We present a new high speed full-field optical coherence tomography (OCT) instrument, the first full-field OCT system that is capable of in vivo ocular imaging. An isotropic resolution of ~ 1 μm is achieved thanks to the use of a xenon arc lamp source and relatively high numerical aperture microscope objectives in a Linnik-type interferometer. Full-field illumination allows the capture of two-dimensional en face images in parallel, using a fast CMOS camera as detector array. Each en face image is acquired in a 4 ms period, at a maximum repetition rate of 250 Hz. Detection sensitivity per en face image is 71 dB. Higher sensitivity can be achieved by image correlation and averaging, although frame rate is reduced. We present the first preliminary results of in vivo imaging in the anterior segment of the rat eye, which reveal some cellular features in the corneal layers.
© 2005 Optical Society of America
The technique known as optical coherence tomography (OCT) is now widely used in medicine and biology for high resolution imaging . The micrometer resolution, high sensitivity and non invasive nature of OCT make it well adapted to imaging of a wide variety of biological tissues. The technique has enjoyed its first and continuing success in ophthalmology [2–4]. The OCT principle relies on the Michelson interferometer, illuminated with a source of large spectral width. The conventional instrument performs imaging in the longitudinal (xz or yz) orientation by scanning the illumination beam over the sample in the x or y directions, acquiring depth profiles at successive transverse locations, to build up a two-dimensional image. Systems operating in the time domain acquire depth profiles, or a-scans, by scanning the reference mirror in the axial direction [1–6], whilst frequency domain systems hold the reference mirror stationary and instead use parallel detection to detect the spectrum, which via Fourier transformation leads to the same a-scan information as in the time domain but at higher speed and sensitivity [7–13]. Although less common, en face OCT can also be performed to produce images in the xy orientation by fast scanning of the illumination beam in both transverse directions [14,15]. Full-field OCT produces en face oriented images by an alternative method [16–20]. An incoherent white light source is employed to provide full-field illumination with high sectioning ability so that an associated detector array may capture a full two-dimensional image in parallel, eliminating the need for scanning. An advantage of en face techniques is the possibility of using high numerical aperture optics in order to achieve transverse resolution comparable to the axial resolution level, i.e. isotropic resolution may be achieved. In addition, the en face view can supply new information which may complement that provided by the longitudinal view.
We have previously described full-field OCT imaging of ex vivo biological tissues, using a halogen source and a CCD camera as detector array [21–24]. This instrument offered a low-cost alternative to ultrahigh-resolution femtosecond laser-based OCT systems, with comparable spatial resolution (~ 1 μm). Ex vivo animal ocular tissue was examined in detail . Detection sensitivity for a single en face image was 78 dB with an acquisition time of 0.13 seconds. High detection sensitivity (> 90 dB) could be achieved through averaging of accumulated images. Averaging was permitted as only ex vivo and hence immobile samples were imaged. This instrument was not suited to in vivo imaging, since sample movement, inevitable over the relatively lengthy period of acquisition, blurs the interference signal. Recently, we introduced a technique named stroboscopic full-field OCT which dramatically reduced image acquisition time to 10 μs per en face image, allowing in vivo full-field OCT imaging for the first time . This instrument used a flash source of 10 μs pulse duration and a pair of CCD cameras to capture two phase-opposed interferometric images simultaneously. It was successfully applied to cellular-level imaging in weakly scattering media. However, the use of polarized light in this instrument meant that its application to birefringent objects was problematic . As the cornea exhibits a birefringence which varies in magnitude from subject to subject, compensation for the polarizing effects introduced by the cornea would be difficult to realize effectively in a clinical setting. In addition, the use of a camera pair imposes fine alignment of the instrument prior to each imaging session, which again would complicate its clinical use.
We introduce in this article a new full-field OCT instrument with an acquisition speed fast enough to allow in vivo biological imaging, a sufficient detection sensitivity level to penetrate to interesting depths in the eye, and an indifference to polarizing effects of birefringent tissue. A series of images of the anterior segment of the in vivo rat eye illustrate the interest of this instrument for in vivo ultrahigh resolution biological imaging applications.
The experimental set-up is represented in Fig. 1. As for previous full-field OCT instruments, the core interferometer is based on the Linnik microscope . Instead of using a tungsten halogen or flash source, illumination in this new instrument is provided by a continuous 300 W Xenon arc lamp, channelled into the interferometer via a multimode fiber optic cable and reversed microscope objective (air, 10×, 0.25 NA) which acts as entry lens. The fiber serves to homogenize the beam for uniform illumination across the field. An image of the fiber exit face is projected by the entry lens onto the field of the interferometer microscope objectives. The polished surface of a zinc selenide (ZnSe) fragment acts as reference mirror to provide a reflectivity of 9 % in water. An achromatic doublet lens of focal length 75 cm focuses the light exiting from the interferometer onto the sensor of a CMOS camera. Detection in the original full-field OCT instrument was performed by a 15 Hz CCD camera. For the rapid acquisition instrument presented in this paper, we required a camera which was rapid enough to freeze sample movement during the acquisition of each tomographic image, whilst still retaining our relatively low-cost goal. A CMOS camera manufactured by Basler, model A504, 1280 × 1024 pixels, 8 bits, 500 Hz, was selected for this purpose. A piezoelectric translation stage sinu-soidally oscillates the reference mirror at 250 Hz, in synchronization with image capture by the CMOS camera at 500 Hz. Two consecutive images are therefore captured per modulation period of the reference mirror. On calculating the difference squared of these two phase-opposed images, we eliminate the light that has not interfered and retain only that which originates from the coherence volume, forming the signal of interest. We thereby extract a two-dimensional en face tomographic image in an acquisition period of 4 ms. A custom interface written in Visual C++ permits image acquisition, calculation and display in real-time.
Axial resolution is determined in OCT by the coherence length of the illumination source. The product of the Xenon arc source spectrum and the CMOS camera’s spectral response yields λ of 300 nm centered at λ = 650 nm, giving a theoretical axial resolution of 0.5 μm in water. Axial resolution may be degraded due to dispersion effects on travel through tissue . Immersion microscope objectives are therefore used to minimize dispersion mismatch between the two interferometer arms. The axial point spread function (PSF) was measured by moving a mirror through the axial focus in 0.1 μm steps on a motorized micrometric translation stage. The full width at half maximum (FWHM) of the PSF, which corresponds to the experimentally measured axial resolution, was equal to 1.0 μm in water.
The use of relatively high numerical aperture (NA) microscope objectives (0.3 NA) gives a theoretical transverse resolution of 1.1 μm. In practice, the transverse resolution may be degraded slightly when imaging at large depths due to optical aberrations. In order to measure the transverse resolution experimentally, we imaged a 100 nm gold bead embedded just below the surface of an agarose gel. The FWHM of its PSF, which corresponds to the experimental transverse resolution, was 1.5 μm.
Full-field OCT offers the advantage of high axial resolution with a simple white light source, and achieves higher transverse resolution than cross-sectional OCT techniques. Frequency domain OCT instruments typically have axial resolution of a few microns [12,13] and relatively low transverse resolution, whilst high speed scanning en face OCT instruments typically offer both axial and transverse resolutions on the order of 10 μm .
The wavelength range used is suitable for imaging of the cornea as risks of blue light and thermal injury are eliminated due to the filtering of wavelengths below 500 nm and above 800 nm. ANSI laser standards  and ICNIRP guidelines on limits of exposure to broadband incoherent optical radiation  only specify corneal risk at wavelengths below 400 nm and higher than 800 nm. We therefore must measure retinal irradiance as our limiting factor for light levels. The radiant power at the position of the cornea is 2.3 mW, giving a corneal irradiance of 900 mW/cm2 and a retinal irradiance of 27 mW/cm2 for the rat or 6.1 mW/cm2 for humans. At this retinal irradiance level, constant illumination, according to ANSI , is permitted for 20 minutes for the rat and over an hour for humans, when reckoning on the worst case scenario, i.e. with all energy concentrated at the most harmful wavelength of 500 nm. Illumination is permitted for longer periods, according to ICNIRP , when one accounts for the broad spectrum of the source. Were the instrument to be applied to retinal imaging (in which case the illumination spot would be focused on the retina rather than the cornea), the allowed illumination period would be limited to the order of half a second  in the worst case. This is long enough nevertheless to capture over 100 consecutive tomographic images. The power per pixel for our instrument (i.e. corresponding to the irradiance over the full-field) is lower than the power per pixel of scanning OCT systems.
The velocity of sample movement is critical to the success of our imaging technique. During the 4 ms acquisition period of each tomographic image, the sample must remain stationary to within a fraction of the imaging central wavelength in order for an image to be successfully captured at maximum sensitivity. Sample movement in the z direction causes the interference signal contrast to progressively diminish with increasing speed of movement . Image artifacts may also be produced by a fast moving sample, seen as “shadow” doubles of sample structures, resulting from taking the difference of a pair of images in which the sample has changed position from one image to the next. This effect is essentially due to transverse motion. In all imaging examples demonstrated here, neither fringe blurring nor artifacts were witnessed. Sample movement was therefore effectively “frozen” during the acquisition time of each consecutive image pair needed to construct a tomographic image.
Images are viewed on the computer screen in real time so that the operator may orient himself within the sample. A few frames preceding the one being viewed are temporarily guarded in memory so that individual en face frames of particular interest may be transferred to the hard drive. Once the operator chooses to acquire an image stack, the real-time viewing function is paused. The frame rate and data storage capacity of the CMOS camera is such that 1600 consecutive tomographic images may be recorded in a 6.4 second period. Each tomographic image was thus captured in a 4 ms acquisition period, at a repetition rate of 250 frames per second. The object is translated axially at constant speed on a motorized translation stage and images recorded continuously during translation. In order to avoid fringe blurring, the path difference due to the translation of the object during acquisition of each image must be small. If we tolerate a path variation of 100 nm during each image acquisition (4 ms), the acquisition of 1600 images (limited by the memory of our frame grabber) leads to a scan of 150 μm in depth. The three-dimensional (3D) data set acquired in a 6.4 second period hence has (x × y × z) dimensions of (1280×1024×1600) pixels, (300×300×150) μm. The acquisition of larger image stacks would be made possible by enlarging the frame grabber memory capacity. In comparison to the acquisition rate of frequency-domain OCT instruments [12, 13] for a 3D volume of equivalent pixel number (though larger physical size), our instrument is faster by up to a factor of 10.
The high pixel sampling in the z direction (10 images per micron) means that images may be averaged in groups of 10 during post-processing to increase sensitivity, without loss of resolution. Averaging decreases the frame rate to 25 frames per second. In addition to image averaging, we may also choose to bin camera pixels (which consequently reduces the x-y sampling of the data set, and hence the equivalent number of pixels) in another measure to improve sensitivity. The diffraction spot is adequately over-sampled by the detector array, so that pixel binning (of up to 4 × 4 pixels) does not degrade the transverse resolution. The need for averaging and binning with the full-field OCT technique is sample dependent. For example, in the images shown in Fig. 2, averaging in groups of 10 images and 4 × 4 binning were carried out, producing a final 3D stack of (320×256×160) pixels in a 6.4 second acquisition time. In terms of speed, this is comparable or slightly inferior to performance achieved by scanning en face OCT systems  or state-of-the-art confocal scanning laser ophthalmoscopes .
Prior to binning, correction of residual camera noise is required. The CMOS camera used is not shot noise limited and exhibits a noise pattern which is apparent in the form of quasi-periodic randomly located horizontal bands. This was removed by subtracting from each pixel the average intensity of its row sum.
Detection sensitivity for a single en face image was measured to be 71 dB, after 4×4 pixel binning was applied. This sensitivity level proved sufficient to image to penetration depths of a few millimeters in moderately scattering tissues. Averaging of 10 consecutive images improved this figure to 80 dB. In terms of sensitivity, high speed full-field OCT has lower performance than its competitors (98 dB with frequency domain instruments [12, 13] when operating at maximum speed or ~83-85 dB with high speed scanning en face systems ). The reason why our images display an inferior sensitivity level to frequency-domain OCT images is because our area-scan CMOS camera has a lower dynamic range than the line-scan CCDs or photodetectors used in frequency-domain systems. Our CMOS camera saturates at an illumination level that in comparison to scanning systems corresponds to a far lower power per pixel. In our instrument, increasing the illumination optical power would therefore not improve the sensitivity.
Image averaging was performed subject to a correlation condition. The 10 images were displayed and an image of satisfactory appearance selected as reference. Each image in the series was then compared and added to this reference image if correlation between the pair was sufficiently high. Movement between consecutive images in the transverse (xy) direction could be corrected by realignment of the images in post-processing. None of the images shown in this paper were realigned for averaging, as transverse movement was insignificant since the animal was anesthetized. As for z movement, image correlation techniques must be sensitive enough to reject images recorded at depths which differ by more than 1 μm, as their summation would degrade axial resolution. The animal’s respiration caused significant axial sample movement during the acquisition period of the 3D stack, which meant that several scans were required in order to acquire a full depth scan series, with images being ordered into a stack of consecutive steps post-acquisition. This means that the precise depth of each en face image is unknown. Use of an eye tracking device would eventually allow automated reordering.
Image processing post-acquisition involved redistribution of the grey-scale histogram over all 256 levels and application of slight Gaussian smoothing of width 0.5 pixels to reduce speckle noise. Speckle, resulting from interference between structures located within the coherence volume, is present in all OCT images, although is greatly reduced by using a spatially incoherent white light source in comparison to laser sources.
Figure 2 shows a series of images taken in the anterior segment of the rat eye in vivo. Animal experimentation was conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) statement on the use of animals in Ophthalmic and Vision Research and performed following a protocol reviewed and approved by our local animal care committee. The rat was 15 days old, of type WISTAR Rj Han:WI (from Elevage Janvier, 53940, Le Genest-St-Isle, France), and was anesthetized by intramuscular injection with a mixture of ketamine xylazine (Ketamine 100mg/kg (from Virbac, France), xylazine 10 mg/kg (from Bayer, France)).
For the animal’s comfort, it was placed in a foam bed on a heated platform, maintained at body temperature using temperature controllable heating resistances connected to the underside of the metal platform. Hydroethylcellulose (Goniosol© from Alcon Labs), a water-based ophthalmic gel, was used as immersion liquid for index matching as it has a similar refractive index to the cornea. In addition, its gel consistency allowed the contact between eye and objective to be easily maintained. The platform on which the animal rested was displaced in the z-direction using a piezoelectric micrometric translation stage to progressively move downward in depth through the anterior segment of the eye. Displacement of the sample with respect to the interferometer for depth scanning was possible in this experiment due to the small size of the animal. For imaging larger samples, the interferometer may instead be displaced with respect to the sample. Figure 2 shows selected steps of this series: the epithelial cell layer and the basal membrane, the stroma, Descemet’s membrane and the endothelial cell layer, then after passage through the vitreous (in which no signal was detected) the lens capsule and interior lens. Cellular details are suggested in some of the corneal layers. Penetration depth was approximately 1.2 mm, i.e. to the level of the crystalline lens. The annular appearance of certain layers occurs because of the high curvature of the rat eye. If larger (e.g., human) eyes were imaged, their lower curvature would mean that each cellular layer could be visualized individually over this 300 μm × 300 μm field size, so reducing difficulty in image interpretation and allowing optimization of image contrast in each individual layer to improve visibility of morphological features.
As for sample movement, transverse movements were almost entirely suppressed due to the anaesthetic. The animal’s respiration caused noticeable z displacement on the time scale of the stack acquisition (6.4 seconds), and so several depth scans were performed and Fig. 2 formed from a composite stack. Image averaging was however possible, with up to 10 images averaged for each image of Fig. 2 (40 ms for 10 consecutive images). Should the technique be applied to imaging of the human eye in vivo, transverse movements would place limitations on the length of time that the eye remains in a sufficiently stable position to allow image acquisition. In particular, high frequency reflex movements such as tremor (30-100 Hz, 50 - 60 arcsec ) would negate the possibility of the acquisition of image stacks if some stabilization were not provided. A possible solution to this problem would be to synchronize image capture with movement, or to follow the eye with a pupil tracking device during acquisition of an image series .
These preliminary results suggest that for the moment, the quality of images captured using full-field OCT is not as good as that achieved by confocal microscopy for in vivo imaging of the cornea. For example, whilst some familiar details within the cornea are distinguishable (e.g. the keratocytes in the stroma), some cellular details that are visible to confocal microscopy are not visible to full-field OCT (e.g. epithelial cells). In the cornea, confocal microscopy  achieves spatial resolution of 4 μm × 2 μm (axial × transverse) and gives impressive results. OCT however can theoretically achieve higher axial resolution than confocal techniques. As its axial and transverse resolutions rely on independent variables (source coherence length and numerical aperture, respectively), isotropic resolution is possible with full-field OCT. In addition, fullfield OCT should in principle be capable of achieving higher sensitivity than confocal imaging since it uses interferometry to measure the amplitude rather than the intensity of backscattered light.
We have introduced a new high speed full-field OCT instrument and presented its first application to in vivo imaging.
Compared to fast scanning en face OCT techniques, full-field OCT presents the advantage of high axial resolution obtained with a simple white light source. It can be performed at comparable or higher speed though more moderate sensitivity. In comparison to results obtained on ex vivo animal eyes using our original full-field OCT instrument, results are not as good using the rapid instrument due to its lower sensitivity.
We conclude therefore that the instrument requires further technological development to improve its sensitivity level and increase its speed. Using a faster camera with a shot-limited noise level and a larger memory capacity would considerably improve the technique. For imaging of human eyes, some mechanical modification of the instrument is necessary.
It is hoped that in the future, full-field OCT imaging of pathological cornea might complement existing ophthalmological imaging techniques to aid diagnostics by offering real-time information on the state of tissue. Use of this technique in follow-up tests after corneal surgery procedures such as laser ablation could allow detailed examination and longitudinal study of the healing process.
Eventual extension of this technique to retinal imaging could allow cellular level resolution of retinal features. Were this instrument to be used for retinal imaging, its numerical aperture and hence transverse resolution would then be limited by the eye’s optics to a few μm. To improve transverse resolution, adaptive optics could be combined with the OCT instrument, in a similar way to that demonstrated in references [33,34].
K. Grieve acknowledges the support of the European Research Network “Sharp Eye” (HPRN-CT-2002-00301), part of the EU Human Potential program. The authors thank G. Moneron for fruitful discussions, and acknowledge the help of M. Barritault and P. Kuchynka with image processing. This work was supported by the Centre National de Recherche Scientifique.
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 J. G. Fujimoto, “Optical Coherence Tomography.,” Science 254, 1178–1180 (1991). [CrossRef] [PubMed]
2. C. A. Puliafito, M. R. Hee, C. P. Lin, E. Reichel, J. S. Schuman, J. S. Duker, J. A. Izatt, E. A. Swanson, and J. G. Fujimoto, “Imaging of macular diseases with optical coherence tomography,” Ophthalmol. 102, 217–229 (1995).
3. M. R. Hee, J. A. Izatt, E. A. Swanson, D. Huang, J. S. Schuman, C. P. Lin, C. A. Puliafito, and J. G. Fujimoto, “Optical coherence tomography of the human retina,” Archiv. Ophthalmol. 113, 325–332 (1995). [CrossRef]
4. W. Drexler, U. Morgner, R. K. Ghanta, F. X. Kärtner, J. S. Schuman, and J. G. Fujimoto, “Ultrahigh-resolution ophthalmic optical coherence tomography,” Nature Medicine 7, 502–507 (2001). [CrossRef] [PubMed]
5. W. Drexler, U. Morgner, F. X. Kärtner, C. Pitris, S. A. Boppart, X. D. Li, E. P. Ippen, and J. G. Fujimoto, “In vivo ultrahigh-resolution optical coherence tomography,” Opt. Lett. 24, 1221–1223 (1999). [CrossRef]
6. B. Považay, K. Bizheva, A. Unterhuber, B. Hermann, H. Sattmann, A. F. Fercher, W. Drexler, A. Apolonski, W. J. Wadsworth, J. C. Knight, P. S. J. Russell, M. Vetterlein, and E. Scherzer, “Submicrometer axial resolution optical coherence tomography,” Opt. Lett. 20, 1800–1802 (2002). [CrossRef]
7. M. Wojtkowski, T. Bajraszewski, P. Targowski, and A. Kowalczyk, “Real-time in vivo imaging by high-speed spectral optical coherence tomography,” Opt. Lett. 28, 1745–1747 (2003). [CrossRef] [PubMed]
8. R. Leitgeb, C. K. Hitzenberger, and A. F. Fercher, “Performance of fourier domain vs. time domain optical coherence tomography,” Opt. Express 11, 889–894 (2003), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-8-889 [CrossRef] [PubMed]
9. J. F. de Boer, B. Cense, B. H. Park, M. C. Pierce, G. J. Tearney, and B. E. Bouma, “Improved signal-to-noise ratio in spectral-domain compared with time-domain optical coherence tomography,” Opt. Lett. 28, 2067–2069 (2003). [CrossRef] [PubMed]
10. M. A. Choma, M. V. Sarunic, C. H. Yang, and J. A. Izatt, “Sensitivity advantage of swept source and Fourier domain optical coherence tomography,” Opt. Express 11, 2183–2189 (2003), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-18-2183 [CrossRef] [PubMed]
12. M. Wojtkowski, V. J. Srinivasan, T. H. Ko, J. G. Fujimoto, A. Kowalczyk, and J. S. Duker “Ultrahigh-resolution, high-speed, Fourier domain optical coherence tomography and methods for dispersion compensation,” Opt. Express 12, 2404–2422 (2004), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-12-11-2404 [CrossRef] [PubMed]
13. B. Cense, N. A. Nassif, T. C. Chen, M. C. Pierce, S. H. Yun, B. H. Park, B. E. Bouma, G. J. Tearney, and J. F. de Boer, “Ultrahigh-resolution high-speed retinal imaging using spectral-domain optical coherence tomography,” Opt. Express 12, 2435–2447 (2004), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-12-11-2435 [CrossRef] [PubMed]
14. A. G. Podoleanu, G. M. Dobre, and D. A. Jackson, “En face coherence imaging using galvanometer scanner modulation,” Opt. Lett. 23, 147–149 (1998). [CrossRef]
15. C. K. Hitzenberger, P. Trost, P. Lo, and Q. Zhou, “Three-dimensional imaging of the human retina by high-speed optical coherence tomography,” Opt. Express 11, 2753–2761 (2003), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-21-2753 [CrossRef] [PubMed]
16. S. Bourquin, P. Seitz, and R. P. Salath, “Optical coherence topography based on a two-dimensional smart detector array,” Opt. Lett. 25, 102–104 (2000). [CrossRef]
17. E Bordenave, E. Abraham, G. Jonusauskas, N. Tsurumachi, J. Oberl, C. Rullire, P.E. Minot, M. Lassgues, and J.E. Surlve Bazeille, “Wide-field optical coherence tomography: imaging of biological tissues,” App. Opt. 41, 2059–2064 (2002). [CrossRef]
18. A. F. Fercher, C. K. Hitzenberger, M. Sticker, E. Moreno-Barriuso, R. Leitgeb, W. Drexler,, and H. Sattmann, “A thermal light source technique for optical coherence tomography,” Opt. Comm. 185, 57–64 (2000). [CrossRef]
19. L. Vabre, A. Dubois, and A. C. Boccara, “Thermal-light full-field optical coherence tomography,” Opt. Lett. 27, 530–532 (2002). [CrossRef]
22. A. Dubois, G. Moneron, K. Grieve, and A. C. Boccara, “Three-dimensional cellular-level imaging using full-field optical coherence tomography,” Phys. Med. Biol. 49, 1227–1234 (2004). [CrossRef] [PubMed]
23. A. Perea-Gomez, A. Camus, A. Moreau, K. Grieve, G. Moneron, A. Dubois, C. Cibert, and J. Collignon, “Initiation of Gastrulation in the Mouse Embryo Is Preceded by an Apparent Shift in the Orientation of the Anterior-Posterior Axis,” Current Biology 14:3, 197–207 (2004).
24. K. Grieve, M. Paques, A. Dubois, J. Sahel, A. C. Boccara, and J.-F. Le Gargasson, “Ocular tissue imaging using ultrahigh resolution full-field optical coherence tomography,” Invest. Ophthalmol. Visual Sci. 45, 4126–4131 (2004). [CrossRef]
26. K. Grieve, G. Moneron, A. Dubois, and A. C. Boccara, “Ultrahigh resolution ex vivo ocular imaging using ultrashort acquisition time en face optical coherence tomography,” J. Opt. A: Pure Appl. Opt. 7: 368–373 (2005). [CrossRef]
27. C. K. Hitzenberger, A. Baumgartner, W. Drexler, and A. F. Fercher, “Dispersion effects in partial coherence interferometry: Implications for intraocular ranging,” J. Biomed. Opt. 4, 144–151 (1999). [CrossRef]
29. Health Phys.73, 539–554 (1997). [PubMed]
31. S. Martinez, S. Macknik, and D. H. Hubel, “The role of fixational eye movements in visual perception,” Nature Reviews Neuroscience , 5:229, (2004). [CrossRef]
33. B. Hermann, E. J. Fernndez, A. Unterhuber, H. Sattmann, A. F. Fercher, W. Drexler, P. M. Prieto, and P. Artal, “Adaptive-optics ultrahigh-resolution optical coherence tomography,” Opt. Lett. 29, 2142–2144 (2004). [CrossRef] [PubMed]
34. Y. Zhang, J. Rha, R. S. Jonnal, and D. T. Miller, “Adaptive optics parallel spectral domain optical coherence tomography for imaging the living retina,” Opt. Express 13, 4792–4811 (2005), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-13-12-4792 [CrossRef] [PubMed]