Open Access
Add to BibSonomy
Abstract
Whole-eye optical coherence tomography (OCT) imaging is a promising tool in ocular biometry for cataract surgery planning, glaucoma diagnostics and myopia progression studies. However, conventional OCT systems are set up to perform either anterior or posterior eye segment scans and cannot easily switch between the two scan configurations without adding or exchanging optical components to account for the refraction of the eye’s optics. Even in state-of-the-art whole-eye OCT systems, the scan configurations are pre-selected and cannot be dynamically reconfigured. In this work, we present the design, optimization and experimental validation of a reconfigurable and low-cost optical beam scanner based on three electro-tunable lenses, capable of non-mechanically controlling the beam position, angle and focus. We derive the analytical theory behind its control. We demonstrate its use in performing alternate anterior and posterior segment imaging by seamlessly switching between a telecentric focused beam scan to an angular collimated beam scan. We characterize the corresponding beam profiles and record whole-eye OCT images in a model eye and in an ex vivo rabbit eye, observing features comparable to those obtained with conventional anterior and posterior OCT scanners. The proposed beam scanner reduces the complexity and cost of other whole-eye scanners and is well suited for 2-D ocular biometry. Additionally, with the added versatility of seamless scan reconfiguration, its use can be easily expanded to other ophthalmic applications and beyond.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Optical coherence tomography (OCT) is a non-invasive biomedical imaging technique that provides high-resolution cross-sectional images of a sample by detecting low-coherence, near-infrared light, backscattered by the sample’s refractive index inhomogeneities. OCT found its prime clinical application in ophthalmology [1,2], where it became the gold standard for morphological assessment of the posterior eye segment and related structural biomarkers and manifestations of disease. OCT is also important for similar morphological assessments of the anterior eye segment to image pathologies and surgical anatomy [3]. Other clinical applications require structural information from the eye as a whole. A common example is ocular biometry, i.e., the measurement of the physical dimensions of the eye, which is crucial in cataract surgery planning [4,5] to estimate the power of the intraocular lens to be implanted. OCT is the gold standard for such measurements, either employing low-coherence interferometry [6] or, more recently, Fourier-Domain FD-OCT, especially with the advent of long-coherence length source in swept-source SS-OCT [7,8]. When it comes to determining the axial length (the distance between the anterior corneal surface and the retina), the measurement is obtained by a single A-scan, usually along the foveal fixation axis, precluding lateral imaging of the retina [9].
However, there are several clinical and vision science applications where it is required to have both anterior and posterior OCT images and whole-eye biometry in 2-D for both segments. Some examples are narrow-angle glaucoma and myopia progression studies [9,10]. In narrow-angle glaucoma, an anterior segment OCT scan provides information about the irido-corneal angle [11] but only a posterior segment OCT scan can show the retinal nerve fiber layer thickness. Similarly, for a patient with high myopia, a biometric measurement of the axial length is important to evaluate the associated risk of myopia progression; however, peripheral eye length measurements coupled with lens biometry data could help evaluate peripheral defocus, and understand its role as driver of further myopia. Also, clinically, a retinal OCT scan can inform on potential myopic maculopathy and pathologic myopia [12]. Lastly, even in ocular biometry for cataract surgery planning, imaging of the posterior segment is important to provide accurate and repeatable measurements, and this includes ruling out patient’s motion during acquisition. A small-angle OCT scan around the retinal foveal pit is used to evaluate motion blur, and therefore, ensure the patient was still during the measurement of axial length [13–15].
The reason why whole-eye OCT scanners are not common is that different focus and scanning configurations are needed to image each ocular segment, as the anterior segment is effectively a compound lens that refracts and therefore alters the scanning configuration used for imaging the anterior segment, thus preventing its use for posterior segment imaging. While for imaging the anterior segment of an eye a typical scanning configuration involves displacing a beam, focused on the anterior segment, parallel to the OCT sample lens axis; for imaging the posterior segment of an eye a typical scanning configuration involves pivoting a collimated beam at the nodal point of the eye (ideally) such that the light beam is focused on the retina by the cornea and the crystalline lens without affecting the scanning direction [16]. Thus, conventional OCT systems are set up to perform either anterior or posterior segment scans and cannot easily switch between the two modes unless some optical components of the system are changed or added to account for the refraction of the eye’s optics, leading to the common separation between anterior and posterior segment OCT systems.
Another important consequence of this separation is cost. In fact, in most cases, the ophthalmic or optometry practice will have to separately procure an anterior segment OCT system, a posterior segment OCT system and an OCT biometer. This is particularly taxing for practices in low-resource settings and emerging economies, where, for example, the prevalence of cataract is higher than in developed countries [17], and the cost of the newer swept-source OCT biometers alone can be in excess of 40 kUSD [18].
A wide variety of approaches have been proposed to overcome this physical limitation by imaging each ocular segment either sequentially or simultaneously within the same acquisition of a subject’s eye. For sequential imaging of each ocular segment the use of mechanical actuators (e.g., a bi-stable rotary solenoid [19], or a fold mirror [20]) has been proposed to switch between two static sample arm configurations suitable for each segment. Alternatively, the split between two static sample arm configurations has been achieved optically. Kim et al. proposed a polarization-encoding approach to separate the anterior and posterior segment scan paths, and an optical switch has been used to interlace the acquisition of each interferometric A-scan signal at the detector, a necessary step due to the short axial range of spectrometer-based OCT systems [21]. For simultaneous imaging of both segments of the eye, Fan et al. [22] used dichroic mirrors and two wavelength bands for the anterior and posterior segment, respectively. However, this approach requires two sources and two detectors. This requirement was relaxed by Kim, et al., when using the previously-proposed polarization-encoding approach in combination with two synchronized photodetectors in a swept-source OCT system [23]. Owing to the longer coherence length of MEMS-swept VCSEL sources and the coherence revival effect, McNabb, et al. [24] demonstrated a polarization- and pathlength-encoding approach that required only one source and one detector and that scanned a wide field of view for anterior and posterior segment, respectively.
The system of McNabb, et al. [24] included a lens mounted on a motorized translation stage to provide dynamic posterior segment optical beam divergence control and therefore diopter correction for patient defocus without adjusting any optics used for imaging the anterior segment. Beam divergence control for diopter correction and more generally for focus tuning has also been used in an example of dynamic whole-eye scanner by Grulkowski, et al. [25], but via an electro-tunable lens (ETL), instead. Taking advantage of its high focal-tuning speed, the ETL has been set to sequentially attain a focused beam in the anterior segment and in the retina, respectively. However, in this case, only a single static scanning configuration was present, which was a trade-off between the aforementioned typical scanning configurations for either segment. By pivoting the beam axis in front of the cornea, the OCT signal-to-noise ratio (SNR) at the cornea is reduced, and the retinal scan angle is likely limited by the aperture of the iris.
Therefore, having the same flexibility afforded by focus tuning in switching between two or more scanning configurations, without the need for duplicating or requiring expensive components, would result in improved image quality at a reduced cost. Moreover, a dynamically reconfigurable scanning configuration would also allow customization for specific applications even beyond whole-eye imaging. In order to incorporate an additional degree of versatility in the selection of the scanning configuration, i.e., the control of beam axis displacement and deflection, several non-mechanical components for beam steering could be used [26]. ETLs are the most promising for lower-cost applications, and they have already been implemented in devices such as an adjustable beam expander [27], where two ETLs were arranged on axis in a telescope configuration, and a wide-angle beam steering system [26,28], where two or more ETLs were arranged at an offset from their respective axes.
In this work, we present the working principle and experimental characterization of a novel, low-cost, optical beam scanner based on the combination of three ETLs for performing both dynamic focusing and changes in the scanning configuration [29]. To the best of our knowledge, our proposed optical beam scanner is the first optical beam scanner that, without requiring any mechanical displacement of the optical components, allows to independently select the divergence and direction of the output beam. Thus, it allows to image a 2-D cross section of both ocular segments in the desired scanning configurations, by, iteratively and rapidly switching between the anterior and posterior imaging configurations. We designed and assembled a prototype of our reconfigurable beam scanner, and we measured beam profiles for each scan configuration to characterize its performance. Finally, we connected it to the sample arm of a SS-OCT system and recorded whole-eye OCT images in a human phantom eye and in an ex vivo rabbit eye.
2. Instrumentation
2.1 Working principle and system design
The low-cost, non-mechanical, reconfigurable, optical beam scanner we propose can dynamically control the output beam divergence, and reconfigure the scanning configuration, by providing independent control of the beam axis displacement and deflection. The main optical components of the beam scanner are a fiber collimator, three ETLs, a hollow-roof mirror (HRM), a flat mirror (M), and a periscope system (P).
To better explain the working principle, we will analyze various sets of components before combining them. For example, only two ETLs are needed to control the displacement and deflection of the output light beam. The two ETLs ($ET{L_1}$ and $ET{L_2}$) are placed on axis at a fixed distance d between them. For the system to work, the input light beam from a collimator must be incident at an offset ${h_{in}}$ from the optical axis of the two ETLs (Fig. 1(a)). We obtained the equations for the desired transversal displacement of the beam at the principal plane of $ET{L_2}$, ${h_{out}}$, and the output deflection angle, ${\theta _{out}}$, with respect to the optical axis of the ETLs, from the ABCD formalism under the paraxial approximation (Eqs. (1),(2)) [30].
Fig. 1. Optical schematic of the proposed beam scanner with step-by-step illustration of its working principle: (a) different combinations of the focal length of $ET{L_1}$ and $ET{L_2}$ allow the control of the output beam displacement and deflection angle, while (b) the addition of $ET{L_3}$ on axis with the input beam allows the control of the output beam divergence. An example of (c) telecentric displacement of the output beam in a single pass setup of the beam scanner, and (d) in a double pass setup of the beam scanner, where the variation of the beam magnification along the transverse direction is compensated for. (e) An example of angular displacement of the output beam in a double pass configuration.
From these formulae, we can derive the combinations of ${f_1}$ and ${f_2}$ required to perform either a telecentric scan, as typically performed for anterior segment scans; or an angular scan, as typically performed for posterior segment scans. For simplicity, we will only consider the case where ${\theta _{in}} = 0$.
For a telecentric scan, the output angle ${\theta _{out}}$ must be always zero, and the transverse displacement ${h_{out}}$ must vary linearly with time. This leads to the condition that the sum of the focal length ${f_1}$ and ${f_2}$ is equal to the distance d between the two ETLs, i.e., ${f_1} + {f_2} = d$. $ET{L_1}$ and $ET{L_2}$ form a telescope and the output transverse displacement ${h_{out}}$ varies as a function of the telescope magnification (${f_2}$/${f_1}$), i.e., ${h_{out}} ={-} \frac{{{f_2}}}{{{f_1}}}{h_{in}}$. It is useful to rearrange Eqs. (1),(2) and solve for the variables that we directly control, i.e., the focal lengths ${f_1}$ and ${f_2}$, and explicitly introduce the dependency with time (Eqs. (3)–(5)).
For an angular scan, the desired distance ${d_p}{\; }$ of the output beam pivoting point from the principal plane of $ET{L_2}$ must be constant and the output angle ${\theta _{out}}$ must vary linearly with time. This implies that the output transverse displacement ${h_{out}}({{t_i}} )$ must vary accordingly, as dictated by trigonometry. It is also convenient to choose the same conditions as in the telecentric scan for the initial output transverse displacement ${h_{out}}({{t_0}} )$ and for the initial output deflection angle ${\theta _{out}}({{t_0}} )$, e.g., ${h_{out}}({{t_0}} )={-} {h_{in}}$ and ${\theta _{out}}({{t_0}} )= 0$ (Eqs. (6)–(8)).
The next feature of the proposed beam scanner is the control of the output beam divergence for focus tuning. In fact, an anterior segment scan requires to focus the output beam into the anterior segment of the eye at a desired distance ${d_\textrm{f}}$ from the principal plane of $ET{L_2}$. Conversely, a posterior segment scan requires to make the output beam incident on the eye pupil collimated. Independent control of the beam divergence from control of the beam direction can only be attained using an additional ETL. $ET{L_3}$ must be placed on axis with the input light collimator before $ET{L_1}$, at a distance ${d_3}$ from it (Fig. 1(b)).
Figure 1(c) shows three transverse steps of an anterior segment scan, where the focal lengths of the three tunable lenses are selected to generate a telecentric displacement $\Delta {h_{out}}$ of the beam while the focal distance ${d_\textrm{f}}$ is maintained at the cornea. In this schematic, we can observe a variation in the beam numerical aperture, as the scan progresses. This variable magnification generates an undesirable variation of the transverse resolution during the scan [31].
This drawback is reduced by converting the system from single pass through the $ET{L_1} - ET{L_2}$ sub-system into a double-pass configuration. Figure 1(d),(e) show the full system design. Firstly, light from a collimator is incident on axis on $ET{L_3}$ and propagates over a distance ${d_3}$ until $ET{L_2}$. The propagation direction along this path is redirected orthogonally to the initial direction as the beam reflects off the mirror M to then propagate on axis through $ET{L_2}$ and $ET{L_1}$. Secondly, the hollow-roof mirror HRM, at a distance ${d_{HRM}}$ from $ET{L_1}$, induces the offset ${h_{in}}$ as it redirects the beam back towards $ET{L_1}$ and $ET{L_2}$ parallel to their optical axis. Lastly, a periscope P system compensates for the previously induced offset ${h_{in}}$. This double-pass propagation compensates, in part, the undesirable magnification, as it can be seen in Fig. 1(d) compared to Fig. 1(c). However, a residual beam numerical aperture variation during a scan is still present in the current design both for the anterior and the posterior segment scans (Fig. 1(d),(e)), respectively).
Using the thin lens equation of the geometrical optics formalism under the paraxial approximation [30], the equation of the focal length ${f_3}$ of this third lens $ET{L_3}$, Eqs. (9a)–(9e), was obtained as a function of the desired working distance ${d_f}$ from $ET{L_2}$ and the fixed distances of the system (${d_3}$, d and ${d_{HRM}}$) as well as the previously obtained focal lengths ${f_1}$ and ${f_2}$ for a given displacement and deflection of the beam.
In summary, we can set combinations of the focal lengths of the three ETLs to perform both a telecentric scan while focusing the output light beam at a constant working distance and an angular scan while maintaining a collimated output light beam pivoting at a specific distance. This shall prove particularly useful for whole-eye imaging.
2.2 System components selection and parameters optimization
We chose low-cost, off-the-shelf components for an experimental prototype of the optical beam scanner (see Table 1).
An analysis of the system parameters, such as the fixed mechanical distances between system components, was performed to optimize the design of optical beam scanner. This analysis was performed using Eqs. (1)–(9a), with the goal of maximizing the scanning range of the optical beam scanner in both scanning configurations, i.e., $\mathrm{\Delta }{h_{out}}$ and $\mathrm{\Delta }{\theta _{out}}$ for the telecentric and angular scans, respectively. During this analysis, two parameters of the beam scanner in the double pass setup were considered: the input offset ${h_{in}}$ and the distance d between $ET{L_1}$ and $ET{L_2}$. The detailed results of this analysis are presented in Supplement 1. In summary, the largest output transverse displacement is obtained for ${h_{in}}$ equal to 2.5 mm, as shown in Fig. S1, which is half of the semi-aperture of $ET{L_1}$. In both scanning configurations, the scan ranges $\mathrm{\Delta }{h_{out}}$ and $\mathrm{\Delta }{\theta _{out}}$ increase with the distance d until d equals 150 mm, as shown in Fig. S2. However, when d becomes bigger than 150 mm the limiting factor is no longer the shortest available focal length of $ET{L_1}$ but the clear aperture of $ET{L_2}$ that starts vignetting the beam.
Table 1. List of the selected off-the-shelf components for the optical beam scanner
2.3 Expected system specifications
The optimal design of the beam scanner was based on the optical components listed in Table 1 and the analysis of the system parameter presented in Supplement 1. The input offset ${h_{in}}$ and the distance d between $ET{L_1}$ and $ET{L_2}$ were set to be 2.5 mm and 150 mm, respectively. We chose a comfortable working distance ${d_f}$ equal to 105 mm, similar to other ophthalmic scanners [32]. A summary of all set values for the mechanical distances is presented in Table 2.
Table 2. List of the optimized parameters for the optical design of the beam scanner when considering the commercial components in Table 1
Using the values in Table 2, we evaluated the focal lengths of $ET{L_1}$-$ET{L_3}$, according to Eqs. (3)–(9a), required for imaging the anterior and posterior segment, respectively. We also validated the required focal lengths with a simulation of the designed beam scanner in Zemax (OpticStudio).
Figure 2 shows the combination of the focal length of $ET{L_1}$, $ET{L_2}$ and $ET{L_3}$ as a function of the desired displacement of the beam for both scanning configurations, telecentric and angular, and the desired divergence at the working distance ${d_f}$ and ${d_p}$, respectively. Under these conditions, and considering the limiting factors, including the clear aperture of the ETLs, the beam scanner was expected to have a telecentric scan range of 2.5 mm and an angular scan range of 1.2 deg. Zemax models for six specific transverse and angular displacements were simulated, and the corresponding focal lengths of $ET{L_1}$-$ET{L_3}$ are represented with red dots in Fig. 2(a),(b). In both cases the Zemax simulations were in good agreement with the calculations from the analytical equations.
Fig. 2. Calculated (solid lines) and simulated (red dots) focal lengths of $ET{L_1}$-$ET{L_3}$ as a function of the (a) transverse displacement in the anterior segment imaging configuration, and (b) angular displacement in the posterior segment imaging configuration, using the values of Table 2 and Eqs. (3)–(9a) and Zemax models, respectively.
We also analyzed the expected (residual) variation of transverse resolution along the transverse or angular position. To do this, we analytically calculated the beam diameter along the beam axis (z) as the collimated Gaussian input beam is transmitted through the beam scanner components, using Eqs. (3.2-5)–(3.2-9a) of [30]. The variation of the output beam spot size, i.e., the beam diameter at the waist position, corresponding to the working distance ${d_f}$ and ${d_p}$, is shown in Fig. 3(a) and Fig. 3(c) as a function of the transverse position for the anterior scan configuration and as a function of the angular position for the posterior scan configuration, respectively. Figure 3(b) shows the output beam diameter profile along the beam axis in the range of the working distance ${d_f} \pm 10$ mm for different transverse positions for the anterior scan configuration. Figure 3(c) shows the residual variation of the output beam spot size from 1.28 mm (at 0 deg) to 0.96 mm (at 1.2 deg) for the angular scan configuration at the pivoting point, ideally set on the eye pupil. To work out the beam spot size at the retina for the posterior scan, we used Eqs. (3.2-17) of [30] and we assumed a relaxed eye effective focal length of 22.2 mm [33].
Fig. 3. Theoretical transverse resolution of the beam scanner for the telecentric scan (a-b) and the angular scan (c). (a) The variation of the spot size at the working distance ${d_f}$ for the telecentric scan. (b) The axial beam diameter profile for the telecentric scan at various transverse output positions. (c) The variation of the spot size for the angular scan at the pupil plane.
Lastly, the spot size on the focal plane in the anterior segment and on the retina corresponds to the transverse resolution for either imaging configuration. As a result, the transverse resolution varies from 35 ${\mathrm{\mu} \mathrm{m}}$ (at 0 mm) to 102 ${\mathrm{\mu} \mathrm{m}}$ (at 2.5 mm) for the anterior imaging configuration and from 23 ${\mathrm{\mu} \mathrm{m}}$ (at 0 deg) to 31 ${\mathrm{\mu} \mathrm{m}}$ (at 1.2 deg) for the posterior imaging configuration. A summary of the theoretical imaging specs is presented in Table 3.
2.4 Experimental beam scanner setup
A prototype of the optical beam scanner was experimentally implemented in the laboratory with the commercial components from Table 1 set according to the mechanical distances and values defined on Table 2. The control of the three ETLs was performed with a low-cost Arduino Nano board in combination with a custom Matlab GUI code that allowed to independently select the scan mode (telecentric or angular) and the working or pivoting distance of the output beam, ${d_f}$ or ${d_p}$. The focal length or optical power of the ETLs was set by the duty cycle of a pulse width modulation (PWM) signal (Visualization 1) sent by the Arduino Nano board with 8-bit resolution to the lenses through a lens driver (DRV8833, Texas Instruments). Due to the high frequency of the PWM signal, the lenses are effectively subject to a voltage between 0-5 V proportional to the PWM signal. The focal length variation with voltage of each ETLs was characterized using a high speed focimeter [34] and adjusted when aligned in the beam scanner experimental setup. The resulting calibration curves are presented in Fig. S3 of Supplement 1. The total cost of the three ELTs and their control electronics was ∼900 USD compared to ∼2000 USD for a standard 1-axis galvanometer mirror scanner with control electronics, resulting in over 50% cost reduction for the beam scanning engine.
The optical performance of the experimental beam scanner was characterized by using a CCD camera (DCC1545M, Thorlabs) to profile the output beam around the working or pivoting distance. This CCD was axially displaced over a range of 25 mm to profile the output beam at different axial position. The analysis of the size and position of the beam in the CCD images allowed to characterize the two main features of the developed optical beam scanner: its capability of controlling the divergence and the direction of the output beam. This analysis was performed for the two scan configurations of interest for ocular imaging.
2.5 Whole-eye OCT experimental setup
The experimental prototype of the new optical beam scanner was integrated into the sample arm of a custom developed SS-OCT system to acquire whole-eye cross-sectional images of different ocular samples. The custom SS-OCT system was based on a fiber Mach-Zender interferometer configuration, similar to [32], as schematically shown in Fig. 4. The light source was a MEMS-based vertical-cavity surface-emitting laser (VCSEL) swept-source centered at 1060 nm with a sweep rate of 60 kHz over a spectral 10 dB-bandwidth of 100 nm (SL100060, Thorlabs), producing a theoretical axial resolution of 11 ${\mathrm{\mu} \mathrm{m}}$. The interferometric signal was acquired with a dual balance photodetector with a 3 dB-bandwidth from 3 kHz to 1 GHz (PDB481C-AC, Thorlabs) and digitized by a 12-bit 8-lane PCI Express digitizer (ATS 9360, Alazartech). The experimental axial depth range was ∼33 mm, when sampling the interferogram at 1 GS/s.
Fig. 4. Schematic of the custom SS-OCT system coupled with the proposed optical beam scanner, where SS: swept laser source, Ci: circulators, FC: fiber coupler, PC: computer, DBP: dual balanced photodetector. Sequential acquisition of the anterior and posterior segment scans is recorded in alternated B-scans.
We compared the cross-sectional images acquired with the proposed optical beam scanner with those acquired with a standard telecentric scan system, a typical approach in anterior segment scans. Such scan system was based on galvanometric scanning mirrors (Saturn 1B, ScannerMAX, Pangolin, USA) placed at the back focal distance of a 2” aperture f-theta telecentric scan lens (LSM05, Thorlabs, USA), which allowed to perform a telecentric scan over a transverse field of view of 8 mm. The custom SS-OCT system could be connected interchangeably with either sample arm and the paired pathlength-matched reference arm by flipping a mirror. A custom LabView program was used for synchronizing the OCT acquisition with the control of either scanner. With the proposed optical beam scanner, the two scan configurations were sequentially alternated without any mechanical adjustment to obtain a whole-eye B-scan image. Table 4 reports the image acquisition specifications for either system.
Table 4. Image acquisition specifications for the proposed whole-eye beam scanner and a standard galvo-based anterior segment scanner used for comparison
To visualize the beam profiles we made use of a nano-particle-embedded point spread function (PSF) phantom (Tomlins Analytics, UK) [35]. To compare the images with either scanner we used an OCT model eye (Modell-Augen Manufaktur, Germany), as it mimics realistic components of the human eye in both the anterior segment (the cornea and the crystalline lens) and the posterior segment (the main retinal layers as well as the macula and the optic nerve). We also acquired images of an ex vivo rabbit eye, obtained from a local slaughterhouse, to demonstrate whole-eye scan of an animal eye.
3. Results
Figure 5 shows an overlay of a picture of the experimental prototype with ray tracing of the beam at the starting position of the anterior scan of the model eye. Visualization 2 shows the same view as in Fig. 5, with a sequence of four different transverse displacements for the anterior scan and, subsequently, four different angular displacements for the posterior scan.
Fig. 5. Overlay of a picture of the experimental beam scanner prototype with the ray tracing of the beam through the beam scanner for the starting position of the anterior segment scan configuration. Visualization 2 shows the sequence of eight discrete positions for the anterior and posterior scan configurations.
Figures 6,7 show the results of the experimental characterization of beam size and direction via beam profiling at several axial positions, as described in Sec. 2.4, for four displacements in each of the anterior and posterior segment scan configurations.
Figure 6 shows a perspective view of the beam profile images at the three axial positions, namely 12.5 mm before, at, and 12.5 mm after the working distance, ${d_f}$ = 105 mm from $ET{L_2}$ and also equal to the pivoting distance ${d_p}$. Figure 6(a) refers to the anterior segment scan configuration. We can observe the beam divergence and the focus at the working distance by noticing the variation in the spot size between the central and the two other axial positions. It is also evident that the beam axis is maintained parallel to itself and to the optical axis of the system throughout the scan. Figure 6(b) depicts the posterior segment scan configuration, only between two angular positions. In this case, we can observe that the beam diameter is much larger than for the anterior scan configuration and rather constant in size for the three axial positions, denoting a collimated output beam pivoting about the central position.
Fig. 6. A perspective view of the beam profile images at the three axial positions, namely 12.5 mm before, at, and 12.5 mm after the working or pivoting distance, ${d_f}$ or ${d_p}$. (a) Anterior segment scan overlay of four transverse displacement profiles. (b) Posterior segment scan overlay of two angular displacement profiles. The arrows represent the scan direction with time. The scale bars represent 1 mm.
The dynamic performance of the beam scanner can also be seen in Visualization 3, which displays a video of 120 beam profile frames, corresponding to each A-scan position for both scan configurations, recorded at the working distance position, ${d_f},$ (set equal to the pivoting distance ${d_p}$). Visualization 4 shows a macro-style video of two consecutive anterior and posterior scans and corresponding close-ups on the PSF phantom, where the scattering by the dispersed iron oxide nanoparticles creates a visible trace of the profile with depth inside the phantom on a video camera sensitive to infrared light.
The quantitative analysis of the beam diameter and direction from the beam profiles recorded at seven and five different axial positions for the anterior and posterior segment scan, respectively, is shown in Fig. 7. The origin of the axial distance axis is set at the working distance, ${d_f}$. Figure 7(a) shows a graph of the experimental axial profile of the beam diameter (evaluated for the 1/e2 value of the peak intensity) for four transverse displacements in the anterior segment scan. The experimental data was also fitted to the theoretical axial profile of a Gaussian beam with the minimum root-mean-square error from the experimental data points. The focal planes for the four transverse displacements all lie within a range of 2.5 mm, confirming our ability to set a constant working distance throughout the scan. The beam diameter increased and, conversely, the transverse resolution decreased during the transverse scan, from 58 ${\mathrm{\mu} \mathrm{m}}$ to 122 ${\mathrm{\mu} \mathrm{m}},$ due to the variation in the focal length of $ET{L_3}$ required to maintain the focal plane at the set working distance. The spot size at each transversal position was in relatively good agreement with theory, as the measured spot size deviated less than 30 ${\mathrm{\mu} \mathrm{m}}$ from the analytical results of Fig. 3. The transverse resolution degradation by a factor of 2.1 was less severe than the 2.6 factor expected from the analytical results.
A side effect of the reduction of transverse resolution is the increase in depth of field, equal to twice the Rayleigh range of the associated beam. Figure 7(b) shows the beam axis for the four beams at the increasing transverse displacements considered in Fig. 7(a). This demonstrates the ability of the beam scanner to perform a telecentric scan as the deviation of the beam axis from a propagation direction parallel to the optical axis of the system was less than ${\pm} 0.1\textrm{\; deg}$ for every beam transverse displacement.
Figure 7(c) shows a graph of the experimental axial profile of the beam diameter for four angular displacements in the posterior segment scan. In this case, the experimental data was fitted to a line. The divergence of the beam was less than 0.3 deg for all the angular scan positions, demonstrating the ability to produce a highly collimated output beam throughout the angular scan. We also observed the expected reduction of the beam diameter during the scan, as predicted by the analytical analysis of Sec. 2.3. The experimental spot sizes were smaller, by 0.46 mm on average, than the ones expected from the analytical results in Fig. 3(c). Figure 7(d) shows the beam axis for the four beams at the increasing angular displacements considered in Fig. 7(c). This demonstrates the ability of the beam scanner to perform an angular scan pivoting at a constant position. The pivoting point was measured at a distance of 103 mm from $ET{L_2}$, only 2 mm off the expected position, and all measured beam axes crossed each other within a range of 1.5 mm.
Fig. 7. Experimental axial profiles of (a), (c) the beam diameter and (b), (d) the beam axis direction for four displacements in (a), (b) the anterior segment telecentric scan and (c), (d) the posterior segment angular scan, respectively.
Figure 8 shows an overlay of the OCT B-scans of the model eye obtained with the proposed beam scanner and the standard telecentric galvanometric scanner described in Sec. 2.5. The scans obtained with the experimental prototype of the optical beam scanner were overlapped with the larger B-scan obtained with the galvanometric scanner.
Fig. 8. Overlay of the OCT B-scans of the model eye obtained with the proposed beam scanner and the standard telecentric galvanometric scanner described in Sec. 2.5 for comparison. (a) B-scan of the entire eye. The boxed areas represent the whole-eye scans obtained with the proposed beam scanner, in green and yellow for the anterior and posterior segment scan configurations, respectively. (b) Close-up on the anterior segment. (c) Close up on the posterior segment, where only the image in the yellow box is artifact-free. (d) Co-location of image features in the posterior segment scan with model retina landmarks. The scanned area is at the edge of the model optic nerve.
Figure 8(a) shows the comparison for the entire eye. The anterior segment of the model eye, including the cornea, iris and crystalline lens, can be observed in the partially transparent background image acquired with the standard scanner. In the green inset, we can see the corresponding anterior segment scan performed with the proposed beam scanner. A close-up on the anterior segment scans in Fig. 8(b) permits to verify that all anatomical features and biometric dimensions clearly match. However, it is in the posterior segment that the value of the proposed scanner becomes evident when seamlessly reconfigured for posterior segment scanning. Without the ability to switch to an angular scan, the standard telecentric scan rays get refracted towards a small area, and the beam itself diverges, creating a defocused spot on the retina. This leads to the artifactual image of a low-resolution, laterally unresolved, retinal depth profile, as seen in Fig. 8(a) or in the posterior segment close-up in Fig. 8(c), where the same issue affects the image of the retina in the green box, obtained with our scanner set for anterior segment scanning.
However, when the scanning configuration is switched to posterior segment scanning (yellow box), different features of the model retina were clearly observed. To verify and co-locate those features with a larger field of view on the retina, we also scanned the posterior segment with the standard telecentric scanner. This was possible, as the model eye is composed of two halves, comprising the anterior and posterior segments, respectively, that can be unscrewed and separated. Figure 8(d) illustrates that the area scanned by the proposed beam scanner perfectly matches the edge of the model optic nerve.
Fig. 9. Whole-eye OCT B-scan of an ex vivo rabbit eye acquired with the proposed beam scanner only. (a) B-scan acquired with the anterior segment scan configuration. (b) subsequent B-scan acquired with the posterior segment scan configuration; (c) close-up view of the anterior segment from the boxed area in (a). (d) close-up view of the posterior segment from the boxed area in (b), where the aspect ratio and scale are maintained as in (c) for reference.
We also imaged an ex vivo rabbit eye with the proposed beam scanner. Figure 9(a) shows the B-scan recorded with the anterior scanning configuration. Figure 9(b) shows the subsequent B-scan recorded with the posterior scanning configuration. Figure 9(c) is a close-up view of the anterior segment in Fig. 9(a), where the cornea, iris and lens of the rabbit eye are clearly seen. Similarly, Fig. 9(d) shows a close-up view of the posterior segment of Fig. 9(b). As the posterior segment scan configuration pivots in the eye pupil, details of the inner and outer retina are now visible, unlike in Fig. 9(a) where the posterior segment image was hindered by the shadow cast by the iris and otherwise a repeated artifactual image of a retinal depth profile was caused by the eye refraction.
4. Discussion
We presented an optical beam scanner with the ability to reconfigure the scanning and focusing configuration via low-cost non-mechanical components, and we employed it to provide quasi-simultaneous whole-eye imaging, albeit over a limited field-of-view. The device employs three ETLs, and several mirrors. We characterized its dynamic scanning and focusing capabilities in an experimental proof-of-concept setup. We benchmarked whole-eye OCT images of a model eye against those produced with a standard galvanometric scanner, and then acquired whole-eye OCT images of an ex vivo rabbit eye, revealing details comparable to those seen in conventional anterior and posterior OCT scanners.
We derived the analytical equations describing the relationship between the focal length of the three ETLs and the desired output beam position, direction and divergence and optimized the mechanical distances between off-the-shelf components to maximize the transverse and angular scanning range of the proposed beam scanner. The optical characterization of the experimental prototype showed a good agreement between theoretical and measured beam profiles. However, the measured beam spot sizes in the anterior segment scan configuration were slightly larger than expected, especially for the smallest spot size, hinting to the presence of aberrations preventing a diffraction limited spot size. The predominant aberration appears to be astigmatism, followed by spherical aberration and coma. This is not entirely unexpected, as the working principle requires the beam to be incident off axis from the ELT axis, and the scan progresses toward the periphery of $ET{L_2}$, where the impact of aberrations is higher. While these aberrations may affect the resolution, we verified that they did not affect the scanning accuracy and range. Nevertheless, larger ETL clear apertures would be beneficial in reducing the effect of system aberrations.
The variation of transverse resolution along the scan direction is undesirable. Even though the double pass configuration through $ET{L_1}$ and $ET{L_2}$ provides a smart way to partly compensate this effect, as the magnification in a single pass is undone in the double pass, that is only valid for a collimated beam impinging on the $ET{L_1}$-$ET{L_2}$ telescope. Therefore, for the scanning configurations considered here, there is some residual beam size variation, and hence, transverse resolution variation. One solution to prevent this variation would be the addition of a variable beam expander (i.e., a telescope made of two additional ETLs) to place between the collimator and $ET{L_3}$ in order to fully pre-compensate this residual variation. However, this solution adds to the device cost and footprint.
The minor experimental imprecisions found in the position of the focal spots or beam axes intersection with respect to the working and pivoting distance could be removed by employing a more precise control of the ETL focal lengths by using a microcontroller with better digital resolution than the Arduino (e.g., the ESP32, 16-bit resolution) and implementing a temperature control of the electrotuneable lenses [36].
A limitation of the current experimental implementation of the beam scanner is the switching frequency between adjacent scan positions. Although the experimental data reported in this work was acquired at 500 Hz, the selected ETLs could respond at faster rates, up to 1 kHz and beyond for small diopter increments [37]. Future works will explore their use at greater speeds, achieving motion artifact-free in vivo images. This will be important to ensure that biometric readings are accurate, both within a single scan and across anterior and posterior scans.
Another important limitation of the current experimental implementation is the limited transverse and angular scan range. The characterization of each scan configuration led to a field of view of 2.5 mm and 1.2 deg for the telecentric (anterior segment) and angular (posterior segment) scan configurations, respectively. Again, larger ETL clear apertures and wider focal ranges, especially towards smaller focal lengths, would enable larger transverse and angular scan ranges. For example, there are off-the-shelf ETLs with a clear aperture 6 mm larger than the one we used for $ET{L_2}$ in our prototype, and which could achieve a similar focal range to $ET{L_2}$, with the help of an offset lens, leading to more than twice the scan ranges we presented.
Another option to increase the field of view of the beam scanner would be the addition of a magnifying two-lens telescope at the output of the beam scanner, such that the first telescope lens would be placed a focal length away from the working distance (and pivoting point) of the proposed beam scanner. This telescope would increase by the magnification factor the transverse anterior scan range and the transverse resolution for the posterior segment scan. However, it would also reduce the transverse resolution for the anterior segment scan and the posterior angular range by the same factor. Therefore, the only effective solution for increasing both transverse and angular ranges, at the expenses of transverse resolution though, is the use of an adaptive telescope, where the second lens is another ETL, such that it creates a magnifying and demagnifying telescope for the anterior and posterior segment scan configurations, respectively.
It is also important to notice that in our experimental setup we only used a semi-aperture of the ETLs. By rotating the hollow roof mirror HRM and the periscope P by 180 deg, we could double the scan range for both anterior and posterior segment configurations, as the full aperture of the ELTs would be exploited. This would already lead for the current setup to a field of view of 5 mm for the telecentric scan and 2.4 deg for the angular scan. This feature would require to simultaneously rotate HRM and P about the beam scanner axis, which can be done with a simple low-cost DC or stepper motor (and a transmission system). Such actuation unlocks the potential of performing full 3-D scans by building radial sections or circular/spiral scan patterns through synchronization of the motor and the ETLs control (see Visualization 5).
Nevertheless, the current implementation can already find good use in low-cost biometry by providing anterior segment 2-D central anatomy for detection of lens tilt and decentration, for example, and posterior anatomy, such as the foveal pit for fixation checks. In fact, commercial biometers, such as the Zeiss IOL Master 700 [13], already implement a small angular scan around the foveola, a central pit of the fovea, about 0.35 mm in diameter (or ∼1.2 deg) [38], where only cone photoreceptors are present and which is specialized for maximum visual acuity [39]. Hence, it is the part of the retina engaged in fixation. By detecting a distorted image of the foveal pit, one can assume poor fixation, likely caused by the patient’s eye motion, and therefore apply a quality check on the biometry reading. With the proposed beam scanner, fixation check and central 2-D anterior segment biometry will be readily available. Additionally, by displacing the transverse field of view to the corneal limbus, regions of diagnostic interest, such as the iridocorneal angle, could be evaluated for close-angle glaucoma diagnostics.
5. Conclusion
In summary, we presented a novel optical beam scanner with reconfigurable non mechanical control of beam position, angle and focus using three low-cost ETLs and applied it to quasi-simultaneous whole-eye OCT imaging. We derived the analytical equations for the focal lengths of $ET{L_1}$-$ET{L_3}$ to control the beam axis position, angle and focal distance from the device, and we set them for alternately providing a telecentric scan of a focused beam for anterior segment imaging and an angular scan of a collimated beam for posterior segment imaging. The expected scan range was 2.5 mm and 1.2 deg, and the expected transverse resolution was varying in the range 35–102 μm and 23–31 μm, for the anterior and posterior segment scans, respectively. We characterized the dynamic scanning and focusing capabilities of an experimental setup through beam profiling, after a thorough calibration procedure. The experimental measurements were in good agreement with the analytical expectations. We integrated the proposed beam scanner in a SS-OCT system and acquired whole-eye OCT images of a model eye and of an ex vivo rabbit eye, revealing details comparable to those seen in conventional anterior and posterior OCT scanners.
The proposed beam scanner reduces the complexity and cost of other proposed whole-eye scanners based on mechanical switching or polarization multiplexing between the two segment scan configurations. Additionally, it provides the ability to reconfigure the scan configuration to best suit the specific application, i.e., it is not limited to using only pre-selected scan configurations. The current system would already prove useful in low-cost biometry applications to provide 2-D anterior segment anatomy and fixation checks for improving the repeatability of biometry readings and leading to fewer refractive surprises. Sourcing larger aperture ETLs will enable to overcome the current limitations on the transverse and angular scan ranges and expand its use to diagnostic and scientific application to glaucoma, myopia studies and beyond, in several other fields.
Funding
Comunidad de Madrid (IND2019/BMD-17262); Agencia Estatal de Investigación (PID2020-115191RB); Fundacja na rzecz Nauki Polskiej (MAB/2019/12); Research to Prevent Blindness (Unrestricted grant); National Eye Institute (P30EY001319).
Acknowledgments
We acknowledge the useful help of Maciej Wielgo at ICTER in processing the OCT data. The International Centre for Translational Eye Research (MAB/2019/12) project is carried out within the International Research Agendas Programme of the Foundation for Polish Science, co-financed by the European Union under the European Regional Development Fund.
Disclosures
MPU, EG, SM, AC: Patent (P).
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. M. Wojtkowski, B. Kaluzny, and R. J. Zawadzki, “New directions in ophthalmic optical coherence tomography,” Optometry Vis. Sci. 89(5), 524–542 (2012). [CrossRef]
2. R. Chopra, S. K. Wagner, and P. A. Keane, “Optical coherence tomography in the 2020s-outside the eye clinic,” Eye 35(1), 236–243 (2021). [CrossRef]
3. R. J. L. Branco, L. Yan, and H. David, “Clinical and research applications of anterior segment optical coherence tomography – a review,” Clin. Exp. Ophthalmol. 37(1), 81–89 (2009). [CrossRef]
4. P. Aristodemou, N. Knox Cartwright, J. Sparrow, and R. Johnston, “Improving refractive outcomes in cataract surgery: A global perspective,” World J. Ophthalmol. 4(4), 140–146 (2014). [CrossRef]
5. A. Sahin and P. Hamrah, “Clinically relevant biometry,” Curr. Opinion Ophthalmol. 23(1), 47–53 (2012). [CrossRef]
6. A. Epitropoulos, “Axial length measurement acquisition rates of two optical biometers in cataractous eyes,” Clin. Ophthalmol. 8, 1369–1376 (2014). [CrossRef]
7. A. Akman, L. Asena, and S. G. Güngör, “Evaluation and comparison of the new swept source OCT-based IOLMaster 700 with the IOLMaster 500,” Br. J. Ophthalmol. 100(9), 1201–1205 (2016). [CrossRef]
8. C. M. Yang, D. H. Lim, H. J. Kim, and T.-Y. Chung, “Comparison of two swept-source optical coherence tomography biometers and a partial coherence interferometer,” PLoS One 14, e0223114 (2019). [CrossRef]
9. I. O. Koumbo Mekountchou, F. Conrad, P. Sankaridurg, and K. Ehrmann, “Peripheral eye length measurement techniques: a review,” Clin. Exp. Optometry 103(2), 138–147 (2020). [CrossRef]
10. G. Schmid, “Axial and peripheral eye length measured with optical low coherence reflectometry,” J. Biomed. Opt. 8(4), 655–662 (2003). [CrossRef]
11. N. Porporato, M. Baskaran, and T. Aung, “Role of anterior segment optical coherence tomography in angle-closure disease: a review,” Clin. Exp. Ophthalmol. 46(2), 147–157 (2018). [CrossRef]
12. D. S. C. Ng, C. Y. L. Cheung, F. O. Luk, S. Mohamed, M. E. Brelen, J. C. S. Yam, C. W. Tsang, and T. Y. Y. Lai, “Advances of optical coherence tomography in myopia and pathologic myopia,” Eye 30(7), 901–916 (2016). [CrossRef]
13. W. Sekundo, “The IOL Master 700,” Cataract and Refractive Surgery Today Europe, 2015.
14. J. Ma, C. Pathak, A. Ramaprasad, and S. Basti, “Utility of IOLMaster 700 Swept-Source Optical Coherence Tomography in Detecting Macular Disease for Preoperative Cataract Surgery Patients,” Clin. Ophthalmol. 15, 3369–3378 (2021). [CrossRef]
15. A. P. Rein, A. Abulafia, D. Zadok, J. Hanhart, E. Gelman, and Y. Weill, “The Additive Value of Foveal OCT-Based Biometry to Fundus Biomicroscopy for Detecting Macular Pathology Prior to Cataract Surgery,” Am. J. Ophthalmol. 228, 8–15 (2021). [CrossRef]
16. A. N. Kuo, R. P. McNabb, and J. A. Izatt, “Advances in Whole-Eye Optical Coherence Tomography Imaging,” The Asia-Pacific J. Ophthalmol. 8, 99–104 (2019). [CrossRef]
17. W. Wang, W. Yan, A. Müller, and M. He, “A Global View on Output and Outcomes of Cataract Surgery With National Indices of Socioeconomic Development,” Invest. Ophthalmol. Visual Sci. 58(9), 3669–3676 (2017). [CrossRef]
18. A. Borgia, R. Raimondi, T. Sorrentino, F. Santoru, M. Buzzi, V. Borgia, V. Scorcia, and G. Giannaccare, “Swept-Source Optical Coherence Tomography-Based Biometry: A Comprehensive Overview,” Photonics 9(12), 951 (2022). [CrossRef]
19. D. Nankivil, G. Waterman, F. LaRocca, B. Keller, A. N. Kuo, and J. A. Izatt, “Handheld, rapidly switchable, anterior/posterior segment swept source optical coherence tomography probe,” Biomed. Opt. Express 6(11), 4516–4528 (2015). [CrossRef]
20. A. Rodríguez-Aramendía, F. Díaz-Doutón, J. Fernández-Trullàs, P. Falgueras, L. González, J. Pujol, I. Grulkowski, and J. L. Güell, “Whole anterior segment and retinal swept source OCT for comprehensive ocular screening,” Biomed. Opt. Express 12(3), 1263–1278 (2021). [CrossRef]
21. H.-J. Kim, P. U. Kim, M. G. Hyeon, Y. Choi, J. Kim, and B.-M. Kim, “High-resolution, dual-depth spectral-domain optical coherence tomography with interlaced detection for whole-eye imaging,” Appl. Opt. 55(26), 7212–7217 (2016). [CrossRef]
22. S. Fan, L. Li, Q. Li, C. Dai, Q. Ren, S. Jiao, and C. Zhou, “Dual band dual focus optical coherence tomography for imaging the whole eye segment,” Biomed. Opt. Express 6(7), 2481–2493 (2015). [CrossRef]
23. H.-J. Kim, M. Kim, M. G. Hyeon, Y. Choi, and B.-M. Kim, “Full ocular biometry through dual-depth whole-eye optical coherence tomography,” Biomed. Opt. Express 9(2), 360–372 (2018). [CrossRef]
24. R. McNabb, J. Polans, B. Keller, M. Jackson-Atogi, C. James, R. Vann, J. Izatt, and A. Kuo, “Wide-field whole eye OCT system with demonstration of quantitative retinal curvature estimation,” Biomed. Opt. Express 10(1), 338 (2019). [CrossRef]
25. I. Grulkowski, S. Manzanera, L. Cwiklinski, F. Sobczuk, K. Karnowski, and P. Artal, “Swept source optical coherence tomography and tunable lens technology for comprehensive imaging and biometry of the whole eye,” Optica 5(1), 52–59 (2018). [CrossRef]
26. D. Benton, “Non-mechanical beam steering: ways and means,” Proc. SPIE 10797, 107907H (2018). [CrossRef]
27. N. Savidis, G. Peyman, N. Peyghambarian, and J. Schwiegerling, “Nonmechanical zoom system through pressure-controlled tunable fluidic lenses,” Appl. Opt. 52(12), 2858–2865 (2013). [CrossRef]
28. M. Zohrabi, R. H. Cormack, and J. T. Gopinath, “Wide-angle nonmechanical beam steering using liquid lenses,” Opt. Express 24(21), 23798–23809 (2016). [CrossRef]
29. M. P. Urizar, A. de Castro, E. Gambra, O. Cetinkaya, S. Marcos, and A. Curatolo, “Design of a low-cost, versatile, whole-eye scanner for optical coherence tomography,” in Biophotonics Congress: Biomedical Optics 2022 (Translational, Microscopy, OCT, OTS, BRAIN), Technical Digest Series (Optica Publishing Group, 2022), CS2E.5.
30. B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics (Wiley, 2019).
31. O. M. Carrasco-Zevallos, C. Viehland, B. Keller, R. P. McNabb, A. N. Kuo, and J. A. Izatt, “Constant linear velocity spiral scanning for near video rate 4D OCT ophthalmic and surgical imaging with isotropic transverse sampling,” Biomed. Opt. Express 9(10), 5052–5070 (2018). [CrossRef]
32. A. Curatolo, J. S. Birkenfeld, E. Martinez-Enriquez, J. A. Germann, G. Muralidharan, J. Palací, D. Pascual, A. Eliasy, A. Abass, J. Solarski, K. Karnowski, M. Wojtkowski, A. Elsheikh, and S. Marcos, “Multi-meridian corneal imaging of air-puff induced deformation for improved detection of biomechanical abnormalities,” Biomed. Opt. Express 11(11), 6337–6355 (2020). [CrossRef]
33. D. Atchison, “Schematic eyes,” in Handbook of Visual Optics: Fundamentals and Eye Optics, P. Artal, ed. (Routledge, 2017), pp. 235–248.
34. C. Dorronsoro, X. Barcala, E. Gambra, V. Akondi, L. Sawides, Y. Marrakchi, V. Rodriguez-Lopez, C. Benedi-Garcia, M. Vinas, E. Lage, and S. Marcos, “Tunable lenses: dynamic characterization and fine-tuned control for high-speed applications,” Opt. Express 27(3), 2085–2100 (2019). [CrossRef]
35. P. D. Woolliams, R. A. Ferguson, C. Hart, A. Grimwood, and P. H. Tomlins, “Spatially deconvolved optical coherence tomography,” Appl. Opt. 49(11), 2014–2021 (2010). [CrossRef]
36. Y. Marrakchi, X. Barcala, E. Gambra, I. Martinez-Ibarburu, C. Dorronsoro, and L. Sawides, “Experimental characterization, modelling and compensation of temperature effects in optotunable lenses,” Sci. Rep. 13(1), 1575 (2023). [CrossRef]
37. L. Chen, M. Ghilardi, J. J. C. Busfield, and F. Carpi, “Electrically Tunable Lenses: A Review,” Front. Robot. AI 8, 678046 (2021). [CrossRef]
38. D. M. Sampson, A. M. Dubis, F. K. Chen, R. J. Zawadzki, and D. D. Sampson, “Towards standardizing retinal optical coherence tomography angiography: a review,” Light: Sci. Appl. 11(1), 63 (2022). [CrossRef]
39. A. M. Dubis, B. R. Hansen, R. F. Cooper, J. Beringer, A. Dubra, and J. Carroll, “Relationship between the Foveal Avascular Zone and Foveal Pit Morphology,” Invest. Ophthalmol. Visual Sci. 53(3), 1628–1636 (2012). [CrossRef]
