Far-Red, High-Resolution, Reflection-Free Images of the Anterior Segment in Retro-Illumination

Abstract

Purpose: Examination of the anterior segment using retro-illumination is part of the routine clinical examination, but capturing images that can be used by the clinician remains a challenge because reflections are always present, resolution is insufficient, and the patient’s glare often causes motion blur. Our aim was therefore to overcome these limitations.

Methods: We modified a slit lamp by using a far-red LED, a high-resolution camera, and by modifying the light path using polarizers and an obturator to suppress reflections while optimizing the light flux. We used this prototype to image various ocular conditions to illustrate its potential.

Results: With the ×40 magnification of the slit lamp, the prototype provided images with a resolution of 6.2 µm, and a field of view of 9.7 mm. We obtained images without motion blur in all patients. The vast majority of images were reflection-free. Only the images of a few patients with pseudophakia showed persistent small reflection. The resolution was sufficient to highlight, for example, all the Descemetic excrescences (Guttae) constitutive of a Fuchs endothelial corneal dystrophy (FECD).

Conclusions: Our retro-illumination device prototype provides high-resolution images of the different structures of the anterior segment of the eye, easy to acquire mainly because they do not dazzle patients. We believe that these images have the potential to facilitate the diagnosis and monitoring of many different diseases both in routine use and in clinical trials where they could be used as objective endpoints.

Translational Relevance: Our prototype revisits retro-illumination with a potential to become a new imaging device.

Introduction

In optical imaging of the anterior segment of the eye, several modalities are commonly used to obtain detailed images of ocular structures. Although techniques such as optical coherence tomography (OCT), confocal microscopy, which involves direct contact with the eye, and specular microscopy have their advantages, they also come with limitations. For instance, OCT is associated with high acquisition costs, whereas confocal microscopy and specular microscopy restrict the field of view. In this context, slit-lamp biomicroscopy offers a significant advantage by allowing complete visualization of the anterior segment of the eye at a low cost, without requiring direct contact with the cornea. 

Retro-illumination observation of the anterior segment of the eye is a routine examination method as old as the slit lamp. After pharmacological dilation of the pupil, a beam of white light usually slightly smaller than the diameter of the pupil is used to illuminate the back of the eye. The practitioner then tilts the slit until the red reflex generated by the pigmented epithelium of the retina is obtained and observes the various structures of the anterior segment directly. This mode of illumination is particularly suitable for observing details of the lens, iris, and cornea and complements the other methods of illumination available on the slit lamp. Nevertheless, capturing good quality images is very challenging because of patient glare and eye movements, the presence of inevitable reflections in the cornea and/or in the crystalline lens, and limited resolution, even with last generation of digital slit lamps (Fig. 1). 

Figure 1.

 

Representative examples of a backlight image using a modern standard photographic slit lamp. Resolution is limited, the white light inevitably dazzles the patient, and the reflections of the light source on the various ocular structures mask certain parts of the image.

Currently, when it is necessary to analyze the entire surface of a reflection-free image from a slit lamp, the practitioner must capture several images, select two with very different reflections (one on the left and one on the right), and assemble the reflection-free parts.^1–4 However, image processing, in addition to being time-consuming, may alter the reality, introducing bias, especially because of the unavoidable eye movements between image captures. These methods are less suitable for conditions requiring the observation of small details and the analysis of their spatial distribution across the whole observation surface. In current practice, because of all these difficulties, digital retro-illuminated images captured with common slit lamps are not used, although they could provide a lot of information to clinicians. They are used only for research purposes, mainly in two domains: Fuchs endothelial corneal dystrophy (FECD)⁴ and posterior capsule opacification⁵ but with all the disadvantages mentioned above. FECD is the most frequent disease of the most posterior layer of the cornea, called the endothelium. It affects 4% to 12% of adults in the Occident and is the first indication for corneal transplantation.⁶ Posterior capsule opacification is a common evolution after cataract surgery (which may affect 5% to 20% of the 25 millions of eyes operated worldwide yearly)⁷ and correspond to the growth of residual crystalline lens cells between the intra ocular lens and the natural lens bag. According to the opacification extent, it may require laser treatment to free the optical axis. 

Some authors have therefore already proposed technical solutions to improve these clinical images. Pande et al. 26 years ago developed a digital coaxial retro-illumination photography system based on a modified slit lamp.⁸ It provided high resolution images but without succeeding in completely eliminating the central Purkinje reflection and without widespread adoption by practitioners. More recently, Weber et al. have developed two retro-illumination imaging modalities. The first is asymmetric retro-illumination microscopy developed for the anterior segment.⁹ It consists in capturing the inside of an unilluminated area to avoid reflections. However, this method comes with a limited field, which is incompatible with the need to capture the whole surface of observation. The second uses transcranial illumination: the patient’s temple is illuminated by a near-infrared light source that penetrates the skin, bones, and muscles, and scatters to the retina, which reflects the light and becomes the sole source of illumination of the imaging system.¹⁰ Although it shows promise for achieving reflect-free corneal imaging, transcranial illumination faces limitations due to high light intensity requirements on the skin. This restricts its application, given regulatory constraints and the need for expensive equipment to compensate for its insufficient light intensity for imaging. In addition, this device has not been used to image the anterior segment of the eye. 

Materials and Methods

Principles of the Slit Lamp Modification

We developed a system to remove undesired reflections in retro-illumination by modifying a slit lamp (SL990, CSO Scandicci, Italy; Fig. 2). This adjustment applied to both the imaging and illumination channels, retaining the standard mechanical functionalities, including independent rotations of the observation and illumination pathways, along with focus and mechanical centering of the observation plane. This ensured ease of use for ophthalmologists. 

Figure 2.

 

Modified slit-lamp and its optical scheme.

Lighting Path

The lighting system (Fig. 3) of this device used a non-polarized 780 nm LED. Its beam was collimated by a parabolic mirror and then linearly polarized. The focus lens optically conjugated the diaphragm and obturator plane with the observation plane, resulting in a ring-shaped illumination. This design effectively avoided illuminating the central zone of the cornea, where specular reflections were directed toward the imaging system. Moreover, the focus lens ensured efficient light concentration on a small planar mirror, providing divergent illumination for the patient. 

Figure 3.

 

Lighting path modelization.

The removal of reflections was accomplished through various steps of light manipulation, involving both spatial and polarization state adjustments (Fig. 4). As the light reached the cornea, it split into two types of beams. One beam passed through the eye to the retina, where it was partially backscattered and absorbed. The backscattered light, depolarized, served as the primary illumination for the ophthalmologist's observation area (retro-illumination). The second more intense beam was reflected by the cornea, generating specular reflections with the same initial polarization. The light passing through the imaging path combined these two types of beams. This light was filtered by an analyzer, which blocked the initial polarization direction, allowing only the corresponding polarization of the backscattered light to pass through. 

Figure 4.

 

Principle of light source modulation.

Imaging Path

The imaging path (Fig. 5) relied on an optical module from the SL990 slit lamp model. This component allowed for various magnification changes, particularly with a wheel composed of multiple Galilean doublets projecting the image of the observation plane to infinity, focused by an objective on a sensor. A camera with a sensor (12.4 × 8.8 mm size) and a 50 mm focal length lens was positioned behind this module to ensure an 8 mm diameter field of view on the sensor, using the ×40 position of the slit lamp magnification changer. A polarizer on an adjustable mount was aligned to cross its orientation with that of the imaging path. A 3D piece, specially modeled and printed in our laboratory, secured the camera and polarizer to the optical part of the slit lamp. We imaged a positive USAF 1951 target (R1DS1P, Thorlabs, USA) to estimate the resolution of our imaging system (Fig. 6), utilizing the plotted intensity profile of the sixth group with ImageJ software. We then used a target (#54-440, Edmund Optics, USA) to evaluate the depth of field at a 15 line-pairs (lp)/mm pattern in our system. The intensity profile along a blue line was analyzed using ImageJ software to measure the signal contrast of the 15 lp/mm object. The markings on the image provided a direct readout of the depth of field in millimeters. 

Figure 5.

 

Imaging path modelization.

Figure 6.

 

Zoomed-in image of the USAF 1951 target, along the intensity profile of blue line from the second to the sixth elements of the sixth group, each blue point represents a pixel.

Clinical Investigation

In a clinical investigation of a medical device validated by an ethics committee (IDRCB: 2021-A01496-35) and the Agence Nationale de Sécurité du Medicament et des produits de Santé (ANSM), we acquired images of different diseases or anatomic conditions after pupillary dilation, including cataract, posterior capsule opacification, penetrating keratoplasty, FECD, corneal neovascularization, and even cells in the retrolental vitreous. 

Results

Optical Performances

Each measurement was carried out using the ×40 position of the slit lamp magnification changer. With the USAF 1951 target, the resolution appeared to be constrained to the sixth group, third element, corresponding to a resolution of 80.6 lp/mm (6.2 µm), the sixth and the fifth elements were not resolved (see Fig. 6). 

The depth of field measured was 758 µm (Fig. 7). We imaged the cross-sectional view of the illumination beam in the observation plane (Fig. 8), using a scattering surface to confirm the removal of light at the center of the beam. The analysis of the intensity profile along the blue line with ImageJ software reveals the brightest point of illuminated area (255) was 35 times the maximum value of the dark area. The horizontal axis represents the distance in millimeters in the observation plane, the dark area sized 4.1 mm, and the outer diameter of the illumination ring sized 7.4 mm. 

Figure 7.

 

Image of a depth of field measurement target along the blue line with its intensity profile as a function of its depth of field value.

Figure 8.

 

Cross-sectional image of the illumination beam in the observation plane with the intensity profile traced using ImageJ software. The horizontal axis represents distance in millimeters in the object plane.

Clinical Imaging

Our innovative prototype device produced images of anterior segment with a 9.7 mm field of view, when using the ×40 position of the slit lamp magnification changer. In 100% of the cases, one or more clear images on the area of interest were obtained without glare, and without disturbing reflections in most of the cases. Some patients with pseudophakia showed minimal to moderate light reflection. In this version, the diaphragm and obturator diameters were fixed. We were able to obtain images from all patients, but image quality could be diminished in cases of pupillary dilatation of less than 6 mm (darker image and longer acquisition time exposing to the risk of motion blur). We did not observe any particular difficulty with patients with high myopia with significant macular atrophy. The amount of reflected light even reduced the acquisition time. The images were sufficiently resolved to allow precise analysis of diseased areas, such as each guttae of the FECD of even cells present in the anterior vitreous (Figs. 9, 10). The first application we will develop will be to specify the correspondences between this new retro-illumination modality and corneal thickness mapping in FECD, the preliminary results of which are illustrated in Figure 11. 

Figure 9.

 

Far red, high resolution, reflection free anterior retro-illuminated pictures of diverse anterior segment diseases or specific conditions, exemplified for highlighting the device’ potentiality. Iris posterior synechiae occur in intraocular inflammation. Both pictures illustrate the capacity to resolve cells and how it may help monitoring the inflammation level during the course of the disease. Implication of posterior capsular opacification after cataract surgery in visual acuity decrease may sometimes be difficult to objectify. Our retro-illuminated pictures will help clinicians and researchers. Subtle scarring of the cornea may also be complex to photography. This example once again demonstrates the device’s ability to record the finest details of corneal healing. Fuchs endothelial corneal dystrophy and Lattice dystrophy (a stromal dystrophy) are representative of the interest of our device, which will make it easy to follow their evolution, which is extremely slow. Comparison of high-resolution images will make it possible to objectively assess the accumulation of corneal lesions.

Figure 10.

 

Other examples of the various clinical applications of high-resolution retro-illumination: monitoring of recurrences of epithelial stromal dystrophies after keratoplasty, precise mapping of epithelial basement membrane, and Bowman’s membrane dystrophies, improved analysis and traceability of posterior capsular opacities and cataracts (to understand certain declines in visual acuity, and to keep track and for educational purposes), analysis of subtle endothelial lesions (in this case, iatrogenic lesions) and, once again, better characterization of Fuchs’ endothelial corneal dystrophy (this imaging has made it possible to describe a totally unknown aspect of this dystrophy: the radial organization of the Guttae in certain patients, the significance of which remains to be determined).

Figure 11.

 

Example of complementarity between retro-illumination and posterior thickness and elevation mapping in Fuchs Corneal Endothelial Dystrophy (FECD). Here, a 68-year-old patient presenting with a central form of FECD, with particularly confluent Guttae, classified 3 in the modified Krachmer classification and showing clinical edema on optical coherence tomography. Only high-performance retro-illumination images, as provided by our prototype, will enable us to analyze the correspondences between the different imaging modalities.

Discussion

This simple new imaging device allows the full clinical information of the retro-illumination that is part of the basic anterior segment examination taught to every ophthalmologist to be fully exploited. The use of a non- or barely visible wavelength eliminates the glare of the dilated patient, thus reducing motion blur and palpebral blink, and allows for very accurate focusing. It makes the examination very easily acceptable to all patients, including the most photophobic. Its only disadvantage is that it provides a greyscale image, but, in retro-illumination, the color provides practically no additional information. The only colored structures that needs to be identified are the pigmented (brown) deposits which are very easily visible in black with far-red illumination and can be distinguished from other non-pigmented deposits which appear in grey. The “optical” suppression (without post-processing) of the various reflections allows direct analysis of the entire backlit surface, from the size of the dilated pupil. Only certain reflections remain in certain patients with pseudophakia. Analysis of a large series of patients is underway to clarify why only some patients with pseudophakia are affected and to further develop the device (NCT05717543). The high resolution allows imaging of the finest details needed by clinicians, such as pigmented deposits that measure in the order of a few micrometers. The overall quality of the images should allow the development of powerful image analysis methods involving, for example, the quantification of specific lesions and the comparison of repeated images during patient follow-up of unwanted reflective artifacts. 

The cleanliness of the flat mirror is crucial to reduce light diffraction on its surface and prevent any unintentional artifacts of the dark area. The obturator has the advantage of completely blocking light; however, by obstructing the central part of the beam, we lose the brightest portion, thus requiring a powerful light source. A conceivable approach would be to investigate the modification of the beam shape to concentrate light intensity along its edges. 

Capturing images of curved biological tissues, such as the cornea and lens, lowers the resolutions previously measured as achievable. Nevertheless, our device remains suitable for observing medically relevant elements, such as cells, sized approximately 20 × 20 µm or pigments, sized approximately 10 × 10 µm. Additionally, the depth of field enables the observation of elements measuring 70 µm or more across a thickness greater than the central cornea (500 µm), rendering it well-suited for examining elements such as deposits scattered on the entire posterior corneal surface. 

None of the other devices proposed to exploit the principle of anterior segment imaging by retro-illumination combine these advantages: coaxial digital retro-illumination photography is in white light and does not suppress glare as effectively⁸; asymmetric retro-illumination provides very high resolution images on 820 × 580 µm only and can therefore be described as retro-illumination microscopy.⁹ It does not provide the same information as the retro-illumination used by clinicians. 

Our retro-illumination device prototype provides high-resolution images of the different structures of the anterior segment of the eye, easy to acquire mainly because they do not dazzle patients. Even with the modifications we propose, retro-illumination will remain a frugal technology compared with other imaging methods – which are also undeniably useful – such as OCT, confocal microscopy, and their derivatives. We believe that these images have the potential to facilitate the diagnosis and monitoring of many different diseases both in routine use and in clinical trials where they could be used as objective endpoints. 

Acknowledgments

The authors would like to thank Carine Labruyere (Clinical Research Unit, Saint-Etienne University Hospital) for her hard work in submitting the clinical investigation application to the health authorities. 

Author Contributions: A.A. participated in the conceptualization of the work, interpretation of data, analysis, and drafting of the manuscript. S.P. and O.D.C. participated in patient recruitment and clinical data collection. T.L., S.U., and J.M.P. participated in the conceptualization of the work and interpretation of data. S.U. and J.M.P. participated in the software programming. P.G. and G.T. participated in the study design, data collection, data analysis and interpretation, and manuscript writing. G.T. participated in the project administration, supervision, and methodology. All authors approved the final version of the manuscript. 

Disclosure: A. Ain, retro-illumination device prototype (P); S. Poinard, None; T. Lepine, retro-illumination device prototype (P); O. Dorado-Cortez, None; S. Urbaniak, None; J.-M. Papillon, None; P. Gain, retro-illumination device prototype (P); G. Thuret, retro-illumination device prototype (P)