Tscherning-type aberroscope. Aberroscopy includes indirect ophthalmoscopy of a pattern of laser spots that have been projected through the mask (M) to the retina. The aberroscope lens (L1) images the mask perforations onto the retina. L2 is the ophthalmoscope lens, and L3 images the virtual image of the retinal spots onto the low-light camera (CA) (S indicates shutter; CO, collimator).
Typical spot pattern before (left) and after (right) photorefractive keratectomy. The deviations of the spots from the ideal rectangular pattern are proportional to the gradients of the wavefront aberration. The postoperative spot pattern is obviously more distorted, indicating an increase in wavefront aberrations.
Preoperative and postoperative mean-square wavefront error as a function of the diameter of the entrance pupil. The surgery-induced increase was statistically significant for pupil diameters greater than 5 mm (asterisk indicates P<.05). PRK indicates photorefractive keratectomy.
Seiler T, Kaemmerer M, Mierdel P, Krinke H. Ocular Optical Aberrations After Photorefractive Keratectomy for Myopia and Myopic Astigmatism. Arch Ophthalmol. 2000;118(1):17-21. doi:10.1001/archopht.118.1.17
To study the effects of photorefractive keratectomy on ocular optical aberrations and to establish correlations with glare vision and low-contrast vision.
Preoperative ocular aberroscopy of 15 eyes undergoing photorefractive keratectomy was compared with aberroscopy at 3 months postoperatively by means of a newly developed automated aberroscope of the Tscherning type. The correlation of the wavefront errors with best spectacle-corrected visual acuity, low-contrast visual acuity, and visual acuity under glare conditions was analyzed.
In any individual treated, the total wavefront error increased. On average, the total wavefront error increased by a factor of 17.65; this increase was highly statistically significant (P = .001). Also, the correlation with best-corrected visual acuity, low-contrast visual acuity, and glare visual acuity was statistically significant (P = .02, P = .001, and P = .03, respectively). The increase in ocular aberrations was significantly related with the virtual pupil size.
Photorefractive keratectomy increases the ocular aberrations, impairing the visual performance of the eyes treated. In detail, scotopic visual measures such as low-contrast visual acuity and glare visual acuity suffer most from the myopia correction. Aberroscopy-guided photorefractive keratectomy may avoid such effects.
ON THE BASIS of the mosaic structure of the retina, the visual acuity (VA) of the human eye could be 20/10 or better. "Subpixel" image processing in the inner retina should even increase VA; however, such good acuity is rarely obtained. Two optical conditions account for this reduction in VA: diffraction originating from the entrance pupil and monochromatic aberrations of the optic apparatus of the eye.1
The limitation of VA caused by diffraction decreases with increasing pupil diameter (Table 1) and may play an important role only for pupils smaller than 2 mm.1 The optical errors of higher order (aberrations) of the human eye, however, demonstrate an opposite behavior and may increase with larger pupil diameters. On the other hand, the shape of the human cornea and lens is naturally designed in a way that these aberrations are minimized, for example, by decreased optical power of the peripheral cornea compared with the center. To our knowledge, the monochromatic aberrations of the human eye so far have not been studied systematically in larger series of individuals2 and, therefore, average values for a standard population are not available.
Refractive surgery for myopia and astigmatism, such as radial keratotomy, photorefractive keratectomy (PRK), and laser in situ keratomileusis, induces a nonphysiological corneal shape with a flat central area and increasing power toward the periphery. This shape induces an increase in optical aberrations3- 5 and may lead to visual losses that are detected under low lighting conditions6 and by low-contrast VA testing.7 These side effects of corneal refractive surgery have the potential for public health problems of a yet unknown dimension.
In this investigation, the aberrations of the human eye were measured preoperatively and compared with those 3 months after PRK for myopia and myopic astigmatism. Because of easier interpretation, we prefer to represent the results in terms of Zernike coefficients, which are standard in optical engineering.8
Fifteen patients scheduled for myopic PRK or photoastigmatic refractive keratectomy (PARK) were enrolled in this prospective study. Other than the refractive error, no ocular or systemic disease was diagnosed; specifically, no patient with diagnosed or suspected keratoconus was included. Best spectacle-corrected VA (BCVA) was 20/25 or better in all eyes. All individuals were aged 18 years or older. The spherical equivalent of the preoperative refraction (±SD) was −4.81 ± 1.21 diopters (D) (range, −2.25 to −7.0 D), and the refractive astigmatism was −0.83 ± 0.8 D (range, −0.25 to −2.5 D). In addition, 10 emmetropic persons without ocular disease and with an uncorrected VA of 20/20 or better served as a control group to study the reproducibility of the aberroscopy.
Written informed consent was obtained from each patient scheduled for refractive surgery after a thorough explanation of the risks and benefits of the operation. Also, the members of the control group and the study group were informed about the potential risks of aberroscopy according to the recommendations of the ethical committee of the Universität Klinikum Dresden, Dresden, Germany, which approved the project.
The laser treatment was performed after manual debridement of the epithelium and careful cleaning of the Bowman membrane. The excimer laser beam was centered halfway between the center of the entrance pupil and the first Purkinje image by means of 2 crossed helium-neon laser beams. The excimer laser used in all patients (Multiscan; Schwind, Kleinostheim, Germany) provided rotating masks for myopia correction and a scanning slit for astigmatism correction. The ablation zones had a diameter of 6.0 mm in most eyes and 6.5 or 7.0 mm in patients with larger pupils.
Postoperative medical treatment included antibiotic ointment (ofloxacin) until the epithelium was healed, and then 0.1% fluorometholone eyedrops 4 times a day for 2 months and tapered during the following months.
The aberrations were measured by an aberroscope of the Tscherning type9 (Figure 1). The collimated parallel beam of a green helium-neon laser (λ = 543 nm, 0.5 mW) illuminated a mask with a 9 × 9 rectangular pattern of round perforations. After passing the aberroscope lens (typically +3.0 D), the laser spots had a diameter of approximately 0.5 mm and a rectangular distance of 1 mm at the cornea. The alignment and centration on the pupil center was achieved by a fixation target in the center of the mask and an additional video camera (not depicted in Figure 1). This spot pattern at the retina was imaged by 2 lens systems onto the sensor array of a low-light charge-coupled device camera (LH 750 LL; Lheritier SA, Clermont-Ferrant, France). The imaging path was cropped by an aperture to guarantee paraxial imaging (diameter, 1 mm). The measurement itself took 65 milliseconds. In both groups, aberroscopy was performed with a dilated pupil (diameter, 7 mm or more in all cases).
To determine the reproducibility of the measurements, from each of the 10 eyes of the control group, 10 separate and independent frame captures were taken in 1 session. From these 10 grids per eye, the repeatability of the total wavefront error and the Zernike coefficients was determined. The eyes of the study group were investigated before surgery and at the 3-month follow-up. The images from the retina were processed and stored in a personal computer.
Preoperatively and at 1 and 3 months after surgery, a complete ophthalmic examination was performed in all individuals in the study group, including uncorrected VA and BCVA, autorefraction (Model 595; Humphrey, San Leandro, Calif), glare and low-contrast VA, corneal topography (C-Scan; Technomed, Baesweiler, Germany), applanation tonometry, slitlamp, and fundus inspection.
With the assumption that the paraxial space of the eye is aberration-free and using the Gullstrand eye model, we calculated the position of spots from rays passing the cornea at 0.5 mm from the pupil center. With this distance as grid constant, the distortion-free reference grid was determined. From the deviations of the retinal spots from these grid points, the gradient of the wavefront aberration in the exit pupil at the corresponding corneal positions was calculated. By using this mathematical model, we approximated the wavefront aberration by means of Zernike polynomials up to the sixth order. Only Zernike coefficients C6 to C27 were taken into consideration to describe higher-order wavefront aberrations (third to sixth order). To accomplish this, at least 28 spots in the retinal image must be identified by specifically developed software with the use of fuzzy logic algorithms (an example is shown in Figure 2). A retinal spot pattern was accepted if at least a 6 × 6 matrix was detected. If more spots were available, we used all additionally identifiable spots to perform the calculations.
From the Zernike coefficients we calculated the mean-square wavefront error for comalike aberrations (third-order component S3 and fifth-order component S5) and sphericallike aberrations (fourth-order component S4 and sixth-order component S6). Because of the linear independence of these terms, the total wavefront error was computed by summing all components.
For the evaluation of the wavefront error dependence on pupil size, the order of polynomials was reduced (up to fifth order) for all pupil sizes (4 to 7 mm) to guarantee an equivalent approximation.
Zernike coefficients and wavefront errors were compared statistically by means of the 2-tailed paired t test. Clinical variables were correlated with Zernike coefficients and wavefront errors with parametric and nonparametric tests, and these correlations were tested for statistical significance, which was accepted at P<.05.
On average, BCVA decreased from 1.09 to 0.98. One eye gained 1 Snellen line, 4 eyes lost 1 Snellen line, and 1 eye lost 2 Snellen lines because of irregular astigmatism (eccentric scar). Average low-contrast VA decreased slightly from 0.75 to 0.68, including a 2-line increase in 1 eye and a 2-line decrease in 1 eye. Also, average glare VA decreased slightly from 0.66 to 0.61. Two eyes gained 2 lines and 4 eyes lost 2 lines.
The spherical equivalent was reduced from preoperative −4.81 ± 1.21 D to +0.41 ± 0.7 D at 3 months after PRK. The preoperative cylinder of −0.83 ± 0.8 D in 8 eyes was reduced to −0.21 ± 0.3 D at 3 months after surgery. This is equivalent to an average reduction of 76% of the cylinder.
In the control group, the total wavefront error's SD (10 independent measurements) was greater than λ/8 in none of the 10 eyes. The variance of the individual Zernike coefficients C6 to C27 was on average ±8.9%, yielding sufficient reproducibility.
The total wavefront error was significantly increased by a factor of 17.65 and for each order by factors ranging from 9.13 to 39.83 (Table 2). The statis tical significance is remarkable, since it is valid despite a high variance in preoperative and postoperative values. However, the wavefront error increased in each individual because of surgery. On the other hand, in some eyes the preoperative wavefront error was greater than in other eyes after surgery. The preoperative and postoperative Zernike coefficients are listed in Table 3. A statistically significant increase was determined for 4 coefficients: C8, C12, C18, and C20.
To study the clinical relevance of these results, the correlation of the increase of the total wavefront error with BCVA, low-contrast VA, and glare VA was analyzed. The correlation coefficients were −0.50 (BCVA), −0.89 (low-contrast VA), and −0.52 (glare VA) with significance levels of .02, .001, and .03, respectively. Therefore, the wavefront error increase resulting from surgery was highly significantly correlated with the loss in low-contrast VA and significantly correlated with the loss in BCVA and glare VA.
The total wavefront error before and after refractive surgery was strongly dependent on the diameter of the virtual entrance pupil (Figure 3). In the preoperative eyes, pupillary dilation from 4 to 7 mm caused a 9-fold increase in total wavefront error, whereas after surgery, the same dilation caused a 13-fold increase. The increase caused by surgery was statistically significant for pupil diameters greater than 5 mm.
The refractive outcome of laser refractive surgery for myopia has gained a remarkable level of efficacy and a fair degree of safety. In corrections up to −6.0 D, PRK demonstrated a refractive success rate of 80% and more,10 and, more recently, laser in situ keratomileusis for corrections up to −10.0 D yielded similar results.11 On the other hand, poor vision and glare under low lighting conditions is a frequent complaint of patients after any type of myopia correction. Two sources have been discussed to account for this reduced visual performance after refractive surgery for myopia: light-scattering structures in the treated cornea, such as haze and scars, and optical aberrations that are more relevant with wider pupils. Increased glare disability after radial keratotomy might be attributed to forward light scattering by the radial scars; however, clinical studies were unable to find a significant effect of radial keratotomy on glare disability.12,13 Glare disability and reduction in contrast sensitivity seems not to be correlated with the severity of subepithelial haze after PRK7,12,14 but has been shown to be significantly correlated with the amount of attempted correction.6,14 Also, other clinical results point toward the optical aberrations being one of the main causes of reduced visual performance after myopia correction: postoperative glare is significantly correlated with the ablation zone used15 and the loss in low-contrast vision is correlated with the amount of decentration of the ablated zone.7
This suspected relation of optical aberrations of the eye after refractive surgery and loss in glare and low-contrast VA has to be substantiated by correlation analysis. We found a highly significant correlation of the wavefront error with low-contrast VA (P = .001) and a significant correlation with BCVA (P = .02) and with glare VA (P = .03). These results may indicate how strong the influence of optical aberrations on visual performance is after PRK.
Many high-order optical errors of optical systems are strongly dependent on the width of the exit pupil. Regarding the human eye, it is therefore expected that the aberrations may increase with the diameter of the pupil. This relationship is depicted in Figure 3, which compares the optical performance of the treated eyes at different virtual pupil diameters and may explain the glare and halo problems of the patients at low lighting conditions, where generally wider pupils occur. In a retrospective study, Martinez and coworkers5 demonstrated a significant increase of corneal optical aberrations after PRK based on analysis of corneal topography alone. The authors also reported a significant dependence on virtual pupil diameter that was even greater than our results. Oliver and coworkers16 demonstrated up to a 3-fold increase in optical corneal aberrations after PRK and up to a 10-fold increase in corneal aberrations when the pupil was dilated from 3 to 5.5 mm. Unfortunately, these results cannot be compared with the data presented herein and those of Martinez et al5 because, in the report of Oliver et al, the order of the Zernike coefficients remained undefined. On the other hand, optical aberrations do not originate only from the outer surface of the cornea but also from other refractive elements, such as posterior cornea and the lens. Therefore, aberroscopic data may be more appropriate to describe the overall aberration of the eye that may impair vision.
The repeatability of our measurements of the ocular aberrations is better than alternative methods by approximately a factor of 3.17 Hitherto, most methods depended on the subjective visual perception of the individuals examined. The method presented herein, however, does not rely on a response of the patient and may be considered more objective. The only subjective variable of influence is the centration of the illumination grid with reference to the pupil. Atchison and coworkers,17 however, demonstrated that this measurement error is of minor importance.
During the past decade, PRK technique and algorithms continued to improve. Most important, the diameter of the ablation zone increased from 4.0 mm to 6.0 mm or more. Today, treatment zones with diameters up to 7.0 mm in patients with large pupils are standard. In the patients described, ablation zones of 6.0 mm or more in diameter were used, and the postoperative optical aberration increase reached statistical significance at pupil diameters of 6.0 mm or more, in contrast to the results of Martinez et al5 and Oliver et al,16 where increased corneal optical aberrations even for 3-mm pupils were found. This difference may be explained by a greater variance of preoperative ocular aberrations originating from refraction surfaces other than the outer corneal surface. Also, the treatment zones used by Martinez et al were as small as 5 mm, creating aspheric aberrations at smaller pupil sizes.
To avoid reduced optical performance of the eye after PRK, an aspheric algorithm including a paracentral correction of the mean induced aberrations could certainly help to reduce surgery-induced aberrations. As an alternative, however, the preexisting individual ocular aberrations could be corrected by using computer-assisted PRK or laser in situ keratomileusis combining spherocylindrical PRK with the correction of the wavefront error. Such a customized PRK, however, is not limited to myopic PRK but may also apply to emmetropic eyes aiming toward a VA exeeding 20/20 and not impaired by optical aberrations according to the values listed in Table 1. Currently we are conducting a trial on aberroscopy-guided laser in situ keratomileusis in our department by using a flying spot laser by means of the computer-assisted input fed with the adjusted individual wavefront deviation. This type of aspheric PRK is in contrast to the currently used multizone treatment in tailoring a paracentral overcorrection. Conventional multizone treatment, however, uses a paracentral undercorrection and thus substantially increases the optical aberrations, as demonstrated by Oliver and coworkers.16
Accepted for publication August 5, 1999.
Reprints: Theo Seiler, MD, PhD, Augenklinik, Fetscherstr 74, 01307 Dresden, Germany.