Refractive surgery by excimer laser ablation continues to develop rapidly. In particular, progressive change has occurred in the methods of beam delivery, from early wide-beam systems using a sequence of apertures or variable diaphragms to control the local ablation exposure, to narrower-beam scanning slit and spot systems. One question of considerable interest is the smoothness and accuracy of the ablated stromal profile produced by any system.1 Some aspects of the roughness of the ablated surfaces have been explored using polymethylmethacrylate (PMMA) as a model for the corneal stroma.2"8 These studies have demonstrated that the quality of the surface produced varies with the laser system used and is generally worse than that of conventional man-made optical surfaces. Less attention has been focused on the question of whether the degree of surface irregularity observed is sufficient to cause significant degradation of vision. We consider here as an example the quality of ablation produced by the Nidek EC-5000 excimer laser system (Nidek Technologies, Gamagori, Japan) in terms of both conventional measurements of surface finish and the effects of the surfaces on vision.
MATERIALS AND METHODS
LASER SYSTEM AND PMMA PLATES
The Nidek EC-5000 excimer laser system uses a Gaussian scanning-slit micro-beam with an expanding diaphragm to control the ablation profile. Systems located in five independent centers were used by Nidek-trained technicians to ablate a series of lenses on 2 -mm thick plane-parallel, calibration plates of PMMA. The Nidek EC-5000 excimer lasers were all DOS-based systems; none were the later Windows-based or current NAVEX systems. In each case, the ablated lenses included a series of negative lenses (intended values -2.00, -4.00, -6.00 and -8.00 diopters [D]) and a similar series of positive lenses (+2.00, +4.00, +6.00, +8.00 D). Five examples each of +3.00 D and -3.00 D lenses were also included. The nominal diameter of the negative lenses was 5.0 mm and that of the positive lenses was 7.0 mm, no transition zones being attempted. Focimetry established that the actual lens powers were close to those intended.8
Surface roughness was measured with two instruments (Talysurf, Rank Taylor Hobson, Leicester, UK; Perthometer S8P, Mahr, Göttingen, Germany) in each of which a mechanical stylus scanned the surface of the ablated lens. The Perthometer S8P was used to provide quantitative roughness data on each lens. Because ablation depth, and possibly surface roughness, varied across the area of each lens, an average estimate of roughness over the central area of the lens was required. For the negative lenses, surface profile was examined along 10 parallel, equally spaced tracks covering the central 2X2 mm of each lens. The instrument yields several measures of surface roughness, expressed in terms of the departures of each height measurement from a best-fit reference curve: the latter is effectively a running average of the local surface heights. For the positive lenses a slightly larger area, 4X4 mm, was scanned. Measurements were also made on the flat, unablated surface of the PMMA and, for comparison, on a piano glass surface and the flat calibration surface supplied with the instrument.
CONTRAST SENSITIVITY AND ACUITY MEASUREMENTS
The principle here was that, if the ablated lenses were optically perfect, neutralizing them with a trial lens whose power was of the same magnitude but opposite sign should yield a piano combination, which, when placed in front of the eye, produced no degradation of contrast sensitivity or acuity. If, however, the ablated lens introduced significant aberration or light scatter, it would degrade visual performance even when its power was neutralized.
Achromatic, sinusoidal gratings were generated on a VGA monitor under computer control. Screen subtense at the viewing distance of 4 m was 2.6×3.6° and the space-averaged screen luminance was 20 cd/mp 2. Gratings were vertical and had spatial frequencies spaced at approximately equal logarithmic intervals, ie, 2.0, 2.8, 4.0, 5.7, 8.0, 11.3, 16.0, and 22.6 c/°. The gratings were viewed monocularly with an eye under cycloplegia (1.0% cyclopentolate) through a centered combination of the ablated lens, a neutralizing glass trial lens of equal but opposite power to the measured power of the ablated lens, an artificial pupil of diameter 3.75 mm (chosen to mimic a typical photopic pupil diameter), and a further lens chosen to correct the refractive error of each subject for the viewing distance in use: in practice the powers of the neutralizing lens and the refractive correction were combined into a single lens. For each lens combination, spatial frequencies were tested in random order and contrast thresholds established by a staircase procedure under computer control. Because some loss in contrast sensitivity might occur due to reflections from lens surfaces, a "piano" condition, in which the neutralizing lens and the ablated PMMA lens were replaced by a plane glass lens and an unablated sheet of PMMA, was also tested for reference purposes.
To ensure that contrast threshold criteria were unchanged, the full contrast sensitivity function (ie, the reciprocal of the contrast threshold as a function of spatial frequency) was measured first with the "piano" combination, then with the negative ablated lens combinations in random order, followed by a further "piano" test. A second similar session consisted of runs with the positive ablated lenses sandwiched between runs with the "piano" combination. Due to the time involved, contrast sensitivity functions were measured through only one lens of each power, as produced by a single Nidek EC-5000 system. The lenses used were from a randomly selected center (the sample used was from the Bristol clinic from our study).
To supplement the information gained by the contrast sensitivity measurements, Snellen acuity was measured for the same lens combinations, using a conventional 90% high-contrast chart and normal clinical testing conditions.
The authors (S. A.N., age 33 years, refraction +0.75 DS and W.N.C. , age 64 years, piano) acted as subjects. Monocular best spectacle-corrected visual acuity was 6/4 for S.A.N, and 6/4+ for W.N.C. Vision was otherwise normal and neither subject had undergone refractive surgery or had any ocular or systemic diseases.
Figure 1A shows a typical Talysurf recording of the central region across a diameter of a negative ablated lens, corresponding to a myopic correction. It is obvious that two types of surface error may be present: a larger-scale departure of the general form of the surface from the ideal sphere and small-scale, local irregularities or roughness.358 The first of these, corresponding to low spatial frequencies errors in surface form (sometimes called waviness), can be thought of as introducing image degradation as a result of aberration, the second as introducing a loss in image contrast primarily due to scattering. Figure IA suggests that the lateral scale on which the local height variations occur goes down to approximately ^0.1 mm, corresponding to spatial frequencies of the order of 10 cycles/mm. This is confirmed by Fourier analysis of the Talysurf trace (Fig IB). This shows the expected strong frequency components at low spatial frequencies, corresponding to the basic form of the concave surface with some relatively smooth distortions, and then significant components at intermediate spatial frequencies, between approximately 4 and 10 cycles/mm, corresponding to the smaller-scale irregularities of Figure IA. Thereafter the amplitude declines steadily with increasing frequency. There is no marked peak in the Fourier spectrum to indicate any strong, regular periodicity in the surface profile.
Figure 1. A) Talysurf trace of the ablated lens surface. Note that the vertical scale (microns), representing the variation in the depth of the ablation, is greatly magnified with respect to the horizontal scale (mm). B) Fourier analysis of the trace showing the relative amplitudes of the surface roughness at different spatial frequencies (c/mm).
With the Perthometer, deviations in surface height are measured from a best-fit reference curve, therefore the roughness data obtained with this instrument largely relate to the smaller-scale (higher spatial frequency) surface irregularities. The measurements on nominally flat surfaces showed that the PMMA was significantly rougher than either flat glass or the test surface. All statistical comparisons were therefore made with the unablated plane PMMA surface. The Table gives the overall mean results for the five test plates from the different centers. Following Argento et al6 we have chosen to describe roughness in terms of the parameter Ra, the arithmetic mean of the absolute departure of the surface from the reference curve (roughness average). Because a large number of comparisons exist, a probability of 0.1% represents a significant difference by the t test.
Overall Mean Values of the Roughness Average (Ra) for the Various Ablated Lenses as Produced by the Five Different DOS Nidek EC-5000 Systems*
The same mean data are plotted in Figure 2, which shows that the values of the roughness index, Ra, tend to increase with the power of the ablated lens, with values being generally higher for negative lenses. If regression lines are fitted to the mean Ra data of Figure 2, over the negative range of lens powers (0 to -8.00 D), the square of the correlation product moment correlation coefficient Rp 2=0.96 (P<.001), and over the positive range Rp 2=0.85 (P<.01).
Figure 2. Changes in roughness average, Ra, with ablated lens power. Round symbols show mean values from all five centers, triangular symbols represent a single laser system (Bristol). Error bars represent standard deviations.
Figure 3. Contrast sensitivity functions for the A) negative and B) positive lens series for one subject (S.A. N.). The results for the other subject (W.N.C.) were similar.
Also included in Figure 2 are the values of Ra for the lenses from the plate from a single center (Bristol), which were used for the contrast sensitivity function measurements. Note that the changes in Ra with lens power for the single-center case are more erratic than in the case of the mean values across the five centers, presumably due to fluctuations in the individual laser's characteristics during the production of the lenses; this effect also occurred with lenses from the other centers.
The contrast sensitivity function results for the two subjects were similar so only the results for one subject are shown in Figure 3. For each subject, the four trials with the "piano" combination gave statistically identical results and the results of these trials were therefore averaged: the contrast sensitivity function takes the normal photopic form. However, it can be seen that, with the negative series of lens combinations, the contrast sensitivity function for both subjects fell markedly as the lens power increased; with the positive lens series, only minor decreases in contrast sensitivity function were recorded with all lens powers.
The visual acuity results for the two subjects are shown in Figure 4, plotted in terms of the changes in the minimum angle of resolution: a minimum angle of resolution of 1 min arc is equivalent to 20/20 acuity. The results follow a similar pattern to the contrast sensitivity function data in that the negative lens combinations had a greater degrading effect than the positive combinations.
We interpret the observed increase in average roughness Ra with lens power, and the fact that it was greater for negative lenses than positive lenses of the same power, as implying that roughness tended to increase with the depth of the ablation. It is clear that, for negative lenses of fixed diameter, central ablation depth increases with lens power and also that more material will be removed from the central area of a negative lens than for a positive lens. As shown in the Appendix, the mean ablation depth over the measured area increases approximately linearly with both positive and negative lens power but, under the conditions of the experiment, would be expected to be approximately two times greater for the negative than the positive lenses of the same absolute power. Note with respect to Figure 1 that the use of a larger measurement area for the average roughness of the positive lenses has the effect of exaggerating their measured roughness in comparison with that of the negative lenses, as it increases the mean effective ablation depth. Even with this exaggeration, the positive lenses were smoother than their negative counterparts. In ocular ablations, the pupil diameter will be similar whether the original eye was myopic or hyperopic, so that, in this sense, it would be reasonable to compare roughnesses over the same area for both signs of ablated PMMA lens, in which case the relative smoothness of the positive lenses would be even more marked.
Figure 4. Variation in minimum angle of resolution (MAR) (minutes of arc) for highcontrast Snellen letters with the power of the neutralized, ablated lens for the two subjects.
The Ra results are reflected in the contrast sensitivity function data, where reductions in contrast sensitivity function are greater with negative lenses than their positive counterparts. These contrast sensitivity function measurements provide useful information on the impact of the lens imperfections on imagery at various object scales. A convenient way of representing the degradation at any spatial frequency is to divide the contrast sensitivity with the lens combination by that found at the same frequency with the "piano" combination to yield the "relative contrast sensitivity." Unit relative contrast sensitivity means that no degradation is associated with the ablation. Figure 5 plots the results in these terms for the two lens series and subjects. Evidently the losses in contrast sensitivity are substantial for the higher-powered negative lenses and extend over the full measured spatial frequency band. The differences observed between the two subjects probably reflect the fact that contrast sensitivity is lost as a result of both small-scale surface roughness, which is quantified by Ra, and of the larger scale departures of the surfaces of the ablated lens from the ideal spherical form. The wavefront aberrations resulting from the latter will couple with the aberrations of the eyes of individual subjects, and the resultant changes in the contrast of the retinal image, and hence in contrast sensitivity function, will therefore be a function of the aberrations of the individual eye as well as those of the lens. Thus, because only two subjects were used, the recorded levels of degradation can only be regarded as being indicative of the effects that might occur and the results may not be truly representative of mean findings with a larger sample of subjects.
A simple contrast sensitivity-based, single-figure, index of performance loss can be derived by taking, for each lens combination and subject, the mean of the individual losses in log10 contrast sensitivity (ie, log10 contrast sensitivity for the piano combination minus log10 contrast sensitivity for the lens combination) at each spatial frequency. Ideally the mean loss should always be zero. Figure 6 shows the results of this procedure. Results are broadly similar for the two subjects and again show that the surface imperfections of the negative lenses have a greater degrading effect than those of the positive lenses. This behavior is similar to that found in the acuity results (see Fig 4).
Figure 5. Relative contrast sensitivity functions (ie, contrast sensitivity function with an ablated lens divided by the contrast sensitivity function with piano PMMA) with A) negative and B) positive lens combinations for subject S.A.N, and C) negative and D) positive lens combinations for subject W.N.C. The results for both subjects were similar.
Although exact correlation would not be expected due to the effect of the larger-scale "waviness" of the surfaces, the common origins of the measurements of Ra and the degradation in contrast sensitivity, as revealed by the mean loss in log10 contrast sensitivity, can be demonstrated by plotting one parameter against the other for each lens combination (Fig 7) for the full series of lenses. The link between the two parameters is obvious, irrespective of whether the ablated lenses were of positive or negative power. The R2 values for least-squares regression line fits between loss in contrast sensitivity and Ra are 0.79 (P<.01) for subject W.N.C, and 0.91 (P<.001) for subject S.A.N.
It is clear that, as suggested by earlier authors,3"8 the surface imperfections of the ablated PMMA lenses can produce significant losses in visual performance. Moreover, differences in the surface roughness may be produced with different individual lasers of the same manufacturer's model when attempting to produce nominally the same surface ablation pattern3,4,6: these differences may vary with time.4
It is, however, reasonable to ask first whether analogous roughness effects will occur with the ablated stroma and second whether, if these are present, they necessarily have similar consequences for visual performance.
We note first that the quasi-linear scaling of Ra with lens power, and hence with ablation depth, may imply that the roughness arises as a result of proportional fluctuations in the local energy delivery of the laser exposure. These would be present whatever the material of the ablated surface, so that both PMMA and stroma would be similarly affected.
It could be argued that laser ablation has adverse effects on the regularity of the surface of ablated PMMA, which do not occur with the corneal stroma. However, it has been demonstrated that a good optical surface can be produced in PMMA after excimer laser ablation.9,10 Moreover, although temperature rises may occur during laser ablation (up to 4°C in the case of the cornea11), PMMA is known to be stable over a much wider temperature range,12 so it seems unlikely that thermal effects can account for the observed PMMA roughness changes. It therefore seems reasonable to attribute them to the characteristics of the beam delivery system used.
Figure 6. Mean loss in log contrast sensitivity as a function of ablated lens power.
Figure 7. Loss in log contrast sensitivity plotted against corresponding values of Ra. The correlation coefficients for leastsquares fits to the data for the two subjects are -0.80 (W.N.C.) and -0.91 (S.A.N.).
The use of a flat test surface may introduce a bias towards interference from the plume in concave ablations that might exaggerate irregularities when compared to effects with a convex corneal substrate. It is possible that this could contribute to some of the differences noted between positive and negative lenses.
A number of factors, which are not present in PMMA ablation, will tend to reduce the optical and visual effects of irregular refraction and light scattering due to stromal roughness after refractive surgery. In photorefractive keratectomy (PRK), partial smoothing effects occur due to the regenerating epithelium and the presence of the tear film. The small-scale irregularities described in this study have a magnitude of 0.02 to 0.08 µ??, whereas epithelial cells have dimensions up to 10 pm. Similar smoothing by the epithelium and tears will occur after laser subepithelial keratomileusis. It seems unlikely, though, that a combined epithelium and tear film thickness of approximately 0.1 mm will completely mask underlying surface irregularity on the spatial frequency scale of 2 to 10 c/mm (see Fig 1).
The model described in this study does not take into account the effects of a peripheral ablation, which may affect the central cornea in an elastic model and therefore sets some significant limitations in generalizing to an in vivo situation. The effects of wound healing are also likely to have a significant impact but are not accounted for using this model.
During laser in situ keratomileusis (LASIK), the replaced flap has a thickness of approximately 160 pm, it is thus also unlikely to exactly reproduce the smaller scale variations in the shape of the underlying ablated stroma. Therefore, it would not be unreasonable to expect the flap in LASIK to mask minor ablation irregularities. On the other hand, any irregularity in the flap, due to surgical problems for example, will introduce a new source of possible optical degradation. Due to the differences in refractive index across the optical interfaces that are involved, t is the presence of postoperative irregularity in the anterior cornea that is important, with irregularity in the posterior surface of the flap or unablated stroma having little effect. The optical path variations introduced by any surface irregularities in the flap/unablated stroma interface will be reduced by a factor of the order of (1.3 76-1. 336)/(1.49-l) in comparison with the effects at an air-PMMA interface, ie, a factor of approximately 0.08 times.
On the whole, it appears probable that on the eye the effects of light scatter due purely to ablation roughness are likely to be diminished in comparison with the PMMA model, although scatter due to healing may negate this advantage. The aberrational effects of larger-scale irregularities are likely to be similar in the PMMA and eye situations.
Although the experiments investigated the alterations in contrast sensitivity function caused by surface roughness in ablated PMMA, it is perhaps not unsafe to assume that similar (although not identical and perhaps smaller) effects may be occurring in real patients. Certainly, cases of reduced contrast sensitivity function have been reported after PRK and LASIK.13,14 It should be noted that although Nidek EC-5000 systems were used in the work described in this article, our own additional studies and those of other authors show that qualitatively similar, and often larger, effects occur with other excimer laser systems.3,4,6 The EC-5000 lasers used in our work were all of the earlier DOS-based type and have been superseded by later Windows-based and NAVEX systems. The newer operating systems used by the Nidek EC-5000 allow different profiles to be made on the corneal surface and have mechanisms that allow smoothing laser pulses to applied.
The detailed surface topography of the corneas of post-refractive surgery patients need to be explored to establish whether surface roughness is present and, if so, its magnitude after treatments with different types of laser system. This cannot be achieved with conventional videokeratoscopes or aberrometers, due to their limited spatial resolution, but might be achieved with interferometric methods.15
1. O'Donnell CB, Kemner J, O'Donnell FE Jr. Ablation smoothness as a function of excimer laser delivery system. / Cataract Refract Surg. 1996;22:682-685.
2. Assouline M, Mo ossavi J, Muller-Stein wachs M, Cohen-Sabban J, Thompson HW, Pouliquen Y. PMMA model of steep central islands induced by excimer laser photorefractive keratectomy. Surv Ophthalmol. 1997;42:S35-S51.
3. Anschütz T, Piéger S. Evaluation of hyperopic photoablation profiles. J Refract Surg. 1998;14:S192-S196.
4. Anschütz T, Piéger S. Correlation of laser profilometry scans with clinical results, f Refract Surg. 1999;15:S252-S256.
5. Naroo SA, Charman WN. Surface roughness after photorefractive keratectomy (PRK) for the correction of refractive error - simulation using PMMA. European Optical Society Digest Series, Topical Meeting on Physiological Optics, Wroclaw, Poland. 1999;23:23-24.
6. Argento C, Valenzuela G, Huck H, Cremona G, Cosentino MJ, Gale MF. Smoothness of ablation on acrylic by four different excimer lasers. / Refract Surg. 2001;17:43-45.
7. Doga AV, Shpak AA, Sugrobov VA. Smoothness of ablation on polymethylmethacrylate plates with four scanning excimer lasers. iRefract Surg. 2 004;20:S 73 0-S 73 3.
8. Hauge E, Naroo SA, Charman WN. Poly (methyl methacrylate) model study of optical surface quality after excimer laser photorefractive keratectomy. / Cataract Refract Surg. 2001;2 7:20262035.
9. Srinivasan R, Braren B, Dreyfus RW, Headell L, Seeger DE. Mechanism of the ultraviolet laser ablation of polymethyl methacrylate at 193 and 248 nm; laser-induced fluorescence analysis, chemical analysis and doping studies. / Opt Soc Am B. 1986;3:785-791.
10. Jitsuno T, Tokumura K, Nakashima N, Nakatsuka M. Laser ablative shaping of plastic optical components for phase control. Appl Opt. 1999;38:3338-3342.
11. Betney S, Morgan PB, Doyle SJ, Efron N. Corneal temperature changes during photorefractive keratectomy. Cornea. 1997;16:158-161.
12. Lytle JD. Polymeric optics. In: Bass M, Stryland EW, Williams DR, Wolffe WL, eds. Handbook of Optics. Vol 2. New York, NY: McGrawHill; 1995:34.1-34.21.
13. Hodkin MJ, Lemos MM, McDonald MB, Holladay JT, Shahidi SH. Near vision contrast sensitivity after photorefractive keratectomy. J Cataract Refract Surg. 1997;23:192-195.
14. Mutyala S, McDonald MB, Scheinblum KA, Ostrik MD, Brint SF, Thompson H. Contrast sensitivity evaluation after laser in situ keratomileusis. Ophthalmology. 2000;107:1864-1867.
15. Licznerski TJ, Kasprzak HT. Reconstruction of fine corneal topography from interferometric measurements. European Optical Society Digest Series, Topical Meeting on Physiological Optics, Wroclaw, Poland. 1999;23:6-7.
Overall Mean Values of the Roughness Average (Ra) for the Various Ablated Lenses as Produced by the Five Different DOS Nidek EC-5000 Systems*