1. Introduction

The examination of pediatric patients with complex retinal diseases presents unique challenges, often requiring anesthesia in the operating room (OR) due to limited cooperation. Additionally, early detection and management of retinopathy of prematurity (ROP), a potentially sight-threatening eye condition that primarily affects premature infants, requires ophthalmology care within the neonatal intensive care unit (NICU). While gold-standard methods like binocular indirect ophthalmoscopy and contact digital fundus photography [1] offer valuable enface retinal fundus images, they may overlook crucial details in the retina sub-surface. Combining OCT with fundus imaging in pediatric patients enhances diagnostic capabilities and improves the management of pediatric retinal diseases by providing detailed cross-sectional views of the retina. [2,3].

OCT has been effectively used for retinal disease screening in NICU and OR settings, using dedicated pediatric imaging systems like Envisu (Leica Microsystems, IL, USA) and iStand (Optovue Inc., CA, USA). These systems have revealed retinal abnormalities not visible in the standard fundus view, underscoring the significance of OCT in managing pediatric retinal disease [4–12]. The systems are non-contact, portable and rely on spectral domain technology (SD-OCT). The Envisu system features a non-contact hand-held scanner, while the iStand comprises a large scanner head mounted on an adjustable stand for imaging patients in a supine position. Both systems have moderate acquisition speeds (ranging from 26,000 to 32,000 A-scans/sec) and relatively bulky probes, leading to challenges in aligning the system with the patient’s retina and the possibility of motion artifacts when acquiring dense volumetric datasets over large fields of views. Recently, there have been significant advancements in the development of non-contact and lightweight pediatric retinal hand-held scanners for OCT imaging [13–23]. These advanced systems employ fast spectrometers or swept-source lasers, achieving scanning speeds ranging from 100,000 to 450,000 Alines per second. The lightweight design and high-speed capabilities of these systems have helped the alignment process with the patient’s eye, enabling fast acquisition of volumetric and angiography data over a wide field of view [17], all while mitigating the impact of motion artifacts. Nevertheless, several challenges persist even with these modern handheld designs.

Locating the region of interest in the retina remains difficult and time consuming due to the absence of patient fixation. Despite advancements in handheld OCT, operators often need to stabilize the subject manually, introducing motion artifacts. To address this challenge, especially in the context of OCTA where motion artifacts significantly impact image quality, contact lens-based approaches have facilitated image stabilization for angiograms acquisition in neonates [24,25].

Some systems have incorporated features such as an iris view (i.e. iStand system) [14], eye-tracking technology [26] or real-time retinal fundus display on a computer screen (i.e. Envisu system) or on a compact screen mounted on the handheld probe itself [17] to aid in precise OCT scan positioning. Nevertheless, these additions complicate system design and increase probe weight.

While OCT is effective in pediatric retinal imaging, it also lacks color information crucial for assessing various retinal pathologies [16,27]. Therefore, there remains a critical need for color fundus photography in conjunction with OCT. La Rocca et al. [16] created a scanning laser ophthalmoscope (SLO) for color fundus imaging and combined it with OCT in a non-contact handheld probe. However, the field of view of the combined SLO (<20 degrees) and OCT (< 10 degrees) was reduced compared to that of commercial and research handheld systems dedicated to fundus photography or OCT individually.

In this paper, we introduce a handheld retinal imaging concept integrating color digital fundus photography and OCT imaging in a contact fashion. By combining these modalities into a single handheld system, we aim to enable simultaneous OCT and color digital fundus imaging. The corneal contact approach, similar to commercial digital fundus cameras for pediatric patients [28] (e.g. RetCam, Clarity Medical System, Inc; ICON Phoenix Clinical, Inc.), reduces motion-related challenges during probe positioning. Furthermore, it offers real-time wide-field fundus display to assist photographers in guiding the OCT scan to the region of interest, ultimately reducing examination duration. This concept eliminates the need for separate, space-consuming systems dedicated to each imaging modality in the operating room and neonatal intensive care unit.

The study focuses on demonstrating the feasibility of the proposed approach, particularly for eventual use in pediatric patients and adults who may be uncooperative during traditional imaging procedures. To do so, we conducted experiments using custom-made eye models that mimic both pediatric and adult eyes. As a preliminary step toward pediatric patient imaging, we conducted in vivo experiments on a rabbit eye, chosen for its similar axial length to that of a newborn. Additionally, we evaluated the system’s ability to generate images on an adult human subject.

2. System for combined color wide-field fundus and OCT imaging

The schematic of the handheld probe is shown in Fig. 1. The probe comprises a front and rear piece. The front piece is adapted from a commercially available pediatric fundus photography system (RetCam, Clarity Medical System, Inc) and includes a multi-element ophthalmoscopy contact lens with a 120-degree field of view (B1200, Clarity medical systems Inc, CA, USA), followed by an electrically actuated adjustable refocusing objective lens for refractive error correction. The ophthalmoscopy lens incorporates a corneal contact element, which is coupled to the patient’s eye using optical coupling gel. Fundus illumination is achieved by directing an annular beam onto the eye's pupil through a ring illuminator positioned behind the contact lens. The ring illuminator consists of a bundle of circularly arranged fiber optics surrounding the optical elements of the ophthalmoscopy lens and connected externally to a Halogen light-source with adjustable intensity. The custom-built rear piece features a dichroic mirror (#64-434, Edmund Optics) that reflects the retinal image relayed by the adjustable refocusing lens to the sensor of a miniaturized level board digital color camera (5 MP, 1/2.5”, MU9PC-MBRD, Ximea Corp.). An infrared (IR) cut-off filter (86-095, Edmund Optics) is positioned in front of the camera sensor to attenuate reflections of the OCT beam into the fundus camera channel. The digital color camera was connected to a computer for real-time display of the fundus image. The OCT beam is delivered to the handheld probe with a single mode optical fiber and subsequently collimated with an aspheric lens (C560TMD-B, Thorlabs Inc. NJ, USA) resulting in a collimated beam with diameter of 2.6 mm (1/e²). After collimation, the OCT beam is directed toward the galvanometer scanner scanning the beam at a pupil conjugate plane. The OCT beam was then focused by an achromatic scanning lens positioned at a focal length distance from the pupil conjugate plane (AC127-019-B, Thorlabs Inc. NJ, USA). To enable OCT scanning of the retina through the ophthalmoscopy contact lens, the OCT scanner was aligned so that the conjugate image of the retina generated by the front piece coincides with the focal plane of the scanning lens, with the adjustable refocusing lens set to a refractive correction of 0 D. To cancel the astigmatism induced by the tilted dichroic mirror in the path of the convergent OCT beam, we introduced an identical dichroic mirror which is tilted in a meridian oriented 90 degrees to the first mirror (Fig. 2(D) – DM2). A CAD rendering of the probe is shown in Fig. 2. The overall weight of the probe was approximately 600 grams.

 

Fig. 1. Schematic of the handheld probe for combined OCT and fundus imaging consisting of an anterior and a posterior piece. Fundus imaging (A) and OCT beam (B) paths are shown. In the anterior piece, an ophthalmoscopy lens (OL) is in contact with the patient’s cornea. An adjustable refocusing lens (RF) corrects the refractive error. The ophthalmoscopy and the adjustable refocusing lens form an image of the retina into the rear piece. In the rear probe, a dichroic mirror (DM1) reflects the fundus image on a color CMOS sensor. A two-axis galvanometer scanner (GXY) scans the collimated OCT beam onto a scanning lens. The OCT beam is focused on a conjugate plane of the retina to enable scanning of the retina. A second dichroic mirror (DM2) cancels astigmatism introduced by the first mirror (DM1). An infrared cut-off filter (IR) reduces OCT reflections in the fundus imaging channel. Fundus illumination is achieved by a circular light guide connected to an external Halogen light-source.

Download Full Size | PDF

 

Fig. 2. CAD models of the integrated system. A) Optical components of the front piece with ophthalmic lens (OL) and the optical cable connected to the external halogen lamp (HLC) for fundus illumination. B) Optical components of the rear piece and custom mechanical mounts supporting the optical components and C) cover of the rear piece. D) Detailed view of the optical components of the rear piece including the dichroic mirrors (DM1 and DM2), the CMOS camera (S), the infrared filter (IR), the galvanometer scanner (GXY), the scanning lens (L), the collimator (C) and the fiber connector (FC). The OCT beam is shown (red shades). Photographs of the probe (E).

Download Full Size | PDF

For the present study, the handheld probe was connected to a custom-built research SD-OCT system employing a fiber-based Michelson interferometer using a super luminescent diode (SLD) with a center wavelength of 840 nm and a FWHM bandwidth of 50 nm (371-HP, Superlum diodes Ltd, Moscow, Russia). The axial resolution of the system is 5 µm in air. The reference arm consists of a collimator and a mirror and was adjusted so that the OCT image appeared on the screen as soon as the probe was placed in contact with the cornea. The OCT interference signal was processed with a research-grade spectrometer (Envisu R2310, Leica Microsystems) that enables acquisition at a speed of 32,000 A-lines per second over an axial range of 3.4 mm in air. Software was developed (LabView) to process and display the OCT and fundus images and to synchronize the operations of all systems, including the OCT spectrometer, the illuminator and the adjustable refocusing lens. The software enables simultaneous or interlaced acquisition of the OCT and fundus data. In the interlaced acquisition mode, the OCT beam is intentionally and temporarily displaced from the optical axis of the probe to further suppress unwanted reflections of the OCT beam into the fundus camera channel. Interlacing occurred after every OCT volume acquisition or 2D OCT scan acquisition, depending on the employed imaging mode. The imaging rate for fundus image capture was set at about 5 fps for both cases.

3. Experiments

A series of experiments were performed on custom-made model eyes to evaluate the optical performance of the handheld probe, including lateral resolution and field of view, and to assess the safety of the device in terms of maximum permissible light exposure. We then conducted in vivo experiments on rabbit and human adult eyes to evaluate the imaging capabilities and usability of the system.

3.1 Lateral resolution measurements

To evaluate the probe's on-axis lateral resolution at the retina, we employed two custom-made physical eye models — one for the adult eye and another for the pediatric eye. Both eye models feature a plano-convex lens positioned in front of a negative air-force resolution target (USAF Resolving Power Test Target 1951), simulating the optics of the eye and the retina, respectively. For the adult eye model, we used a plano-convex (PCX) lens with an 18 mm focal length (47-331, Edmund Optics Inc., NJ, USA), while for the pediatric eye model, we used a 12 mm focal length lens (47-329, Edmund Optics Inc., NJ, USA). These focal lengths were chosen to replicate the optical powers of the emmetropic adult (∼ 60D) and pediatric eye (∼80D) [29]. The contact lens of the probe was positioned as close as possible to the PCX lens in the eye models. The resolution target was mounted on a translation stage, allowing for axial displacement to replicate the spherical refractive error of the eye. A 7 mm diameter aperture, representing the eye's pupil, was positioned behind the PCX lens. The translation stage was adjusted incrementally to simulate myopic and hyperopic defocus shifts, encompassing a total of 46 refractive error steps ranging from -20 D to +25 D for both eye models. Each shift was calculated with the following formula:

3.2 Field of view measurements

We measured the Field of View (FOV) for both the OCT and digital fundus imaging systems as the angle α subtended by the image as viewed from the acquisition device (Fig. 3). The FOV was determined separately for both the vertical and horizontal axes for both imaging modalities. The setup resembled the one used for lateral resolution measurements, as we employed the same eye models used in section 3.2. To replicate the curvature of the retina, we fabricated a custom target with a radius of curvature of 12 mm for emulating the adult eye, and another with radius of curvature of 7.5 mm to mimic the pediatric eye [29]. The surface of the target was marked with millimeter-graph paper for precise FOV measurements (Fig. 4(B)). The fundus camera was configured to capture images in full format (1944 × 1944 pixels). For the OCT, we employed a raster scanning pattern consisting of 1000 × 1000 A-lines. Figure 4(B) reports representative volume intensity projections (VIPs) and fundus photographs obtained from the adult and pediatric eye models, respectively, under a refractive error of 0D. The translation stage was incrementally adjusted to replicate myopic and hyperopic defocus shifts, encompassing a total of 10 refractive error steps spanning from -20 D to +25 D for both eye models.

 

Fig. 3. (A) Schematic of the eye used for calculating the field of view (α). r_eye is the radius of the retina, f the focal of the eye, x the displacement of the target to the simulate a refractive error, l the linear length of the arch on the target. (B) Photographs of the custom-made targets for FOV measurements for the adult (top) and the pediatric (bottom) eye. CAD drawings of the eye models including the PCX lens and iris are reported. OCT volume intensity projections and fundus photographs of the emmetropic adult and pediatric eye models configured for measurements along the horizontal axis (right). FOV is measured along the marked lines (red).

Download Full Size | PDF

Field-of-view measurements were then conducted for both the horizontal and vertical axes by rotating the curved target 90 degrees relative to the optical axis of the system. All images were processed to measure the arc length (l) in mm (Fig. 3) in the OCT VIP images and fundus photographs under all refractive error conditions and for both eye models. The field of view (α) in degrees corresponding to each measured arc length (l) was then calculate with the procedure described in Appendix A.

3.3 In vivo experiments on the rabbit’s eye

As an initial step to assess the feasibility of pediatric patient imaging with the current probe, we performed experiments on a rabbit, chosen for its similarity in eye dimensions to that of a newborn human subject, with an axial eye length of approximately 18 mm [30]. The use of our experimental setup for in vivo measurements in rabbits was approved by the local Institutional Animal Care and Use Committee. An ophthalmologist conducted the rabbit’s retina examination with the handheld probe while simultaneously using the real time OCT and color fundus image as feedback to position the probe. Following anesthesia with a cocktail of xylazine, acepromazine and ketamine adjusted according to the rabbit weight, the pupil was dilated with tropicamide 1%. The operator gently applied the probe in contact with the rabbit cornea using a lubricant eye gel (Goniotaire, Altaire Inc, NY), a technique similar to that employed in pediatric fundus imaging. The examination included the combined acquisition of fundus images and OCT linear or volumetric scans, along with video recordings from both imaging modalities. Prior and following each experiment, the contact lens was sterilized with a solution based on sodium hypochlorite to avoid contamination and rinsed with sterile water.

3.4 In vivo experiments on an adult human subject

Experiments on an adult human subject were approved by the local Institutional Review Board and adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from a 26-year-old human subject. The subject had a refractive error (mean spherical equivalent) of -0.75 D but no known ocular pathology. The subject was imaged in supine position after pupil was dilated with tropicamide 1%. A topical anesthetic (tetracaine 0.05%) was instilled immediately before the imaging procedure to reduce the discomfort produced by the contact of the probe with the cornea. During the procedure, the operator applied eye gel to the cornea and conducted retinal imaging with a procedure similar to the one described in section 3.3. To minimize discomfort to the subject caused by the contact probe during the imaging procedure, a faster 2D scan was chosen over a longer 3D image acquisition.

3.5 Light safety

Before testing the device in vivo, we conducted a light safety evaluation to verify that the combined light irradiance generated by the near-infrared OCT and visible fundus illumination adheres to the most recent ISO standard 15004-2 Ophthalmic instruments — Fundamental requirements and test methods — Part 2: Light hazard protection– [31]. We assume the most conservative case scenario of continuous irradiation at a fixed position. For the spectral ranges (Fig. 4) corresponding to the SLD and fundus illumination, we assessed three key parameters:

 

Fig. 4. Normalized (unitary area under the curve) spectrum of the (A) fundus light-source (Halogen lamp) and (B) OCT light-source (SLD).

Download Full Size | PDF

Table 1. Compliance of the system illumination with the ISO standard

View Table

4. Results

4.1 Optical performance

Figure 5 shows the lateral resolution, measured in line pairs per millimeter, plotted against simulated refractive error in Diopters for both the adult (Fig. 5(A)) and pediatric (Fig. 5(B)) eye models. The data points are discrete due to measurements using an air-force target. In the adult eye model, the resolution of the OCT and fundus system increased from 10 lp/mm at -20 D to 22.6 lp/mm at +25 D and from 12.7 lp/mm at -20 D to 28.5 lp/mm at +25 D, respectively. In the pediatric eye model, the resolution of the OCT and fundus system increased from 18lp/mm at -20 D to 32 lp/mm at +25 D and from 22.6 lp/mm at -20D to 51.8 lp/mm at +25 D, respectively. The blue and orange solid lines represent linear fits for the OCT and fundus camera data, respectively. The fundus images acquired with the digital camera consistently exhibit superior lateral resolution compared to the OCT for both the pediatric and adult models, with resolution improving as the spherical refractive errors increase. The increase in resolution with refractive error is attributable to the greater image magnification, which occurs because the object subtends a wider angle as the refractive error increases or the distance between the lens and the object decreases. Overall, the pediatric eye model, characterized by a higher optical power, demonstrates greater resolving power compared to the adult eye model.

 

Fig. 5. On axis lateral resolution in line pairs per mm as a function of the simulated refractive error in Diopters for the adult (A) and the pediatric (B) eye models. The linear fit of the data samples and R² coefficient are displayed for the OCT (blue)_and digital fundus camera (orange). Example of the images of a negative high-resolution target (USAF 1951) taken with the OCT and the fundus imaging systems for the adult (C) and the pediatric (D) eye. Each row shows fundus images taken with OCT and the digital camera for three refractive errors corresponding to sphere values of -20, 0 and +25 Diopters. The group and element of the resolution target was reported for the three refractive errors. The target was shifted laterally between imaging sessions to prevent reflections caused by the ring illuminator on the surface of the plano-convex lens and to align it to the optical axis of the handheld unit. As a result, certain images captured with the digital camera and OCT may appear with different brightness.

Download Full Size | PDF

Figure 6 shows the plots of the FOV for both the OCT and the fundus camera using the adult and pediatric eye models from the horizontal and vertical measurements. For the fundus camera, the FOV remains relatively consistent across different refractive errors, with mean values of 87 and 75 degrees for the pediatric and the adult eye model, respectively. For the OCT system, the FOV closely aligns with that of the fundus camera for positive refractive errors, yielding mean values of 88 degrees for the pediatric eye model and 74 degrees for the adult eye model. However, for myopic shifts below approximately -5 D (as indicated in Fig. 6 – shaded area), the curvature of the eye models extends beyond the axial range of the OCT system, thereby limiting the FOV (as demonstrated in Fig. 7). With an extended OCT axial range, the FOV will remain relatively consistent over the entire myopic range, similarly to the FOV observed with the fundus camera (as shown in Fig. 7 for Sph = -20D).

 

Fig. 6. Field of view measurements (degrees) of the OCT (blue) and fundus camera (orange) as a function of the simulated refractive error (D) for the adult (A) and pediatric (B) eye models. Each dot represents the average FOV calculated along the horizontal and vertical directions. The shade area represent the myopic region below approximately -5D where the FOV of the OCT system is limited by the axial range.

Download Full Size | PDF

 

Fig. 7. OCT images obtained from the adult eye model (left) and pediatric eye model (right) for refractive errors in spherical values of -20, 0, and +25 D. As the sphere value increases, the target representing the retina adopts a flatter appearance. At a high myopic shift (-5 to -20 D), the retinal target exhibits a steep curvature that surpasses the depth range of the OCT system, thereby constraining the system’s field of view. Consequently, the image of the target folds back into the OCT mirror image space (red arrows).

Download Full Size | PDF

4.2 In vivo imaging with the handheld device

Figure 8(A) and 8(B) show a fundus image and an OCT volume intensity projection generated from a volumetric dataset, respectively. The volumetric OCT data consists of 1,000 frames of 1,000 A-lines each, acquired in approximately 32 seconds over a field of view of about 85 degrees. The fundus image covers an approximate field of view of 87 degrees and was acquired simultaneously with the acquisition of the volumetric OCT data. Key features as retinal layers in the OCT image, and the optic disc and vasculatures in the color fundus image can be distinctly observed. Two-dimensional OCT images along the fast (Fig. 8(C)) and slow (Fig. 8(D)) axes are shown. Despite the IR filter placed in front of the camera effectively blocking the scanning OCT scan from appearing on the fundus image, residual reflections and glare generated by the OCT beam at the optical elements of the probe are visible in the fundus when acquired simultaneously with OCT. To address this issue, interlaced acquisition was used (Fig. 8(E – I)). Figures 8(G) and 8(I) show the fundus and 2D OCT images acquired across the optic disc of the same rabbit’s eye, respectively. Figure 8(J) shows a photograph of the probe in contact with the rabbit eye during an imaging session.

 

Fig. 8. Fundus (A) and OCT VIP image (B) simultaneously acquired on the rabbit’s eye (see Visualization 2 and Visualization 3). Residual reflections and glare generated by the OCT beam at the optical interfaces of the probe are visible in the fundus image (A – white arrows). 2D OCT images extracted along the fast (C) and slow axis (D) of the volumetric OCT dataset. Image artifacts due to residual axial motion during imaging are visible along the slow axis (D). Fundus image (E), 2D OCT (F) and 3D OCT (G) (see Visualization 4) recorded with interlaced acquisition across the optic disc. No reflections or glare are visible in the fundus image during interlaced acquisition. Additional fundus and 2D OCT image acquired away from the optic disc using interlaced acquisition (H, I). Photograph and video (Visualization 1) of the imaging probe in contact with the rabbit eye (J).

Download Full Size | PDF

Throughout the in vivo imaging procedure, the operator stated that the hand-held probe was sufficiently light to enable maneuverability to complete fundus image and volumetric OCT acquisitions. A movie displaying the imaging procedure is reported (see Visualization 1). Additionally, the real-time fundus image served as valuable feedback for precise probe positioning and for locating the OCT in various regions of interest. To show the real-time capabilities of the imaging system and the operator’s ability in maintaining probe stability during extended imaging session, we acquired simultaneous sequences of fundus and 2D OCT images (comprising 1,200 A-lines with an 80 A-line flyback) on the rabbit’s eye for approximately 20 seconds. These sequences, available in Visualization 1 and Visualization 2, show the ability of the operator to maintain the OCT image approximately at the same imaging depth. While the contact of the probe with the eye ensured stability during imaging, residual axial motions become perceptible when recording large volumetric OCT datasets (acquired over 32 seconds and consisting of 1,000,000 A-lines) (Fig. 8(D).

Figures 9(A) and 9(B) show a fundus and a 2D OCT image acquired on the human subject. The OCT scan had a density of 1,000 A-lines and was acquired over a field of view of approximately 40 degrees. The fundus image was acquired with full resolution over a field of view of approximately 75 degrees. The fovea and the optics disc are visible in both fundus and OCT image as well as the retinal layers in the OCT image. The experiment demonstrates the ability of the system to acquire fundus and 2D OCT images on human subjects.

 

Fig. 9. Fundus (A) and 2D OCT image (B) acquired on the right eye of a 26-year-old subject. The location of the OCT image on the fundus is indicated (dashed line). The fundus image was acquired automatically and immediately after the OCT scan was completed. 3D OCT scan is reported (see Visualization 5)

Download Full Size | PDF

5. Discussion

The combination of OCT and fundus photography, as complementary imaging modalities, holds significant potential for improving pediatric retinal examination. Currently, there is no commercial or research handheld system that combines high-quality wide-field fundus photography and OCT imaging for pediatric retinal examination in the operating room or neonatal intensive care unit. Our proof of concept establishes the foundation for a multimodal imaging solution that addresses key challenges in the development of a technique for routine diagnosis and management of pediatric retinal diseases. By integrating wide-field fundus photography and OCT into a single system, we replace the need for two separate devices solely dedicated to fundus photography and OCT, streamlining clinical space usage. Simultaneous image acquisition of fundus photography and OCT scans also reduces examination time. Currently, both modalities are acquired independently, which elongates the diagnostic process. Simultaneous acquisition also ensures precise alignment and registration between the OCT scan and fundus image. The real-fundus time also provides an intuitive guide for photographers to accurately position the OCT scan over the region of interest enhancing diagnostic capabilities. The probe's contact with the cornea, similar to existing commercial handheld camera probes, enhances probe stability during imaging. This approach minimizes motion-related challenges associated with handheld OCT probes. Typically, OCT systems prioritize non-contact imaging to prevent patient discomfort, contamination, and the potential risk of probe-induced eye damage. However, introducing contact OCT imaging for pediatric patients does not pose additional risks with the proposed approach because pediatric fundus photography inherently involves eye contact, and both examinations are performed concurrently. Ultimately, the current technology offers a platform for both OCTA and Fluorescein Angiography (FA), which is the gold standard for retinal vascular visualization. This dual approach might leverage the unique strength of each technique.

Our handheld prototype combines the optics of a contact digital fundus camera with custom-made optics for OCT scanning and fundus image acquisition. A key challenge from using a contact lens designed for fundus imaging with OCT is its optimization to enhance imaging in the visible rather than near-infrared spectral range. We identified an OCT beam geometry that maximizes lateral resolution (on-axis) across a wide range of refractive errors (i.e. -25 D to +20 D) for both adult and pediatric eye using the current contact lens. Nevertheless, the lateral resolution of the OCT system still averaged 4 and 10 lines per mm lower than that of the fundus camera for the adult and pediatric eye, respectively. Future evaluations should consider off-axis lateral resolution to assess retinal periphery imaging.

Another drawback associated with employing this contact lens is the suboptimal optical coating for minimizing OCT beam reflections. Despite the use of an infrared blocking filter, unwanted reflections and glare persisted in the fundus image when acquired simultaneously with OCT. To address this issue, we implemented an interlaced acquisition approach for both modalities, which effectively enhance fundus image clarity without significantly extending the acquisition time.

Additionally, the suboptimal coating of the optical elements within the contact lens limited the maximum OCT power exiting the probe to approximately 0.5 mW, corresponding to a loss in sensitivity of 3.9 dB. Based on our calculations following the ISO standard 15004-2 (Table 1), the optical power at the retina could be increased to improve the signal-to-noise ratio of the OCT image, possibly allowing an increase in imaging speed. This improvement can be achieved by using a higher-powered SLD and/or by optimizing the coating of the contact lens for the NIR spectral range.

These challenges emphasize the importance of optimizing the contact lens’s optical design for both OCT and color fundus imaging modalities.

While theoretically feasible, our system encountered challenges in providing real-time guidance by visualizing the OCT beam on the fundus image. Achieving this visualization requires precise synchronization between the two acquisition processes, which was not implemented in our system. Retro-reflections also impacted the clarity of the scanning OCT beam on the fundus image. As previously mentioned, this issue can be addressed by designing a lens with NIR coating. Alternatively, employing a fiducial marker on the fundus image to indicate the location and area of the OCT scan could circumvent synchronization complexities.

Our results indicate that the field of view remains consistent between the OCT and fundus imaging system (∼ 87 degrees for the pediatric eye and 75 degrees for the adult eye degrees) over a wide range of refractive errors (-20 D to +25 D). Although the field of view was measured over a wider range of refractive errors, in practical scenarios, refractive errors typically range from -10D to +5D, encompassing the imaging needs of more than 95% of subjects. The field of view of the OCT system was primarily constrained in the highly myopic region due to its limited axial range. This limitation could potentially be addressed by employing an OCT system with an extended axial range. The field of view was assessed as the angle subtended by the image as viewed from the acquisition device (i.e., visual angle). Commercial wide-field systems typically specify the field of view as the angle subtended by the image region at the spherical center of the eye (i.e., eye angle) [32]. This distinction might in part contribute to the difference between the field of view measured in our study (approximately 87 degrees) and the one specified for the contact lens (120 degrees).

In our experimental setups, we estimated the optical performance without using coupling gel between the probe and eye models. While gel is used in actual imaging scenarios to create contact and protect the cornea via a thin layer, our experiments encountered challenges with gel dispersion and control while using the eye model, rendering its implementation impractical. Our physical models rely on PCX lenses that do not directly replicate the corneal curvature, resulting in a gap between the contact lens and eye model reaching a maximum of 0.2 mm and 0.4 mm along the optical axis of the pediatric eye and adult eye, respectively. This gap prevented uniform gel filling. Optical simulations (Zemax) were conducted to assess the potential impact of gel on resolution and field of view measurements on the eye model. These simulations indicated negligible changes in spot size on the retina with refocusing but revealed underestimation of the field of view in both pediatric and adult eye models without gel (approximately 4% for the pediatric eye and 8% for the adult eye).

As an initial phase in preparing the system for pediatric patients, we performed in vivo imaging experiments on both an adult subject’s eye and a sedated rabbit’s eye. The rabbit’s eye was specifically chosen for its comparable axial length to that of a newborn. In the experiment involving the adult subject, we deliberately restricted the OCT field of view to 40° and acquired only 32 B-scans. This imaging protocol aimed to minimize acquisition time, thus reducing discomfort for the patient despite the administration of topical anesthesia. We anticipate that sedated patients, the primary target of the system, will tolerate the procedure well. Conversely, imaging of the rabbit under full sedation allowed for the acquisition of complete datasets without such concerns. The stability of the probe-eye contact facilitated contact between the probe and the eye ensured stability during extended imaging sessions (approximately 20-30 seconds) during the experiment on the rabbit, enabling acquisition of long 2D OCT sequences and large volumetric OCT datasets (approximately 1,000,000 A-lines). However, some residual axial motion artefacts were still observed. These residual motion artifacts might be further mitigated by implementing a faster OCT system based on a spectrometer or swept-source operating at A-line rates of 100 kHz or higher.

Although we have demonstrated a proof-of-concept system that can generate images on the rabbit and the adult human eye in vivo, further testing on additional subjects and pediatric patients remains a critical step in its development.

Appendix A: Conversion arc length to field of view

Figure 10 shows the schematic used for converting arc length to field of view in the eye models.

 

Fig. 10. Schematic used for the conversion from arc length to field of view in the eye model.

Download Full Size | PDF

To determine the semi-field of view α/2 in the eye model with given radius of curvature of the retina r_eye, focal length f, shift of the retina x, for a measured semi arc length l/2, the following expression is employed:

(2)$$\begin{array}{{c}} {\sin \frac{\alpha }{2} = \frac{y}{m}\; = \; \frac{y}{{\sqrt {{y^2} + {n^2}} }}} \end{array},$$

where y is the semi chord subtending the semi arc l/2, and m can be represented as $\sqrt {{y^2} + {n^2}} $, where segment n is equal to $f + x - \; ({{r_{eye}} - {r_{eye}}\cos \beta } )$ (Fig. 10). By squaring Eq. (2), substituting m and n with their respective expressions, and considering $y = {r_{eye}}\sin \beta $, we derive:

(3)$$\begin{array}{{c}} {{{\left( {\sin \frac{\alpha }{2}} \right)}^2} = \frac{{{{({{r_{eye}}\; \sin \beta } )}^2}}}{{{{({{r_{eye}}\; \sin \beta } )}^2} + {{({f + x - \; ({{r_{eye}} - {r_{eye}}\cos \beta } )} )}^2}}}} \end{array}$$

The angle β (in radians) can then be computed as the ratio between the semi arc length and the radius of the eye, resulting in the following expression for the entire field of view $\alpha $ (in degrees):

(4)$$\begin{array}{{c}} {\alpha = \frac{{360}}{\pi }{\; \textrm{si}}{\textrm{n}^{ - 1}}\sqrt {\frac{{{{\left[ {{r_{eye}}\; \sin \left( {\frac{l}{{2{r_{eye}}}}} \right)} \right]}^2}}}{{{{\left[ {{r_{eye}}\; \sin \left( {\frac{l}{{2{r_{eye}}}}} \right)} \right]}^2} + {{\left[ {f + x - {r_{eye}} + \; {r_{eye}}\; \cos \left( {\frac{l}{{2{r_{eye}}}}} \right)} \right]}^2}}}} \; \; \; } \end{array}$$

Funding

Stanley J. Glaser Foundation at University of Miami (UM SJG2015-15); National Institutes of Health (Center Core Grant P30EY014801); Florida Lions Eye Bank; Beauty of Sight Foundation; Henri and Flore Lesieur Foundation; Drs. Harry W. Flynn Jr MD, Raksha Urs, PhD and Aaron Furtado; Karl R. Olsen, MD and Martha E. Hildebrandt, PhD; Research to Prevent Blindness (Unrestricted Grant (GR004596-1)).

Acknowledgements

The authors thank Dr. Eric Buckland for material support in the probe development and Dr. Esdras Arrieta Quintero for assistance with the rabbit experiments.

Disclosures

The University of Miami and some of the authors (MR, FM, JMP) stand to benefit from intellectual property in the technology used in this study.

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.

References

1. W. M. Fierson, “Screening examination of premature infants for retinopathy of prematurity,” Pediatrics 142(6), e20183061 (2018). [CrossRef]  

2. P. Hahn, J. Migacz, R. O’Connell, et al., “The use of optical coherence tomography in intraoperative ophthalmic imaging,” Ophthalmic Surg. Lasers Imaging 42(4), 85–94 (2011). [CrossRef]  

3. A. Mallipatna, A. Vinekar, C. Jayadev, et al., “The use of handheld spectral domain optical coherence tomography in pediatric ophthalmology practice: Our experience of 975 infants and children,” Indian J. Ophthalmol. 63(7), 586–593 (2015). [CrossRef]  

4. R.H. Muni, R.P. Kohly, E.H. Sohn, et al., “Hand-held spectral domain optical coherence tomography finding in shaken-baby syndrome,” Retina 30(4), S45–S50 (2010). [CrossRef]  

5. A.W. Scott, S. Farsiu, L.B. Enyedi, et al., “Imaging the infant retina with a hand-held spectral-domain optical coherence tomography device,” Am. J. Ophthalmol. 147(2), 364–373.e2 (2009). [CrossRef]  

6. R.H. Muni, R.P. Kohly, A.C. Charonis, et al., “Retinoschisis detected with handheld spectral-domain optical coherence tomography in neonates with advanced retinopathy of prematurity,” Arch. Ophthalmol. 128(1), 57–62 (2010). [CrossRef]  

7. R.S. Maldonado, J.A. Izatt, N. Sarin, et al., “Optimizing hand-held spectral domain optical coherence tomography imaging for neonates, infants, and children,” Invest. Ophthalmol. Visual Sci. 51(5), 2678–2685 (2010). [CrossRef]  

8. S.H. Chavala, S. Farsiu, R. Maldonado, et al., “Insights into advanced retinopathy of prematurity using handheld spectral domain optical coherence tomography imaging,” Ophthalmology 116(12), 2448–2456 (2009). [CrossRef]  

9. G.T. Chong, S. Farsiu, S.F. Freedman, et al., “Abnormal foveal morphology in ocular albinism imaged with spectral-domain optical coherence tomography,” Arch. Ophthalmol. 127(1), 37–44 (2009). [CrossRef]  

10. D.B. Rootman, E. Gonzalez, A. Mallipatna, et al., “Hand-held high-resolution spectral domain optical coherence tomography in retinoblastoma: clinical and morphologic considerations,” Br. J. Ophthalmol. 97(1), 59–65 (2013). [CrossRef]  

11. H.A. Aziz, M. Ruggeri, and A.M. Berrocal, “Intraoperative OCT of bilateral macular coloboma in a child with Down syndrome,” J. Pediatr. Ophthalmol. Strabismus. 48(6), 37–39 (2011). [CrossRef]  

12. S.S. Ong, T.J. Cummings, L. Vajzovic, et al., “Comparison of optical coherence tomography with fundus photographs, fluorescein angiography, and histopathologic analysis in assessing coats disease,” JAMA Ophthalmol. 137(2), 176–183 (2019). [CrossRef]  

13. F. LaRocca, D. Nankivil, T. DuBose, et al., “Ultra-compact switchable SLO/OCT handheld probe design,” Proc. SPIE 9307, 93071F (2015). [CrossRef]  

14. C.D. Lu, M.F. Kraus, B. Potsaid, et al., “Handheld ultrahigh speed swept source optical coherence tomography instrument using a MEMS scanning mirror,” Biomed. Opt. Express 5(1), 293–311 (2014). [CrossRef]  

15. D. Nankivil, G. Waterman, F. LaRocca, et al., “Handheld, rapidly switchable, anterior/posterior segment swept source optical coherence tomography probe,” Biomed. Opt. Express 6(11), 4516–4528 (2015). [CrossRef]  

16. F. LaRocca, D. Nankivil, S. Farsiu, et al., “True color scanning laser ophthalmoscopy and optical coherence tomography handheld probe,” Biomed. Opt. Express 5(9), 3204–3216 (2014). [CrossRef]  

17. S. Song, K. Zhou, J.J. Xu, et al., “Development of a clinical prototype of a miniature hand-held optical coherence tomography probe for prematurity and pediatric ophthalmic imaging,” Biomed. Opt. Express 10(5), 2383–2398 (2019). [CrossRef]  

18. W. Jung, J. Kim, M. Jeon, et al., “Handheld optical coherence tomography scanner for primary care diagnostics,” IEEE Trans. Biomed. Eng. 58(3), 741–744 (2011). [CrossRef]  

19. J.D. Malone, M.T. El-Haddad, S.S. Yerramreddy, et al., “Handheld spectrally encoded coherence tomography and reflectometry for motion-corrected ophthalmic optical coherence tomography and optical coherence tomography angiography,” Neurophotonics 6(04), 1 (2019). [CrossRef]  

20. S. Ni, X. Wei, R. Ng, et al., “High-speed and widefield handheld swept-source OCT angiography with a VCSEL light source,” Biomed. Opt. Express 12(6), 3553–3570 (2021). [CrossRef]  

21. C. Viehland, X. Chen, D. Tran-Viet, et al., “Ergonomic handheld OCT angiography probe optimized for pediatric and supine imaging,” Biomed. Opt. Express 10(5), 2623–2638 (2019). [CrossRef]  

22. J. Yang, L. Liu, J.P. Campbell, et al., “Handheld optical coherence tomography angiography,” Biomed. Opt. Express 8(4), 2287–2300 (2017). [CrossRef]  

23. O. Nadiarnykh, V. Davidoiu, M.G.O. Gräfe, et al., “Phase-based OCT angiography in diagnostic imaging of pediatric retinoblastoma patients: abnormal blood vessels in post-treatment regression patterns,” Biomed. Opt. Express 10(5), 2213–2226 (2019). [CrossRef]  

24. J.P. Campbell, E. Nudleman, J. Yang, et al., “Handheld optical coherence tomography angiography and ultra-wide-field optical coherence tomography in retinopathy of prematurity,” JAMA Ophthalmol. 135(9), 977–981 (2017). [CrossRef]  

25. S.J. Kim, J. Yang, G. Liu, et al., “Optical coherence tomography angiography and ultra-widefield optical coherence tomography in a child with incontinentia pigmenti,” Ophthalmic Surg. Lasers Imaging Retina 49(4), 273–275 (2018). [CrossRef]  

26. X. Liu, A.U. Kale, N. Capewell, et al., “Optical coherence tomography (OCT) in unconscious and systemically unwell patients using a mobile OCT device: a pilot study,” BMJ Open 9(11), e030882 (2019). [CrossRef]  

27. S.P. Hibbs, A. Smith, L.P. Chow, et al., “Colour photographs for screening in neovascular age-related macular degeneration: are they necessary?” Eye 25(7), 918–921 (2011). [CrossRef]  

28. C. Wu, R.A. Petersen, and D.K. VanderVeen, “RetCam imaging for retinopathy of prematurity screening,” J. AAPOS 10(2), 107–111 (2006). [CrossRef]  

29. R.C. Augusteyn, D. Nankivil, A. Mohamed, et al., “Human ocular biometry,” Exp. Eye Res. 102, 70–75 (2012). [CrossRef]  

30. A. Hughes, “A schematic eye for the rabbit,” Vision Res. 12(1), 123–IN6 (1972). [CrossRef]  

31. “Ophthalmic instruments — Fundamental requirements and test methods —Part 2: Light hazard protection” ISO15004-2:2007.

32. X. Yao, D. Toslak, T. Son, et al., “Understanding the relationship between visual-angle and eye-angle for reliable determination of the field-of-view in ultra-wide field fundus photography,” Biomed. Opt. Express 12(10), 6651–6659 (2021). [CrossRef]