Abstract
Visual acuity is dependent on corneal shape and size. A minor variation in surface geometry can cause a deformation of corneal geometry, which affects its optical performance. In this work we demonstrate an algorithm for the simultaneous measurement of corneal tomography and topography with a traditional point-scanning Fourier domain optical coherence tomography (FD-OCT) system. A modified wavelength scanning interferometry (mWSI) algorithm enabled topographical evaluation of the surface with nanometer-scale resolution, which is superior to the micrometer-scale resolution of traditional OCT structural imaging. We validated the technique with an optically flat mirror, standard roughness gauges, and atomic force microscopy (AFM). The mirror results show nanometer-scale sensitivity (~3.5 nm), and the mWSI measurements were in good agreement (error ~5%) with the specifications of the roughness comparator and AFM, demonstrating the accuracy of the technique. Following validation, the measurements were made on pig corneas in situ at various artificially controlled intraocular pressures (IOP) and before and after cross-linking (CXL). The results show that the mean surface roughness increased by ~65% after removal of the epithelium in preparation for CXL but did not change as a function of IOP. The demonstrated method could be used for simultaneous measurement of tissue tomography with micrometer-precision and topography with nanometer-precision.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Optical coherence tomography (OCT) has emerged as a powerful tool in various fields of biomedical imaging [1]. OCT is based on low coherence interferometry, where the optical path difference (OPD) between the sample and a reference reflector is used to obtain depth-resolved images with micrometer-scale resolution. While the major application of OCT has been in ophthalmology, its capability for rapid noninvasive imaging with micrometer-scale resolution has expanded to other fields, such as embryology [2], cardiology [3], dermatology [4], gastroenterology [5], oncology [6], and dentistry [7]. The first realization of OCT was time-domain OCT (TD-OCT) [1], which utilized a scanned reference mirror. The development of Fourier domain OCT (FD-OCT) [8] enabled single-shot depth-wise measurements with superior sensitivity [9] by acquiring spectral information, removing the need for a moving reference mirror. However, the resolution of quantitative structural information provided by OCT is limited to the axial resolution of the system, which is generally on the order of few micrometers.
On the other hand, wavelength scanning interferometry (WSI) can perform topographical surface measurements with much finer resolution than OCT [10, 11]. The basic principle of WSI is to shift the phase of the interfered beams by scanning the wavelength instead of altering the optical path difference, such as with traditional phase shifting holography [12]. WSI has the advantage of enhanced stability by overcoming mechanical errors caused by piezoelectric transducers such as hysteresis and nonlinearity, which are inherent to any mechanical scanning method. The key feature of WSI is an acousto-optic tunable filter (AOTF), which selects a specific wavelength from a broadband source, thus producing an interferogram only at that specific wavelength. There have been reports utilizing WSI for topographical evaluation of samples such as integrated circuits and fingerprints via phase reconstruction of wavelength specific interferograms using full-field swept source OCT [13, 14]. However, these setups are limited because the AOTF restricts the imaging speed. Additionally, due to the bandwidth limitations of AOTFs, the imaging depth is typically less than a millimeter.
FD-OCT has been extensively used to assess corneal topography [15]. However, only the OCT structural image was used, again restricting the axial displacement sensitivity to a few micrometers. In this paper, we applied a modified WSI (mWSI) algorithm for simultaneous measurement of tomography and topography of the cornea with a traditional point scanning FD-OCT system. Because the demonstrated algorithm does not utilize an AOTF, this method is only restricted in speed by the FD-OCT system and inherits the relatively long imaging range (generally several millimeters). In addition, the mWSI algorithm is applicable for both SS-OCT and SD-OCT systems and equally effective because each A-line is processed independently.
Corneal shape largely determines the optical performance of the eye and even minor variations in surface geometry can alter visual acuity. Irregularities in corneal shape caused by diseases such as keratoconus or scarring due to trauma can severely degrade or impair vision. Most current refractive surgical procedures, such as UV-A/riboflavin corneal collagen crosslinking (CXL), which is used to treat keratoconus [16], work by modifying corneal curvature, and screening for corneal abnormalities prior to these surgeries relies on some form of corneal topography [17–19]. To resolve clinically meaningful features, corneal topography measurement techniques must be capable of sub-micrometer axial resolution and micrometer lateral resolution over a 10-15 mm lateral field. Because axial and lateral resolution are decoupled in OCT imaging, it is a promising approach for corneal topography measurements. However, early attempts to use OCT for corneal topography were inferior to existing approaches because imaging resolution (primarily axial) was insufficient. Newer approaches have overcome limitations due to imaging speed and axial resolution has improved [20].
In this work, we describe and evaluate a new approach for OCT-based corneal topography using a point scanning FD-OCT system to achieve nanometer-scale axial topographic resolution. We also present results from topographic measurements of an in situ porcine cornea before and after therapeutic interventions that modify corneal shape.
2. Materials and methods
2.1 Experimental procedures
For experimental quantification of the surface topography, we used an optically flat mirror (PF07-03, ThorLabs Inc., NJ, USA), standard surfaces roughness comparators (Models 315 and 316, SPI Microsurf Comparator, MSC Industrial Direct Co., NY, USA). Further validation was performed with an atomic force microscope (AFM) (Bioscope Resolve V, Bruker MA, USA). Please note that AFM validation data are presented in Appendix. Fresh juvenile (4 to 6 months) porcine eyes (Sioux-Preme Packing Co., Sioux City, IA) were used to quantify corneal tomography and topography. The eyes were transported overnight in 4°C saline. Experiments were conducted immediately when the samples were received, which was within 24 hours of enucleation. During the experiments, the whole eye-globes were positioned in a custom holder that was mounted on a stage. After removal of extraneous tissues such as muscles, the eye-globes were cannulated with two needles for artificial intra-ocular pressure (IOP) control [21]. One needle was connected to a micro-infusion pump, and the other needle was connected to a pressure transducer to form the closed-loop IOP control system.
CXL was performed by a technique that mimics the standard “Dresden” protocol [16]. The epithelium was removed with a blunt surgical instrument, and a 0.1% riboflavin solution in 20% dextran was topically applied to the cornea every 5 minutes. After 30 minutes, the cornea was irradiated for 30 minutes (3 mW/cm2, 365 nm). During UV irradiation, the riboflavin solution was topically added to the cornea every 5 minutes. The measurements were taken before the treatment, after the epithelium was removed, and immediately after the CXL procedure.
2.2 Experimental setup
The SD-OCT system used in these studies has been previously described in detail in [22]. Briefly, the system utilizes a superluminescent diode (SLD) with a central wavelength at 840 nm and −3 dB bandwidth of 50 nm. The source light was split by a 50/50 fiber coupler to a reference arm with a mirror and the sample arm. The light in the sample arm was scanned across the sample surface with a pair of galvanometer mounted mirrors. The light returning from the sample and reference arms interfered, and the interference pattern was captured with a home-built spectrometer. In OCT, the axial resolution (Lc) is determined by two parameters: the central wavelength of the light source (λ0) and the bandwidth (Δλ) by
The axial optical resolution of the SD-OCT system was ~6 µm, as measured by the PSF of a mirror, and the transverse resolution was ~8 µm as determined by imaging a resolution target. Each frame took 100 ms with a scanning area of 9.5 mm × 9.5 mm with 2000 × 2000 pixels.
2.3 Modified wavelength scanning interferometry algorithm
Considering the lateral dimensions along the surface as the XY plane and depth as the Z direction, the light intensity can be expressed as:
where is the background intensity, is the fringe modulation, k is the wavenumber, and is the absolute OPD, which is twice the actual path difference. The phase of the interference signal, , can be quantified by:Therefore, the dynamic phase change of the interference signal along with the change of wavenumber is given by:
The phase difference corresponding to wavenumber shift can be used to obtain the OPD, , with:
There are many methods for phase calculation that can be implemented in WSI. These algorithms include phase demodulation by a lock-in amplifier [22], phase calculation by a seven-point method used in classical phase-shifting interferometry [23], extremum position counting [24], and Fourier transform based techniques [25]. Here, we utilize a Fourier transform based technique because it is fast, accurate, and insensitive to intensity noise [25]. In Eq. (2), and define the envelope of the fringe. The envelope slowly varies with respect to k because of the spectral intensity of the light source and spectral response of the detector, which is usually a line-scan CCD camera in an SD-OCT system. The path difference of the interferometer was adjusted to be large enough so that the frequency of the cosine term was greater than the modulation frequency of and for easy separation. Equation (2) can be expressed in exponential form:
By taking the Fourier Transform of Eq. (6) with respect to wavenumber, k, we can filter out the first and last terms using a bandpass filter. The first term is the DC signal, and last term is the complex conjugate of the second term, which can both be removed by a bandpass filter. An inverse Fourier transform can then be applied to reconstruct b, which yields the intensity image of the sample under investigation. Accumulating the intensity images frame by frame, we can construct the 3D structural image. Taking the natural logarithm of the newly constructed b value, we can isolate the phase term, where the imaginary term encodes the phase at each pixel:
To obtain the phase from the sample surface, the location of the surface was identified from the OCT intensity image. Due to generally high contrast at the air-sample interface at the target surface, a strong discontinuity appears in the intensity of each A-line at the surface. Since, the interference is only from the sample surface in conventional WSI, the phase information directly provides the surface information. However, in FD-OCT, each individual A-line corresponds to a single spectral interference signal. Therefore, a Hanning window, matched to the point spread function of the SD-OCT system, was defined to select the surface information. The Hanning window increased the phase calculation robustness by removing system noise and information not relating to the sample surface. The implementation of Hanning window to extract the raw surface information is an additional modification of the traditional WSI algorithm.
After locating the surface of the sample in each A-line, we performed an inverse Fast Fourier Transform (FFT) on each A-line to reconstruct b as described earlier. The phase information of the reconstructed b can be extracted from the imaginary part of Eq. (7). Since the mathematically calculated phase is limited from -π to π, the phase distribution has discontinuities along wavenumber. After unwrapping the raw phase information extracted from Eq. (7) along wavenumber, the phase follows an inclined straight line. The slope of the line signifies the dynamic change of phase with respect to the change in wavenumber as expressed in Eq. (5). Therefore, the OPD at each scanned position of the mirror surface can be calculated from the slope of the line.
The advantage of this algorithm is that the sectioning depth can be obtained over the full OCT imaging range, which is particularly useful for biological samples that have irregular geometry. Additionally, the slope of the unwrapped phase can be positive or negative, so there is no sign ambiguity in the OPD between the sample and reference arms.
3. Results and discussion
To test the sensitivity of our algorithm, we first applied it on the interference pattern obtained from a coverslip with the system in a common-path setup (blocking the reference arm of the SD-OCT system). The resulting sensitivity, as quantified by the standard deviation of the OPD of the self-interference, was 3.5 nm at an SNR of 40 dB. We then imaged a simple flat mirror (PF07-03, ThorLabs Inc., NJ, USA). From the intensity data collected from the SD-OCT system, we first tracked the surface of the mirror.
Surface tracking was performed on each A-line independently based on the median-filtered intensity. The intensity plot of a single frame with the tracked surface (in red) is shown in Fig. 1. After tracking the surface of the flat mirror for each frame, we followed the mWSI algorithm to get the OPD for each A-line. The OPDs from each A-scan in a 3D image were then combined.
One limitation of most OCT systems is the presence of the so-called fan (or field) distortion for a non-telecentric scanning system [26]. In general, the effect is related to the scanning architecture of most point-scanning 3-D OCT systems, where perfectly flat surfaces appear curved [27–29]. This phenomenon is due to at least two possible effects: 1) the architecture of the scanning system, which is primarily affected by the spatial separation of the mirrors, and 2) design, position, and alignment of the collimating lens in relation to the scanning mirrors. As the result, the flat surface of the mirror appears as a curved surface. To remove the scanning distortion, we subtracted a quadratic curve that was fitted along the curved surface for each B-scan.
A 3D surface plot of the flat mirror without correction is shown in Fig. 2(a). The roughness of the mirror is superimposed on the scanning curvature. As mentioned earlier, we used a quadratic fit to replicate the scanning curvature for each B-scan and subtracted it from the corresponding B-scan. After subtracting the scanning curvature from the 3D surface, the remaining data was used to calculate surface roughness. The 3D examples of surface roughness over a small ~100 µm by ~100 µm region of the mirror at three arbitrary locations are illustrated in Figs. 2(b-d).
The statistics of the mirror roughness are provided in Table 1. We discarded values at the periphery of the image due to the low structural image SNR. The effect of SNR on the mWSI algorithm is shown in the Appendix in Fig. 7, where the stability decreases drastically when the OCT image SNR drops below 40 dB. An OCT B-Scan of the optically flat mirror and the corresponding SNR and roughness as calculated by mWSI algorithm are shown in the Appendix in Fig. 8. Due to low SNR (<40dB) at the periphery, the roughness increases drastically compared to the central region.
For further verification of the accuracy of mWSI algorithm, we compared the roughness of known standard surfaces from a roughness comparator (Models 315 and 336, SPI Microsurf Comparator, MSC Industrial Direct Co., NY, USA) to the roughness as calculated by the mWSI algorithm (both measurements in accordance with ISO 4287). We calculated mean roughness value (Ra), which is the average of the absolute values of the profile deviations from the mean line of the roughness. We also calculated the mean roughness depth (Rz), which is the average value of greatest height (peak to valley) of the roughness profiles. The comparison is shown in Table 2. Sample 3D surface roughness plots of three surface comparator samples (N2, N3, and N4 of model 315) are shown in Figs. 3(a-c). The roughness of all comparator samples is plotted in the Appendix in Figs. 9 and 10. The surface comparator results show only a small error in the roughness averaged from all samples (5.1 ± 3.5% for Ra and 1.4 ± 11.8% for Rz). Further analysis of the disagreement between mWSI and the specifications of the surface comparator is presented in the Appendix in Fig. 11 with normalized Bland-Altman plots [30]. The average disagreement was 4.4% and 4.5% for Ra and Rz, respectively. The upper limit of agreement (LoA) for Ra was 9.5%, and the lower LoA for Ra was −0.8%. Similarly, the upper LoA for Rz was 9.5%, and the lower LoA for Rz was −0.5%. These results demonstrate that mWSI has a slight positive bias as compared to the manufacturer specifications, but the normalized disagreement decreased as a function of feature size for Ra and Rz. Imaging of sample N8 of Model 315 also demonstrates the limitation of the technique for rougher samples. If the feature size is above the pixel resolution of the OCT system, the mWSI data is then clipped. Hence, other techniques would need to be utilized in this case. Additional validation was performed with atomic force microscopy (AFM), and the methods and results are provided in the Appendix.
In comparison to existing techniques such as AFM, the presented mWSI method can scan a larger sample area with no contact and much more quickly. However, AFM can provide much finer spatial resolution (angstrom-scale) in contrast to the micrometer (transverse) and nanometer (axial) scale resolution demonstrated by the presented technique. Different groups have proposed surface measurements using optical interferometry, such as absolute distance measurements with a measurement error of 13 nm [31] and large stepped structure with an uncertainty of 45 nm [32]. Moreover, the imaging speed of our method can be further improved by using ultra-fast OCT systems [33].
Next, we performed measurements on an in situ porcine cornea at various conditions. We artificially controlled the eye-globe intraocular pressure (IOP) with a closed-loop feedback system [21]. Imaging was performed at IOPs from 10 to 30 mmHg with 5 mmHg increments under three different treatment conditions: 1) a baseline with no treatment, 2) after removing the epithelium with a blunt surgical instrument, and 3) after traditional CXL [16]. The curvature of the untreated porcine cornea at IOP of 10 mmHg is shown in Fig. 4(a). The surface roughness over a ~100 µm by ~100 µm region from the apex of the porcine cornea at an IOP of 10 mmHg with an intact epithelium is plotted in Fig. 4(b). The roughness of the sample after removing the epithelium and after CXL, also at 10 mmHg IOP, are presented in Figs. 4(c) and (d), respectively, with identical axes and colormap to Fig. 4(b) for better comparison to the sample roughness with the epithelium intact. The roughness of the small areas shown was representative of the total scanned area shown in the figures. Clearly, the roughness as measured by mWSI increased after removal of the epithelium and then further increased in roughness after CXL.
As the presented technique has the advantages of providing both topography and tomography, OCT images of the corneas after each treatment step are also presented. The two-dimensional images of the porcine cornea before the treatment, after removal of epithelium, and after CXL are shown in Figs. 5 (a), (b), and (c), respectively. All the presented images were at an artificially controlled IOP of 10 mmHg. Because of the topical addition of dextran, the cornea thins after CXL [34].
A box plot of the absolute surface roughness amplitude of the porcine cornea at various IOPs and treatment conditions is shown in Fig. 6. The horizontal line is the median, whiskers are the 5th and 95th percentiles, the box is the interquartile range, and the small inscribed box is the mean. The detailed statistical analysis of the surface roughness is given in Table 4 in the Appendix. In Fig. 6, the surface roughness of the cornea increased by ~65% after removal of the epithelium, which is reasonable because of the smooth and homogeneous nature of the corneal epithelium [35]. In addition, the Bowman’s membrane is absent in the pig eye [36, 37], so removing the epithelium exposed the rougher anterior stroma. Even though the collagen fibrils and proteoglycans are uniformly distributed in deeper layers of the corneal stroma, the collagen fibrils are randomly distributed in the anterior section of stroma [38], which explains the increase in roughness as measured by mWSI after removal of the epithelium. AFM of the human cornea revealed that the diameter of corneal collagen fibrils is 81.73 ± 19.88 nm with mean axial D-periodicity of 68.50 ± 10.3 nm [39]. Since the spatial resolution of the SD-OCT system is a few orders of magnitude larger than the diameter of collagen fibrils, we are measuring the bulk phase and intensity information from many collagen fibrils rather than individual fibrils.
Additionally, we postulate that the random orientation of collagen fibrils in the anterior stroma is the primary cause for no measurable difference in the surface roughness as a function of IOP. Even though higher IOPs induce measurable strains on the cornea, they do not alter the surface fibril ultrastructure at the micrometer scale, at least in the IOP range we measured. In our future work, we will assess the roughness with alternative modalities to validate our corneal results as well as higher IOP to measure the influence of pathological IOPs on the corneal surface topography because an elevated IOP is a known risk factor for glaucoma [40].
Previous work has shown that CXL increased the diameter of the collagen fibrils [41] and caused the fibrils to become wavy [42]. Our results show that CXL did not increase the roughness of the corneal surface as compared to a de-epithelialized cornea. As stated previously, this may be due to the fact we are measuring bulk information from many collagen fibrils, so any microstructural changes may not be detectable. Similarly, the tear film can also contribute to the bulk phase information. However, use of an OCT system with finer axial resolution would enable isolation of the tear film to provide information about the corneal surface only [43].
As with all coherence imaging techniques, OCT images are well-known for their speckle pattern. However, the speckles in tissues, such as the cornea, are generally underdeveloped due to various parameters, such as multiple scattering and different angles of incident light [44]. Extensive research has been performed on measuring the roughness of samples from the speckle pattern [45–48]. However, this was only possible after a lengthy calibration of various parameters, such as incident intensity, spot size, and incident angle, and is only applicable to materials such as glass and metal, which have fully developed speckle patterns. Here, we performed further analysis on the OCT intensities from the surface of the cornea (formed by speckles) and the mWSI-measured surface profiles (Fig. 14 in Appendix). The correlation between the surface intensities and mWSI-measured surface profiles, as well as the corresponding spatial frequencies, was incredibly weak (R2<0.05), demonstrating that there was no direct correlation between the speckle in the cornea and the mWSI measurements. However, one has to be cautious by correlating phase measurements using coherence-domain techniques with sample roughness directly as speckle-noise might obscure true surface profile in some situations.
Nevertheless, the proposed algorithm was reasonably accurate as demonstrated by the small error from the manufacturer specifications of a roughness comparator. We also demonstrated surface roughness measurement of an in situ porcine cornea, suggesting the feasibility of applying high resolution topographical imaging of biological samples. Our future work will entail analyzing pathological corneal samples because diseases such as keratoconus alter the corneal fibril arrangement [49–52]. Moreover, we will measure the surface roughness of primate samples with mWSI to investigate the influence of Bowman’s membrane on the surface roughness of the cornea.
4. Conclusion
In this work, we presented a technique for simultaneous tomography and topography with a modified wavelength scanning interferometry technique that can be utilized with traditional point-scanning Fourier domain optical coherence tomography systems. This technique does not require external synchronization as in a traditional WSI system, thus removing restrictions on imaging speed while also preserving the long imaging depth of OCT systems. We validated the mWSI technique by imaging a mirror and a standard roughness comparator. Our results showed good agreement between the manufacturer provided and mWSI measured roughness, demonstrating the accuracy of the technique. We also imaged an in situ porcine cornea at various conditions demonstrating the applicability of mWSI for characterizing the roughness of tissues.
Appendix
Speckle is an inherent property of any coherent imaging technique such as OCT. Imaging of surfaces such as glass and metal do not have any obvious speckle pattern in OCT images. Thus, we imaged a sample capable of producing a clear speckle pattern with AFM (Nanoscope Resolve V, Bruker) to further validate mWSI method. A gelatin matrix was deposited on a microscope glass slide. A hole was punched in the gelatin matrix as a landmark for side by side comparison of mWSI and AFM measurements. Figure 12 shows the region imaged by OCT and AFM. Figure 13 shows the comparison between the AFM-measured and mWSI-measured height profiles and roughness of a region across the glass and gelatin. The results show very good agreement and show the limitations in spatial resolution of the OCT scanning system, which was also unable to resolve the individual collagen fibrils in the cornea samples. A summary of the results is provided in Table 3. The relatively large error in the glass sample may be due to interactions between the AFM tip and the glass.
Funding
National Institutes of Health (R01EY022362).
Acknowledgments
The authors would like to sincerely thank Dr. Vijaykrishna Raghunathan at the University of Houston College of Optometry for his aid with the AFM measurements and Dr. Sean Kirkpatrick at the Michigan Technological University for discussion of speckle statistics related to the surface roughness.
Disclosures
The authors declare that there are no conflicts of interest related to this article.
References and links
1. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Hee, T. Flotte, K. Gregory, C. A. Puliafito, and et, “Optical coherence tomography,” Science 254(5035), 1178–1181 (1991). [CrossRef] [PubMed]
2. R. Raghunathan, M. Singh, M. E. Dickinson, and K. V. Larin, “Optical coherence tomography for embryonic imaging: a review,” J. Biomed. Opt. 21(5), 050902 (2016). [CrossRef] [PubMed]
3. K. K. Galougahi, A. Maehara, G. S. Mintz, R. A. Shlofmitz, G. W. Stone, and Z. A. Ali, “Update on intracoronary optical coherence tomography: A review of current concepts,” Curr Cardiovasc Imag 9(6), 1–10 (2016).
4. M. Ulrich, L. Themstrup, N. de Carvalho, M. Manfredi, C. Grana, S. Ciardo, R. Kästle, J. Holmes, R. Whitehead, G. B. Jemec, G. Pellacani, and J. Welzel, “Dynamic optical coherence tomography in dermatology,” Dermatology (Basel) 232(3), 298–311 (2016). [CrossRef] [PubMed]
5. T. H. Tsai, C. L. Leggett, A. J. Trindade, A. Sethi, A. F. Swager, V. Joshi, J. J. Bergman, H. Mashimo, N. S. Nishioka, and E. Namati, “Optical coherence tomography in gastroenterology: A review and future outlook,” J. Biomed. Opt. 22(12), 1–17 (2017). [CrossRef] [PubMed]
6. J. Wang, Y. Xu, and S. A. Boppart, “Review of optical coherence tomography in oncology,” J. Biomed. Opt. 22(12), 1–23 (2017). [PubMed]
7. Y. S. Hsieh, Y. C. Ho, S. Y. Lee, C. C. Chuang, J. C. Tsai, K. F. Lin, and C. W. Sun, “Dental optical coherence tomography,” Sensors (Basel) 13(7), 8928–8949 (2013). [CrossRef] [PubMed]
8. A. F. Fercher, C. K. Hitzenberger, G. Kamp, and S. Y. El-Zaiat, “Measurement of intraocular distances by backscattering spectral interferometry,” Opt. Commun. 117(1–2), 43–48 (1995). [CrossRef]
9. J. F. de Boer, R. Leitgeb, and M. Wojtkowski, “Twenty-five years of optical coherence tomography: the paradigm shift in sensitivity and speed provided by Fourier domain OCT [Invited],” Biomed. Opt. Express 8(7), 3248–3280 (2017). [CrossRef] [PubMed]
10. H. Muhamedsalih, F. Gao, and X. Jiang, “Comparison study of algorithms and accuracy in the wavelength scanning interferometry,” Appl. Opt. 51(36), 8854–8862 (2012). [CrossRef] [PubMed]
11. S. Kuwamura and I. Yamaguchi, “Wavelength scanning profilometry for real-time surface shape measurement,” Appl. Opt. 36(19), 4473–4482 (1997). [CrossRef] [PubMed]
12. I. Yamaguchi and T. Zhang, “Phase-shifting digital holography,” Opt. Lett. 22(16), 1268–1270 (1997). [CrossRef] [PubMed]
13. T. Anna, C. Shakher, and D. S. Mehta, “Simultaneous tomography and topography of silicon integrated circuits using full-field swept-source optical coherence tomography,” J. Opt. A, Pure Appl. Opt. 11(4), 045501 (2009). [CrossRef]
14. S. K. Dubey, D. S. Mehta, A. Anand, and C. Shakher, “Simultaneous topography and tomography of latent fingerprints using full-field swept-source optical coherence tomography,” J. Opt. A, Pure Appl. Opt. 10(1), 015307 (2008). [CrossRef]
15. S. Ortiz, D. Siedlecki, P. Pérez-Merino, N. Chia, A. de Castro, M. Szkulmowski, M. Wojtkowski, and S. Marcos, “Corneal topography from spectral optical coherence tomography (soct),” Biomed. Opt. Express 2(12), 3232–3247 (2011). [CrossRef] [PubMed]
16. G. Wollensak, E. Spoerl, and T. Seiler, “Riboflavin/ultraviolet-a-induced collagen crosslinking for the treatment of keratoconus,” Am. J. Ophthalmol. 135(5), 620–627 (2003). [CrossRef] [PubMed]
17. A. M. Mahmoud, M. X. Nuñez, C. Blanco, D. D. Koch, L. Wang, M. P. Weikert, B. E. Frueh, C. Tappeiner, M. D. Twa, and C. J. Roberts, “Expanding the cone location and magnitude index to include corneal thickness and posterior surface information for the detection of keratoconus,” Am. J. Ophthalmol. 156(6), 1102–1111 (2013). [CrossRef] [PubMed]
18. A. M. Mahmoud, C. J. Roberts, R. G. Lembach, M. D. Twa, E. E. Herderick, T. T. McMahon, and C. S. Group, “CLMI: The cone location and magnitude index,” Cornea 27(4), 480–487 (2008). [CrossRef] [PubMed]
19. M. D. Twa, J. J. Nichols, C. E. Joslin, P. S. Kollbaum, T. B. Edrington, M. A. Bullimore, G. L. Mitchell, K. J. Cruickshanks, and D. J. Schanzlin, “Characteristics of corneal ectasia after LASIK for myopia,” Cornea 23(5), 447–457 (2004). [CrossRef] [PubMed]
20. K. Karnowski, B. J. Kaluzny, M. Szkulmowski, M. Gora, and M. Wojtkowski, “Corneal topography with high-speed swept source OCT in clinical examination,” Biomed. Opt. Express 2(9), 2709–2720 (2011). [CrossRef] [PubMed]
21. M. D. Twa, J. Li, S. Vantipalli, M. Singh, S. Aglyamov, S. Emelianov, and K. V. Larin, “Spatial characterization of corneal biomechanical properties with optical coherence elastography after UV cross-linking,” Biomed. Opt. Express 5(5), 1419–1427 (2014). [CrossRef] [PubMed]
22. X. L. Dai and K. Seta, “High-accuracy absolute distance measurement by means of wavelength scanning heterodyne interferometry,” Meas. Sci. Technol. 9(7), 1031–1035 (1998). [CrossRef]
23. P. Sandoz, G. Tribillon, and H. Perrin, “High-resolution profilometry by using phase calculation algorithms for spectroscopic analysis of white-light interferograms,” J. Mod. Opt. 43(4), 701–708 (1996). [CrossRef]
24. J. Schwider and L. Zhou, “Dispersive interferometric profilometer,” Opt. Lett. 19(13), 995–997 (1994). [CrossRef] [PubMed]
25. M. Takeda and H. Yamamoto, “Fourier-transform speckle profilometry: three-dimensional shape measurements of diffuse objects with large height steps and/or spatially isolated surfaces,” Appl. Opt. 33(34), 7829–7837 (1994). [CrossRef] [PubMed]
26. V. Westphal, A. Rollins, S. Radhakrishnan, and J. Izatt, “Correction of geometric and refractive image distortions in optical coherence tomography applying Fermat’s principle,” Opt. Express 10(9), 397–404 (2002). [CrossRef] [PubMed]
27. Y. Li, “Beam deflection and scanning by two-mirror and two-axis systems of different architectures: a unified approach,” Appl. Opt. 47(32), 5976–5985 (2008). [CrossRef] [PubMed]
28. J. Xie, S. H. Huang, Z. C. Duan, Y. S. Shi, and S. F. Wen, “Correction of the image distortion for laser galvanometric scanning system,” Opt. Laser Technol. 37(4), 305–311 (2005). [CrossRef]
29. S. Ortiz, D. Siedlecki, L. Remon, and S. Marcos, “Optical coherence tomography for quantitative surface topography,” Appl. Opt. 48(35), 6708–6715 (2009). [CrossRef] [PubMed]
30. J. M. Bland and D. G. Altman, “Statistical methods for assessing agreement between two methods of clinical measurement,” Lancet 327(8476), 307–310 (1986). [CrossRef] [PubMed]
31. X. Wu, H. Wei, H. Zhang, L. Ren, Y. Li, and J. Zhang, “Absolute distance measurement using frequency-sweeping heterodyne interferometer calibrated by an optical frequency comb,” Appl. Opt. 52(10), 2042–2048 (2013). [CrossRef] [PubMed]
32. W. Zhang, H. Wei, H. Yang, X. Wu, and Y. Li, “Comb-referenced frequency-sweeping interferometry for precisely measuring large stepped structures,” Appl. Opt. 57(5), 1247–1253 (2018). [CrossRef] [PubMed]
33. T. Klein and R. Huber, “High-speed OCT light sources and systems [Invited],” Biomed. Opt. Express 8(2), 828–859 (2017). [CrossRef] [PubMed]
34. S. A. Greenstein, V. P. Shah, K. L. Fry, and P. S. Hersh, “Corneal thickness changes after corneal collagen crosslinking for keratoconus and corneal ectasia: One-year results,” J. Cataract Refract. Surg. 37(4), 691–700 (2011). [CrossRef] [PubMed]
35. J. J. Chen and S. C. Tseng, “Corneal epithelial wound healing in partial limbal deficiency,” Invest. Ophthalmol. Vis. Sci. 31(7), 1301–1314 (1990). [PubMed]
36. N. Nautscher, A. Bauer, M. Steffl, and W. M. Amselgruber, “Comparative morphological evaluation of domestic animal cornea,” Vet. Ophthalmol. 19(4), 297–304 (2016). [CrossRef] [PubMed]
37. M. D. Merindano, J. Costa, M. Canals, J. M. Potau, and D. Ruano, “A comparative study of bowman’s layer in some mammals: Relationships with other constituent corneal structures,” Eur J Anat 6(3), 133–140 (2003).
38. N. Morishige, W. M. Petroll, T. Nishida, M. C. Kenney, and J. V. Jester, “Noninvasive corneal stromal collagen imaging using two-photon-generated second-harmonic signals,” J. Cataract Refract. Surg. 32(11), 1784–1791 (2006). [CrossRef] [PubMed]
39. D. Meller, K. Peters, and K. Meller, “Human cornea and sclera studied by atomic force microscopy,” Cell Tissue Res. 288(1), 111–118 (1997). [CrossRef] [PubMed]
40. M. O. Gordon, J. A. Beiser, J. D. Brandt, D. K. Heuer, E. J. Higginbotham, C. A. Johnson, J. L. Keltner, J. P. Miller, R. K. Parrish 2nd, M. R. Wilson, and M. A. Kass, “The ocular hypertension treatment study: baseline factors that predict the onset of primary open-angle glaucoma,” Arch. Ophthalmol. 120(6), 714–720, discussion 829–830 (2002). [CrossRef] [PubMed]
41. G. Wollensak, M. Wilsch, E. Spoerl, and T. Seiler, “Collagen fiber diameter in the rabbit cornea after collagen crosslinking by riboflavin/uva,” Cornea 23(5), 503–507 (2004). [CrossRef] [PubMed]
42. H. Y. Tan, Y. L. Chang, W. Lo, C. M. Hsueh, W. L. Chen, A. A. Ghazaryan, P. S. Hu, T. H. Young, S. J. Chen, and C. Y. Dong, “Characterizing the morphologic changes in collagen crosslinked-treated corneas by Fourier transform-second harmonic generation imaging,” J. Cataract Refract. Surg. 39(5), 779–788 (2013). [CrossRef] [PubMed]
43. R. M. Werkmeister, A. Alex, S. Kaya, A. Unterhuber, B. Hofer, J. Riedl, M. Bronhagl, M. Vietauer, D. Schmidl, T. Schmoll, G. Garhöfer, W. Drexler, R. A. Leitgeb, M. Groeschl, and L. Schmetterer, “Measurement of tear film thickness using ultrahigh-resolution optical coherence tomography,” Invest. Ophthalmol. Vis. Sci. 54(8), 5578–5583 (2013). [CrossRef] [PubMed]
44. V. V. Tuchin, “Coherent effects in the interaction of laser radiation with tissues and cell flows,” in Tissue Optics: Light Scattering Methods and Instruments for Medical Diagnosis, 2nd ed., V. V. Tuchin, ed. (SPIE Press, Bellingham, WA, 2007), pp. 289–327.
45. H. M. Pedersen, “Theory of speckle dependence on surface roughness,” JOSA 66(11), 1204–1210 (1976). [CrossRef]
46. H. Fujii, T. Asakura, and Y. Shindo, “Measurement of surface-roughness properties by using image speckle contrast,” J. Opt. Soc. Am. 66(11), 1217–1222 (1976). [CrossRef]
47. H. Fujii and T. Asakura, “Roughness measurements of metal-surfaces using laser speckle,” J. Opt. Soc. Am. 67(9), 1171–1176 (1977). [CrossRef]
48. R. S. Lu, G. Y. Tian, D. Gledhill, and S. Ward, “Grinding surface roughness measurement based on the co-occurrence matrix of speckle pattern texture,” Appl. Opt. 45(35), 8839–8847 (2006). [CrossRef] [PubMed]
49. N. Morishige, R. Shin-Gyou-Uchi, H. Azumi, H. Ohta, Y. Morita, N. Yamada, K. Kimura, A. Takahara, and K. H. Sonoda, “Quantitative analysis of collagen lamellae in the normal and keratoconic human cornea by second harmonic generation imaging microscopy,” Invest. Ophthalmol. Vis. Sci. 55(12), 8377–8385 (2014). [CrossRef] [PubMed]
50. S. Sawaguchi, T. Fukuchi, H. Abe, T. Kaiya, J. Sugar, and B. Y. Yue, “Three-dimensional scanning electron microscopic study of keratoconus corneas,” Arch. Ophthalmol. 116(1), 62–68 (1998). [CrossRef] [PubMed]
51. H. Y. Tan, Y. Sun, W. Lo, S. J. Lin, C. H. Hsiao, Y. F. Chen, S. C. Huang, W. C. Lin, S. H. Jee, H. S. Yu, and C. Y. Dong, “Multiphoton fluorescence and second harmonic generation imaging of the structural alterations in keratoconus ex vivo,” Invest. Ophthalmol. Vis. Sci. 47(12), 5251–5259 (2006). [CrossRef] [PubMed]
52. K. M. Meek, S. J. Tuft, Y. Huang, P. S. Gill, S. Hayes, R. H. Newton, and A. J. Bron, “Changes in collagen orientation and distribution in keratoconus corneas,” Invest. Ophthalmol. Vis. Sci. 46(6), 1948–1956 (2005). [CrossRef] [PubMed]