The spatial resolution of adaptive optics-optical coherence tomography (AO-OCT) enables visualization of retinal components (e.g., photoreceptors), which can advance clinical diagnosis of ocular diseases. Currently, however, variability in AO-OCT system performance suggests a need for standardized physical models, or “phantoms”, that replicate the opto-structural properties of retinal components. Here we employ direct laser writing – a two-photon polymerization-based additive manufacturing technology – using a light scattering titanium (IV) dioxide nanoparticle-laden photomaterial to create novel retinal cone outer segment (OS) phantoms. Fabrication results demonstrate effective mimicry of cone OS diameter and spacing at four retinal eccentricities, while AO-OCT experiments demonstrate cone spacings >4 µm can be resolved. These results suggest a promising methodology for biomimetic photoreceptor phantom development for AO-OCT.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
12 November 2020: A correction was made to a link to Visualization 4.
Over the past two decades, adaptive optics (AO) approaches have been applied to many imaging modalities to correct for optical aberrations and enhance lateral resolution. While AO offers significant benefits to probe microscopic structures and cells for earlier and more precise disease detection [1,2], the technology has not yet seen widespread translation to the clinic. One of the key barriers to AO reaching the clinic is the lack of standardized performance tools and methods for this highly personalized imaging modality, which is configured differently for each subject. One such tool is a phantom model, which can provide standardized performance metrics such as spatial resolution between AO systems, thereby minimizing the likelihood of measurement inconsistencies and clinical variability [3–5]. One of the modalities that has been advanced by AO approaches is optical coherence tomography (OCT), which is a low coherence interferometric imaging modality that is used in ophthalmology to provide three-dimensional (3D) information of retina layers for disease detection . Due to the presence of ocular aberrations, clinical OCT systems often operate with small imaging beacon size (1.5-2mm) at the pupil of the eye, which limits the retinal image lateral resolution . To expand OCT lateral resolution capabilities, AO approaches have been used to visualize cellular-level retinal details, such as individual cone and rod photoreceptor cells [8–10]. However, there remains no standardized retina phantom to evaluate the micro-scale resolution of AO-OCT systems.
Most prior OCT retina phantom studies have focused on modeling the thickness, reflectivity, and macroscopic geometries (e.g., the foveal pit and optic nerve) of the retina and its layers [3,11–13]. While these studies have produced valuable tools to analyze clinical OCT performance, they have not captured localized retinal micro-scale architecture, the detection of which has been enabled by AO. As a result, such layered samples are not ideally suited for the standardization of AO-OCT performance. To address this need, researchers have turned to photolithography to generate retinal cone phantoms that simulate photoreceptor cell waveguiding behavior [14,15]. Inherent planar limitations of photolithography, however, constrain such phantoms to cylindrical geometries, which do not accurately capture the micro-scale morphology of retinal cones . With the advent of additive manufacturing methods for the creation of biomimetic architectures [17–19], one pathway we have recently explored for retinal phantom development is the use of Direct Laser Writing (DLW). DLW is a three-dimensional printing method that uses a femtosecond pulsed infrared laser to solidify liquid-phase photoreactive polymers in a point-by-point, layer-by-layer manner via two-photon (or multi-photon) absorption phenomena . The unique advantage of this technology is the 100-nm minimum feature resolution, which allows for micro- and nano-scale features to be constructed with 3D geometrical complexity . Using DLW, we previously created retinal cone phantoms with tapered cylindrical architectures, which provide improved structural stability over standard cylindrical components [22,23]. Although these preliminary experiments demonstrate the value of DLW in producing controlled 3D geometries, our 3D printed phantoms were not able to capture retinal cone architectures at biological aspect ratios (i.e., height-to-width ratios).
In this work, we expand upon our previous research by using DLW to manufacture a novel microfluidic-based retinal cone OS phantom platform with greater anatomical relevance than has been previously possible. The platform consists of four sections of phantom arrays that are inspired by the anatomical geometry, aspect ratio, and spacing of the outer segment (OS) portions of human retinal cone cells at 1°, 2.5°, 5°, and 10°eccentricities [Fig. 1(a)]. To facilitate AO-OCT visibility of the phantom components, we use a titanium (IV) dioxide (TiO2) nanoparticle-doped photomaterial to provide light scattering directly within 3D nanoprinted components. This builds upon previous research that successfully combined TiO2 nanoparticles and a printed material of interest to generate arbitrary geometries such as woodpile structures [24–27]. For the fabrication of phantom components, we leverage an in situ DLW (isDLW) strategy previously established by our group  to manufacture components directly within enclosed microfluidic channels. This strategy supports the fabrication of our high-aspect-ratio (>15) phantom components by allowing microstructures to be fully adhered to the luminal surface of the microfluidic channel [Fig. 1(b)], thus anchoring them in place during printing and post-processing.
2. Materials and methods
2.1 Microchannel fabrication
For the material composition of the microfluidic devices, cyclic olefin polymer (COP) was selected due its strong adhesion to DLW-printed photopolymers and relative ease of manufacturing . To create COP-on-COP devices, negative microchannel master molds were first DLW-printed with the Photonic Professional GT (Nanoscribe, Karlsruhe, Germany). This was done by using a 25mm × 25mm square silicon substrate to support the manufacturing of the proprietary IP-S photoresist (Nanoscribe) in the dip-in laser lithography configuration  with a 25×, N.A. = 0.8 objective lens at a wavelength of 780 nm and pulse duration of ∼100 fs. The resulting 35-µm tall molds were designed with two sections: (i) 100 × 1000 µm (W × L) end portions with semi-ovular cross-sections and (ii) 500 × 500 µm central portion with a tapered rectangular cross-section [Figs. 1(c)–1(e)]. After printing, the substrate was placed in a bath of propylene glycol monomethyl ether acetate (PGMEA) for 20 minutes to clear the uncured (i.e., liquid phase) photoresist from the mold. Subsequently, the mold was placed in a bath of isopropanol (IPA) for 2 minutes to rinse the PGMEA from the mold. To create the microfluidic channel, a blank monolith of COP (nD = 1.53) was placed over the mold and hot embossed into a ∼15° wedge shape [Fig. 1(c)] at 120°C for 3 minutes. As was found in previous experiments , this wedge atop the sample was pivotal to deflecting light reflections that can saturate the wavefront sensor and prevent AO correction. After hot embossing, the heat was turned off and the COP was held under pressure for 10 minutes and subsequently peeled from the negative master mold, drilled with inlet and outlet holes, and rinsed with IPA and N2 gas. Finally, the devices were sealed through a vapor-based bonding of 100-µm thick COP film. This was done by placing one side of the film over a 30°C cyclohexane bath for 2 minutes and quickly pressing the molded COP wedge onto the cyclohexane-exposed side of the film. The device was then placed in an oven overnight at 60°C to eliminate residual vapor from the film surface and ensure a strong permanent bond between the two layers.
2.2 TiO2-laden photomaterial preparation and testing
To create the TiO2-laden photomaterial, TiO2 nanoparticles (∼21 nm, Sigma-Aldrich, St. Louis, MO, USA) were added to a stock of IP-Dip photoresist (nD = 1.548 , Nanoscribe) at a ratio of 0.1% by mass. The two components were mixed in a three-stage process that consisted of: (i) manual hand mixing, (ii) probe sonication of the mixture for 1 minute, and (iii) continuous stirring of the mixture with a magnetic stir bar for 24 hours at 45°C. For initial testing of particle-doped photoresist homogeneity, a droplet of 0.1% TiO2 photoresist was sandwiched between a microscope slide and a coverslip, spot cured for 60 seconds with maximum intensity light exposure from a brightfield microscope, and imaged under dark-field conditions with a 20×, N.A. = 0.4 plan-neofluor objective lens (Zeiss, Oberkochen, Germany). Subsequently, the fabrication feasibility of the 0.1% TiO2 photoresist was tested by placing a droplet of the material on a #1.5 thick 30mm circular glass coverslip (Bioptechs, Inc., Butler, PA, USA) and printing test structures in the oil-immersion mode using the Photonic Professional GT. After being post processed in a similar manner to the microchannel molds, the test structures were gold sputter-coated and imaged via scanning electron microscopy (SEM) (SU-70, Hitachi, Chiyoda City, Tokyo, Japan).
2.3 Retinal cone OS phantom fabrication
The 3D architecture for the cone OS phantom arrays was designed using Solidworks (Dassault Systèmes, Vélizy-Villacoublay, France) and the laser path used to construct the components via DLW was defined using DeScribe (Nanoscribe). Once mixed, the particle-doped photomaterial was loaded into a COP microchannel by placing a droplet of the photomaterial over the inlet port and applying a gentle vacuum pressure at the outlet port with a syringe [Fig. 1(d)]. The microfluidic device was then loaded into the printer in the oil-immersion configuration and isDLW was conducted with a 63×, N.A. = 1.4 objective lens in a ceiling-to-floor manner such that the tops of the phantoms were fastened to the ceiling, thus stabilizing the structures throughout fabrication [Figs. 1(e), 1(f)]. The 35 µm tall retinal cone OS phantoms were fabricated in two identical rows of four adjacent 75 µm × 75 µm arrays, mimicking retinal cone OS geometries at, 1°, 2.5°, 5°, and 10°eccentricities (i.e., central diameters of 1.48 µm, 1.59 µm, 1.76 µm, and 2.2 µm and center-to-center spacings of 4.63 µm, 6.95 µm, 8.79 µm, and 11.69 µm, respectively) [16,32,33]. Additionally, the two ends of each pillar were designed with an outward taper (diameters of 3 µm, 3 µm, 3.17 µm, and 4 µm) to increase the surface area contacting the COP microchannel for increased phantom adhesion. DLW parameters for each of these sections consisted of a layer height of 200 nm, hatch spacing of 200 nm, laser power of 27.5 mW and scanning speeds of 20, 17.5, 15, and 10 mm/s respectively, all at a wavelength of 780 nm and pulse duration of ∼100 fs. After the completion of isDLW, the device was removed from the printer, connected to a Fluigent Microfluidic Control System (Fluigent, Paris, France), and perfused with PGMEA at 10-50 kPa for 5 minutes to remove uncured photoresist and subsequently washed with IPA at 10 kPa for 2 minutes [Fig. 1(g)]. Lastly, to remove IPA from the channel while reducing the likelihood of stiction-based collapse, the COP device was dried in a critical point dryer (Tousimis, Rockville, MD, USA), [Fig. 1(h)].
2.4 Retinal cone OS phantom imaging
After fabrication, development, and critical point drying of the phantoms, the TiO2-doped sample was first analyzed via fluorescence microscopy with the same 20×, N.A. = 0.4 plan-neofluor objective lens as with dark-field imaging at an excitation wavelength of 487 nm and emission wavelength of 503 nm. This imaging was facilitated by the autofluorescence of the IP-Dip photoresist, which emits a strong fluorescent signal when excited in the UV domain and lower portion of the visible light domain (see Visualization 1). After imaging, ImageJ (NIH, Bethesda, MD, USA) and MATLAB (Mathworks, Natick, MA, USA) were used to analyze the dimensions of the components (i.e., phantom base diameter and center-to-center spacing, see Visualization 2). To do this, first an image of each 75 µm × 75 µm array was isolated and converted to 16-bit format (Visualization 2a,b), and thresholded with the “IsoData” algorithm (Visualization 2c) . This is an iterative algorithm that computes a threshold value that is exactly halfway between the mean gray value of the object and the mean gray value of the background. Once thresholded, the diameter of each circular area was computed by the software, and the centroid location for each phantom was located (Visualization 2d). Lastly, a MATLAB script was written to compute and plot the average Euclidian distance between the centroid of each cone OS phantom and the neighboring phantoms.
Prior to AO-OCT imaging, the wedged microfluidic devices were processed in two steps to facilitate more idealized imaging. First, the flat back side of the COP microfluidic device was roughened to provide a diffuse reflection that fills the AO pupil for more ideal wavefront sensing. Additionally, an immersion oil closely matching the refractive index of COP (nD = 1.536, Cargille Laboratories Inc., Cedar Grove, NJ, USA) was placed within the channel and dropped on top of the wedged surface and a glass coverslip was secured atop the oil droplet. This step helped to smooth the wedged surface and maintain a consistent refractive index throughout the device, both of which were critical for greater AO-OCT resolution. Once prepared, the wedged device was mounted into a model eye composed of entirely off-the-shelf components [Fig. 1(h)], including an adjustable iris (Thorlabs, Newton, NJ, USA), a neutral density filter (Thorlabs), a 19 mm focal length achromatic doublet lens (Edmund Optics, Barrington, NJ, USA), and an X/Y translational stage (Thorlabs). All AO-OCT imaging experiments were conducted with a 1° scan size on a research-grade, custom-built multimodal AO system at the U.S. Food and Drug Administration .
3.1 Fabrication results of TiO2-laden retinal cone OS phantoms
As a preliminary analysis of the homogeneity of particle mixing, the 0.1% TiO2 sample was analyzed via dark-field microscopy. Imaging results (see Visualization 3) demonstrate a largely homogeneous mixing procedure, with very few instances of particle aggregation, which are negligible during fabrication. To initially determine the manufacturing feasibility of the particle-laden material, cello-shaped test components were manufactured outside of microfluidic channels with 0.1% TiO2-laden IP-Dip and control (i.e., particle-free) IP-Dip. SEM results depict the presence of TiO2 particles on and beneath the surface of the 0.1% TiO2 sample (Visualization 4a,c). However, this particle accumulation does not appear to drastically alter the geometry of the component, as the structure is virtually identical to the control sample (Visualization 4b,d). For the fabrication of high-aspect-ratio cone OS phantoms, videos were recorded for both the 0.1% TiO2-laden and control photomaterials. Computer-aided manufacturing (CAM) simulations and corresponding fabrication frames are depicted in Figs. 2(a)–2(c) (see also Visualization 5). Similar to the cello structures, phantom print results suggest that the presence of light scattering nanoparticles [Fig. 2(b)] does not negatively affect the manufacturing of high-aspect-ratio structures. Following fabrication and post-processing, a fluorescence micrograph of TiO2-laden OS phantoms was collected to further analyze fabrication and post-processing success [Fig. 3(a)]. To quantify the fabrication quality for the TiO2-laden phantoms, base diameter and center-to-center spacing were measured from the fluorescence micrograph. Phantom base diameter measurements [Fig. 3(b)] demonstrate fidelity between printed phantoms and computationally designed structures, as the average experimental phantom diameter was 2.71 ± 0.16 µm, 2.80 ± 0.13 µm, 3.00 ± 0.08 µm, and 3.97 ± 0.11 µm for 1°, 2.5°, 5°, and 10° eccentricities respectively, each within 0.30 µm of the designed value for all eccentricity arrays (mean: -0.17 ± 0.11 µm). Furthermore, the center-to-center spacing measurements [Fig. 3(c)] were found to be 4.76 ± 0.05 µm, 7.15 ± 0.07 µm, 9.02 ± 0.08 µm, and 11.99 ± 0.11 µm for 1°, 2.5°, 5°, and 10° eccentricities respectively, with an average discrepancy of 0.21 ± 0.07 µm between experimental results and the designed values.
3.2 AO-OCT results for retinal cone OS phantoms
As stated previously, the back side of each microfluidic device was roughened prior to AO-OCT imaging to facilitate effective wavefront sensing and AO correction. Although scotch tape was molded with PDMS in previous experiments to provide surface roughness , this technique did not work for COP. In this study, 800-grit sandpaper was used to roughen the back side of the device, as it was found to yield a comparable surface roughness to scotch tape (RaTape = 0.68 ± 0.31 µm, RaSP = 0.35 ± 0.62 µm, Visualization 6). For AO-OCT imaging, three samples were tested, each with a distinct material composition based on particle concentration: (i) 0.1% TiO2 in IP-Dip, (ii) 0.05% TiO2 in IP-Dip, and (iii) particle-free IP-Dip as a control. To analyze the quality of the phantom structures, en face AO-OCT images [Figs. 4(a)–4(c)] were produced by averaging the brightness intensity of all pixels through the height of the microchannel. The images shown in Figs. 4(a)–4(c) are presented with a logarithmic intensity scale. The two rows of 0.1% TiO2-laden components [Fig. 4(a)], 1°, 2.5°, 5°, and 10° left to right) depict regular spacing of retinal cone OS phantoms that are visually resolved at the 2.5° eccentricity and above. For the 0.1% TiO2 sample in Fig. 4(a), the difference between pillar diameters in the 2.5°, 5°, and 10° sections appears more drastic than the true diameters known from the design and revealed in the fluorescence image in Fig. 3(a). This direct dependence of AO-imaged pillar diameter on pillar spacing is congruent with previous findings from our group [22,23], which we have attributed to speckle and waveguiding effects. The 1° phantoms are not resolved and exhibit only a random speckle pattern typical of OCT images from bulk scattering material. For the 0.05% TiO2 case [Fig. 4(b)], phantoms remain visible, but appear to have lower brightness, less circularity, and less regular spacing than the 0.1% TiO2 case. It is important to note, however, that the diminished circularity of the phantoms in the 0.05% case may have resulted from astigmatism, as AO correction was not as ideal in this case as in the 0.1% case. For the control case [Fig. 4(c)], visibility of the pillars is largely limited, and the phantom arrays were not clearly identified.
To supplement the en face imaging results, cross-sectional AO-OCT B-scan images [Figs. 4(d), 4(e)] are presented on a linear intensity scale for representative rows of the particle-doped phantoms at each eccentricity. The location of the B-scan profile is identified by blue lines in Figs. 4(a), 4(b). For all cases, the photoreceptor phantoms appear to be vertically oriented, ranging from the ceiling to floor of the microchannel with regular spacing, thus further demonstrating the success of fabrication and post-processing. The 0.1% TiO2 results [Fig. 4(d)] depict distinctly identifiable pillars at the 10° eccentricity. However, at the 5° and 2.5° eccentricities, while some pillars appear distinguishable, the random speckle generated by TiO2 light scattering makes pillar resolution more challenging. Similarly, for the 0.05% TiO2 results [Fig. 4(e)], pillars appear to be weakly resolved at the 10° and 5° eccentricities, whereas pillars become more difficult to distinguish at the 2.5° eccentricity. In either case, the 1° eccentricity image appears as a random speckle pattern with indistinguishable pillars. Although localized TiO2 particle aggregation appears to produce brightness intensity heterogeneities throughout the cross-sections of each sample, there appears to be less brightness (i.e., light scattering) in the cross-sections of the 0.05% TiO2 phantom arrays than in the 0.1% case, as was validated by en face images.
Furthermore, to elucidate the role of particle concentration in phantom visibility, the brightness intensity was measured for each individual pillar within the 2.5°, 5°, and 10° eccentricity arrays in the en face images of the 0.1% and 0.05% TiO2 samples [Fig. 4(f)]. This was done by averaging the logarithmic intensity values at the centroid locations of all phantoms in each 75 µm × 75 µm array. Centroid brightness was selected as a representative metric for eccentricity array comparison due to the challenge in thresholding individual pillars in the en face AO-OCT images from Figs. 4(a), 4(b). In the 0.1% TiO2 sample, results reveal average intensities of 7.2 ± 0.6, 7.1 ± 0.7, and 8.6 ± 0.8 A.U. for 2.5°, 5°, and 10° arrays respectively, whereas in the 0.05% sample, average intensities were found to be 6.5 ± 0.8, 6.9 ± 0.8, and 7.8 ± 1.0 A.U for 2.5°, 5°, and 10° arrays.
In this work, we have used manufacturing recordings [Fig. 2, Visualization 5] and fluorescence microscopy [Fig. 3] to demonstrate successful isDLW fabrication and post-processing of retinal cone OS phantoms with a TiO2-doped photomaterial, as all OS phantoms appear regularly arrayed, vertically oriented, and structurally intact (i.e., no signs of stiction-based collapse). Although quantifications of phantom base diameter and center-to-center spacing demonstrate small variations (<0.3 µm) of the printed components from their designed values, this error is within the tolerance of the printed components and below the sensitivity range of current AO-OCT systems. One potential source of error is shrinkage of the photopolymer after fabrication, which is a well-established natural effect of photopolymers produced with DLW [36,37]. If necessary, this shrinkage could be managed by controlling the microfluidic environment to minimize phantom exposure to air. The successful fabrication of the OS cone phantoms in this study suggest a pathway for alternative tissue phantoms to be generated with isDLW and scattering particle-doped photomaterials. Furthermore, to expand upon the single-material OS phantoms demonstrated here, future experiments could target multi-material DLW approaches [38,39] to produce segmented phantom components with distinct optical, mechanical, and/ or geometric properties at distinct locations in 3D space. For example, more advanced retinal cone phantoms could be produced with two materials that mimic the geometry and scattering properties of the inner and outer segments of the cone.
When considering the AO-OCT performance of particle-laden phantoms, the 0.1% TiO2 particle concentration appears to provide a good combination of phantom brightness and structural detail. En face images [Figs. 4(a)-4(c)] demonstrate that the presence and concentration of light scattering particles is critical to visualization of the phantoms. Although small, irregular incidents of light scattering can be seen in the particle-free control image in Fig. 4(c), the source of the scattering is believed to be from topographical and material inconsistencies within the channel. For instance, the cluster of scattering along the left side and top side of the image appears to come from dust that entered the channel and was trapped along the perimeter of the phantom arrays. Furthermore, the instances of light scattering from OS pillars, particularly in the 10° eccentricity array, do not appear to come from the printed components themselves, but rather from topographical changes induced at the microchannel surfaces by the high energy of the two-photon laser. When comparing the 0.1% and 0.05% TiO2 materials, the greater particle concentration appears more ideal for inducing phantom brightness and AO-OCT visualization [Fig. 4(f)]. B-scan images offer insights into the reason for the brightness differences in en face images, as there appears to be more consistent light scattering throughout the height of 0.1% TiO2 samples than the 0.05% TiO2 samples. This is most notable when comparing 2.5° cross-sectional images, as light scattering appears more inconsistent in individual pillars for the 0.05% sample. These light scattering differences are a direct result of refractive index disparities between TiO2 particles and the surrounding materials. Since the index of IP-Dip photoresist (nD = 1.548) is similar to that of COP (nD = 1.53) and the immersion oil (nD = 1.536), the higher-index TiO2 particles (nD = ∼2.5) act as the chief source of light scattering within the phantoms. Therefore, greater concentrations of particles result in larger refractive index differences and higher intensity in AO-OCT images. The surface roughness of the printed phantoms may also play a role in scattering light for OCT imaging, but is assumed to be minimal, as the roughness is well below the resolution limit of the AO-OCT system and is relatively consistent between each of the samples, as verified in Visualization 4. An important note, however, is that although the 0.1% TiO2 particle concentration provided improved light scattering and AO-OCT resolution over the 0.05% TiO2 case, further increases in particle concentration could have detrimental effects. For instance, although greater particle concentrations appear to offer benefits for AO-OCT visibility, excessive light scattering and brightness could lead the pillars to appear fused together, particularly at lower eccentricities, which would diminish apparent imaging resolution. Future studies could be conducted to elucidate a more ideal particle concentration that simultaneously maximizes AO-OCT resolution and minimizes this expansion effect.
Interestingly, at 1° eccentricity, the difference in material performance was difficult to elucidate, as individual pillars in each sample are unresolvable, as shown in Figs. 4(a), 4(b) and Figs. 4(d), 4(e) (left). There are three potential reasons for this result: (i) the liquid photoresist was not fully cleared from between the tightly packed phantoms, (ii) there were defects that arose within the phantom after infusion of the index-matching oil into the microfluidic channel, or (iii) the diameter and spacing of the phantoms have reached the resolution limit of the AO-OCT system. It is likely that all three played a role. For reason (i), the small fluorescent glow surrounding the 1° phantoms in the fluorescence micrograph in Fig. 3(a) suggests the presence of residual liquid photoresist between the phantoms. This could be addressed in future experiments by conducting more extensive development steps to ensure more complete clearance of liquid photoresist. For reason (ii) there appear to be occasional instances of pillars either missing from the arrays or displaced from their original location in en face images. This could suggest that the oil infusion prior to imaging could have had deleterious effects on the microchannel or printed components. In fact, this conclusion is particularly likely, as the COP was found to be damaged by the index-matching oil after extended periods of exposure (i.e., hours to days). In future experiments, a different index-matching material such as a curable photomaterial could be used in lieu of an index matching oil to address these concerns and create a phantom with greater long-term stability. Associated with reason (iii) is the possibility that the phantoms are either exceeding the practical resolution limit of the imaging system, or that local inconsistencies in the material properties (e.g., transparency and refractive index) within the COP monolith introduced wavefront aberrations which degraded the resolution of the AO-OCT imaging system. Although COP is ideal for the anchoring printed photopolymers within the channel, alternative microfluidic platforms such as glass-on-glass could be explored to circumvent this potential challenge. Furthermore, multi-material DLW approaches could be leveraged to replace the immersion oil with a photocurable material that will seal retinal cone phantoms in place indefinitely to facilitate long-term phantom stability for use in the AO-OCT community.
DLW is an additive manufacturing technique that allows photoreactive materials to be fabricated into complex 3D geometries with submicron-scale precision. We have demonstrated a unique application of the isDLW technique using a novel TiO2-doped photomaterial to create retinal cone OS phantoms with aspect ratios and geometries that mirror anatomic architectures with higher accuracy than has been previously reported. Experimental results demonstrate the need for a light-scattering agent within the photomaterial to permit visibility of phantoms with AO-OCT, as small diameter, tightly packed structures (e.g., at 1°, 2.5°, and 5° eccentricity) did not provide sufficient diffuse light scattering when produced with a standard photopolymer. Alternatively, the inclusion of particles within the phantoms facilitated resolution down to 2.5° eccentricity, particularly for a 0.1% TiO2 particle concentration. Furthermore, the inclusion of particles within the photopolymer did not cause any deleterious effect to the DLW fabrication process. These results signify the advent of a valuable new manufacturing technique with the potential to usher in a new class of phantoms with highly controlled optical scattering and 3D geometric properties at sub-micron scales. Our strategy offers a promising pathway towards the production of anatomically accurate photoreceptor phantoms for AO-OCT and can also be leveraged for other applications. We envision an expansion of this technology for the generation of phantoms for other micro-scale tissues such as retinal ganglion cells, microglia, and microvasculature that can be used to assess and standardize the performance of micro-scale imaging strategies not limited to AO-OCT, including alternative OCT applications and AO-scanning laser ophthalmoscopy.
U.S. Food and Drug Administration.
The authors graciously thank the TerrapinWorks Research Prototyping Laboratory and the Maryland Nanocenter at the University of Maryland, College Park, for granting equipment access and training throughout the course of this study. This project was supported in part by an appointment to the ORISE Research Participation Program at the Center for Devices and Radiological Health (CDRH) administered by the Oak Ridge Institute for Science and Education through an agreement between the U.S. Department of Energy and CDRH.
The authors declare no conflicts of interest.
Disclaimer: The mention of commercial products, their sources, or their use in connection with material reported herein is not to be construed as either an actual or implied endorsement of such products by the U.S. Department of Health and Human Services.
1. K. Kurokawa, Z. Liu, and D. T. Miller, “Adaptive optics optical coherence tomography angiography for morphometric analysis of choriocapillaris [Invited],” Biomed. Opt. Express 8(3), 1803–1822 (2017). [CrossRef]
2. D. T. Miller, O. P. Kocaoglu, Q. Wang, and S. Lee, “Adaptive optics and the eye (super resolution OCT),” Eye 25(3), 321–330 (2011). [CrossRef]
3. A. Agrawal, J. Baxi, W. Calhoun, C.-L. Chen, H. Ishikawa, J. S. Schuman, G. Wollstein, and D. X. Hammer, “Optic Nerve Head Measurements With Optical Coherence Tomography: A Phantom-Based Study Reveals Differences Among Clinical Devices,” Invest. Ophthalmol. Visual Sci. 57(9), OCT413 (2016). [CrossRef]
4. B. W. Pogue and M. S. Patterson, “Review of tissue simulating phantoms for optical spectroscopy, imaging and dosimetry,” J. Biomed. Opt. 11(4), 041102 (2006). [CrossRef]
5. G. Lamouche, B. F. Kennedy, K. M. Kennedy, C.-E. Bisaillon, A. Curatolo, G. Campbell, V. Pazos, and D. D. Sampson, “Review of tissue simulating phantoms with controllable optical, mechanical and structural properties for use in optical coherence tomography,” Biomed. Opt. Express 3(6), 1381–1398 (2012). [CrossRef]
6. M. Adhi and J. S. Duker, “Optical coherence tomography – current and future applications:,” Curr. Opin. Ophthalmol. 24(3), 213–221 (2013). [CrossRef]
7. Y. Jian, R. J. Zawadzki, and M. V. Sarunic, “Adaptive optics optical coherence tomography for in vivo mouse retinal imaging,” J. Biomed. Opt. 18(5), 056007 (2013). [CrossRef]
8. M. Pircher and R. J. Zawadzki, “Review of adaptive optics OCT (AO-OCT): principles and applications for retinal imaging [Invited],” Biomed. Opt. Express 8(5), 2536–2562 (2017). [CrossRef]
9. S. A. Burns, A. E. Elsner, K. A. Sapoznik, R. L. Warner, and T. J. Gast, “Adaptive optics imaging of the human retina,” Prog. Retinal Eye Res. 68, 1–30 (2019). [CrossRef]
10. 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(12), 4792–4811 (2005). [CrossRef]
11. A. Agrawal, C.-W. Chen, J. Baxi, Y. Chen, and T. J. Pfefer, “Multilayer thin-film phantoms for axial contrast transfer function measurement in optical coherence tomography,” Biomed. Opt. Express 4(7), 1166–1175 (2013). [CrossRef]
12. J. Baxi, W. Calhoun, Y. J. Sepah, D. X. Hammer, I. Ilev, T. Joshua Pfefer, Q. D. Nguyen, and A. Agrawal, “Retina-simulating phantom for optical coherence tomography,” J. Biomed. Opt. 19(2), 021106 (2014). [CrossRef]
13. G. J. Greening, R. Istfan, L. M. Higgins, K. Balachandran, D. Roblyer, M. C. Pierce, and T. J. Muldoon, “Characterization of thin poly(dimethylsiloxane)-based tissue-simulating phantoms with tunable reduced scattering and absorption coefficients at visible and near-infrared wavelengths,” J. Biomed. Opt. 19(11), 115002 (2014). [CrossRef]
14. D. Valente and B. Vohnsen, “Retina-simulating phantom produced by chemically amplified photolithography,” in Frontiers in Optics 2016 (OSA, 2016), p. FW3A.3.
15. D. Valente and B. Vohnsen, “Retina-simulating phantom produced by photolithography,” Opt. Lett. 42(22), 4623–4626 (2017). [CrossRef]
16. C. A. Curcio, K. R. Sloan, R. E. Kalina, and A. E. Hendrickson, “Human photoreceptor topography,” J. Comp. Neurol. 292(4), 497–523 (1990). [CrossRef]
17. D. B. Kolesky, R. L. Truby, A. S. Gladman, T. A. Busbee, K. A. Homan, and J. A. Lewis, “3D Bioprinting of Vascularized, Heterogeneous Cell-Laden Tissue Constructs,” Adv. Mater. 26(19), 3124–3130 (2014). [CrossRef]
18. J. Koffler, W. Zhu, X. Qu, O. Platoshyn, J. N. Dulin, J. Brock, L. Graham, P. Lu, J. Sakamoto, M. Marsala, S. Chen, and M. H. Tuszynski, “Biomimetic 3D-printed scaffolds for spinal cord injury repair,” Nat. Med. 25(2), 263–269 (2019). [CrossRef]
19. A. Marino, O. Tricinci, M. Battaglini, C. Filippeschi, V. Mattoli, E. Sinibaldi, and G. Ciofani, “A 3D Real-Scale, Biomimetic, and Biohybrid Model of the Blood-Brain Barrier Fabricated through Two-Photon Lithography,” Small 14(6), 1702959 (2018). [CrossRef]
20. T. Gissibl, S. Thiele, A. Herkommer, and H. Giessen, “Two-photon direct laser writing of ultracompact multi-lens objectives,” Nat. Photonics 10(8), 554–560 (2016). [CrossRef]
21. P.-I. Dietrich, M. Blaicher, I. Reuter, M. Billah, T. Hoose, A. Hofmann, C. Caer, R. Dangel, B. Offrein, U. Troppenz, M. Moehrle, W. Freude, and C. Koos, “In situ 3D nanoprinting of free-form coupling elements for hybrid photonic integration,” Nat. Photonics 12(4), 241–247 (2018). [CrossRef]
22. N. Kedia, Z. Liu, R. D. Sochol, J. Tam, D. X. Hammer, and A. Agrawal, “3-D printed photoreceptor phantoms for evaluating lateral resolution of adaptive optics imaging systems,” Opt. Lett. 44(7), 1825–1828 (2019). [CrossRef]
23. N. Kedia, Z. Liu, R. Sochol, D. X. Hammer, and A. Agrawal, “3D printed phantoms of retinal photoreceptor cells for evaluating adaptive optics imaging modalities,” in Proc. SPIE 10474, Ophthalmic Technologies XXVIII (SPIE, 2018), p. 104740D.
24. Q. Guo, R. Ghadiri, T. Weigel, A. Aumann, E. Gurevich, C. Esen, O. Medenbach, W. Cheng, B. Chichkov, and A. Ostendorf, “Comparison of in Situ and ex Situ Methods for Synthesis of Two-Photon Polymerization Polymer Nanocomposites,” Polymers 6(7), 2037–2050 (2014). [CrossRef]
25. A. Ostendorf, M. Chakif, and Q. Guo, “Laser direct writing of nanocompounds,” in MRS Proceedings (MRS, 2011), Vol. 1365, pp. mrss11-1365-tt01-03.
26. Q. Guo, R. Ghadiri, S. Xiao, C. Esen, O. Medenbach, and A. Ostendorf, “Laser direct writing of high refractive index polymer/TiO2 nanocomposites,” in Proc. SPIE 8243, Laser Applications in Microelectronic and Optoelectronic Manufacturing (LAMOM) XVII (SPIE, 2012), p. 824304.
27. H. Nishiyama, K. Umetsu, and K. Kimura, “Versatile direct laser writing of non-photosensitive materials using multi-photon reduction-based assembly of nanoparticles,” Sci. Rep. 9(1), 14310 (2019). [CrossRef]
28. A. C. Lamont, A. T. Alsharhan, and R. D. Sochol, “Geometric Determinants of In-Situ Direct Laser Writing,” Sci. Rep. 9(1), 394 (2019). [CrossRef]
29. A. T. Alsharhan, R. Acevedo, R. Warren, and R. D. Sochol, “3D microfluidics via cyclic olefin polymer-based in situ direct laser writing,” Lab Chip 19(17), 2799–2810 (2019). [CrossRef]
30. T. Bückmann, N. Stenger, M. Kadic, J. Kaschke, A. Frölich, T. Kennerknecht, C. Eberl, M. Thiel, and M. Wegener, “Tailored 3D Mechanical Metamaterials Made by Dip-in Direct-Laser-Writing Optical Lithography,” Adv. Mater. 24(20), 2710–2714 (2012). [CrossRef]
31. T. Gissibl, S. Wagner, J. Sykora, M. Schmid, and H. Giessen, “Refractive index measurements of photo-resists for three-dimensional direct laser writing,” Opt. Mater. Express 7(7), 2293–2298 (2017). [CrossRef]
32. M. S. Banks, A. B. Sekuler, and S. J. Anderson, “Peripheral spatial vision: limits imposed by optics, photoreceptors, and receptor pooling,” J. Opt. Soc. Am. A 8(11), 1775–1787 (1991). [CrossRef]
33. Z. Liu, O. P. Kocaoglu, T. L. Turner, and D. T. Miller, “Modal content of living human cone photoreceptors,” Biomed. Opt. Express 6(9), 3378–3404 (2015). [CrossRef]
34. T. W. Ridler and S. Calvard, “Picture Thresholding Using an Iterative Selection Method,” IEEE Trans. Syst. Man Cybern. 8(8), 630–632 (1978). [CrossRef]
35. Z. Liu, J. Tam, O. Saeedi, and D. X. Hammer, “Trans-retinal cellular imaging with multimodal adaptive optics,” Biomed. Opt. Express 9(9), 4246–4262 (2018). [CrossRef]
36. A. Ovsianikov, X. Shizhou, M. Farsari, M. Vamvakaki, C. Fotakis, and B. N. Chichkov, “Shrinkage of microstructures produced by two-photon polymerization of Zr-based hybrid photosensitive materials,” Opt. Express 17(4), 2143–2148 (2009). [CrossRef]
37. A. A. Bauhofer, S. Krödel, J. Rys, O. R. Bilal, A. Constantinescu, and C. Daraio, “Harnessing Photochemical Shrinkage in Direct Laser Writing for Shape Morphing of Polymer Sheets,” Adv. Mater. 29(42), 1703024 (2017). [CrossRef]
38. A. C. Lamont, M. A. Restaino, M. J. Kim, and R. D. Sochol, “A facile multi-material direct laser writing strategy,” Lab Chip 19(14), 2340–2345 (2019). [CrossRef]
39. F. Mayer, S. Richter, J. Westhauser, E. Blasco, C. Barner-Kowollik, and M. Wegener, “Multimaterial 3D laser microprinting using an integrated microfluidic system,” Sci. Adv. 5(2), eaau9160 (2019). [CrossRef]