the pyramid (Fig. 10) have large regions, especially at the pupil center, where the wave aberration is flat, whereas those modes in the center have nonzero slope over a larger fraction of the pupil. At Rochester, Li Chen and I have adopted a complementary approach. We use our deformable mirror to introduce wave aberrations into the eye instead of correcting them. For example, we can correct all the aberrations in a patient while at the same time introducing specific amounts of any particular aberration or sum of aberrations we choose. At ARVO
156 |
|
|
|
D.R. Williams |
||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Fig. 10. The pyramid showing each of the Zernike modes in radial orders 2 to 5, along with their names. Tip, tilt, and piston, which would normally cap the pyramid, have been excluded because they do not influence image quality.
last year, we reported our first psychophysical measurements of the subjective blur of individual Zernike modes with this technique and introduced a biologically plausible metric for computing subjective blur from the eye’s point spread function that quantitatively predicts Applegate’s conclusions based on simulations.20 Adaptive optics has potential uses in a large number of ophthalmic applications from research instruments to explore the impact of aberrations on vision, fundus cameras, surgical microscopes, and phoropters, to name a few. Just how useful this technology will ultimately be in the clinic depends in part on how robust and inexpensive we can eventually make instruments that incorporate adaptive optics. Progress in this area has been rapid and I anticipate that many ophthalmic instruments will eventually come equipped with some form of adaptive wavefront control, and that some people, especially refractive surgery patients, will see better as
a result.
Acknowledgments
This contribution has been published earlier in “Review of Refractive Surgery”, volume 3, number 3, August 2002. Reprinted with the permission of the publisher.
Williams’ laboratory is supported by Bausch and Lomb, National Institutes of Health grants
What adaptive optics can do for the eye |
157 |
|
|
EY01319 and EY07125 and the National Science Foundation Science and Technology Center for Adaptive Optics, managed by the University of California at Santa Cruz under cooperative agreement No. AST-9876783.
References
1.Cashell GTW: A short history of spectacles. Proc Roy Soc Med 64:1063-1064, 1971
2.Rubin ML: Spectacles: past, present and future. Surv Ophthalmol 30:321-327, 1986
3.Von Helmholtz H: Helmholtz’s Treatise on Physiological Optics. Southall JPC (ed). New York, NY: Dover, 1896
4.Smimov MS: Measurement of the wave aberration of the human eye. Biofizika 6:687-703, 1961
5.Liang J, Grimm B, Goelz S, Bille J: Objective measurement of the wave aberrations of the human eye using a Shack-Hartmann wavefront sensor. J Opt Soc Am A A11:1949-1957, 1994
6.Liang J, Williams DR: Aberrations and retinal image quality of the normal human eye. J Opt Soc Am A 14:2873-2883, 1997
7.Liang J, Williams DR, Miller D: Supernormal vision and high-resolution retinal imaging through adaptive optics. J Opt Soc Am A 14:2884-2892, 1997
8.Babcock HW: The possibility of compensating astronomical seeing. Pub Astr Sco Pac 65:229236, 1953
9.Max CE, Macintosh BA, Gibbard SG, Gavel DT, Roe HG, de Patier I, Ghez A, Acton S, Stomski P, Wizinowich P: Cloud structures on Neptune Observed with Keck Telescope Adaptive Optics. Astronomical J 125:364, 2003
10.Dreher AW, Bille JF, Weinreb RN: Active optical depth resolution improvement of the laser tomographic scanner. Appl Opt 24:804-808, 1989
11.Hofer H, Chen L, Yoon GY, Singer B, Yamauchi Y, Williams DR: Improvement in retinal image quality with dynamic correction of the eye’s aberrations. Optics Express 8:831-643, 2001
12.Roorda A, Williams DR: The arrangement of the three cone classes in the living human eye. Nature 397:520-522, 1999
13.Roorda A, Romero-Borja F, Donnelly WJ III, Queener H, Hebert TJ, Campbell MCW: Adaptive optics scanning laser ophthalmoscopy. Optics Express 10:405-412, 2002
14.Yoon GY, Williams DR: Visual performance after correcting the monochromatic and chromatic aberrations of the eye. J Opt Soc Am A 19:266-275, 2002
15.Williams DR, Yoon GY, Guirao A, Hofer H, Porter J: How far can we extend the limits of human vision? In: MacRae SM, Krueger RR, Applegate RA (eds). Customized Corneal Ablation: The Quest for SuperVision, pp 11-32. Thorofare, NJ: Slack, Inc 2001
16.Guirao A, Porter J, Williams DR Cox, I: Calculated impact of higher-order monochromatic aberrations on retinal image quality in a population of human eyes: erratum. J Opt Soc Am A 19:62D-628, 2002
17.Bames TB, Thompson CA, Sawyer RM, Silva DA, Wilks SC, Olivier SS, Wemer JS: A highresolution adaptive optics system for vision science, abstr #957. Association for Research in Ophthalmology Annual Meeting, Ft. Lauderdale, FL, 2002
18.Doble N, Yoon GY, Chen L, Bierden P, Singer B, Olivier S, Williams DR: The use of a microelectromechanical (MEMS) mirror for adaptive optics in the human eye. Optics Letters, 2002
19.Applegate RA, Sarver EJ, Khemsara V: Are all aberrations equal? J Refract Surg, Vol 18(5) Sept/Oct 2002
20.Williams DR: Assessment of optical aberrations of the eye: wavefront sensing and adaptive optics, abstr #913. Association for Research in Ophthalmology Annual Meeting, Ft. Lauderdale, FL, 2002