Fundus Photography in the 21st Century--A Review of Recent Technological Advances and Their Implications for Worldwide Healthcare

BACKGROUND

The introduction of fundus photography has impacted retinal imaging and retinal screening programs significantly.

LITERATURE REVIEW

Fundus cameras play a vital role in addressing the cause of preventive blindness. More attention is being turned to developing countries, where infrastructure and access to healthcare are limited. One of the major limitations for tele-ophthalmology is restricted access to the office-based fundus camera.

RESULTS

Recent advances in access to telecommunications coupled with introduction of portable cameras and smartphone-based fundus imaging systems have resulted in an exponential surge in available technologies for portable fundus photography. Retinal cameras in the near future would have to cater to these needs by featuring a low-cost, portable design with automated controls and digitalized images with Web-based transfer.

CONCLUSIONS

In this review, we aim to highlight the advances of fundus photography for retinal screening as well as discuss the advantages, disadvantages, and implications of the various technologies that are currently available.

Automated estimation of choroidal thickness distribution and volume based on OCT images of posterior visual section

A variety of vision ailments are indicated by anomalies in the choroid layer of the posterior visual section. Consequently, choroidal thickness and volume measurements, usually performed by experts based on optical coherence tomography (OCT) images, have assumed diagnostic significance. Now, to save precious expert time, it has become imperative to develop automated methods. To this end, one requires choroid outer boundary (COB) detection as a crucial step, where difficulty arises as the COB divides the choroidal granularity and the scleral uniformity only notionally, without marked brightness variation. In this backdrop, we measure the structural dissimilarity between choroid and sclera by structural similarity (SSIM) index, and hence estimate the COB by thresholding. Subsequently, smooth COB estimates, mimicking manual delineation, are obtained using tensor voting. On five datasets, each consisting of 97 adult OCT B-scans, automated and manual segmentation results agree visually. We also demonstrate close statistical match (greater than 99.6% correlation) between choroidal thickness distributions obtained algorithmically and manually. Further, quantitative superiority of our method is established over existing results by respective factors of 27.67% and 76.04% in two quotient measures defined relative to observer repeatability. Finally, automated choroidal volume estimation, being attempted for the first time, also yields results in close agreement with that of manual methods.

Quantitative shadow compensated optical coherence tomography of choroidal vasculature

Conventionally rendered optical coherence tomography (OCT) images of the posterior segment contain shadows which influence the visualization of deep structures such as the choroid. The purpose of this study was to determine whether OCT shadow compensation (SC) alters the appearance of the choroid and the apparent choroidal vascularity index (CVI), an OCT-derived estimated ratio of luminal to total choroidal volume. All scans were shadow compensated using a previously published algorithm, binarized using a novel validated algorithm and extracted binarized choroid to estimate CVI. On 27 raw swept-source OCT volume-scans of healthy subjects, the effect of SC on CVI was established both qualitatively and quantitatively. In shadow compensated scans, the choroid was visualized with greater brightness than the neurosensory retina and the masking of deep tissues by retinal blood vessels was greatly reduced. Among study subjects, significant mean difference in CVI of -0.13 was observed between raw and shadow compensated scans. Conventionally acquired OCT underestimates both choroidal reflectivity and calculated CVI. Quantitative analysis based on subjective grading demonstrated that SC increased the contrast between stromal and luminal regions and are in agreement with true tissue regions. This study is warranted to determine the effects of SC on CVI in diseased eyes.

Sensitivity of corneal biomechanical and optical behavior to material parameters using design of experiments method

The optical performance of the human cornea under intraocular pressure (IOP) is the result of complex material properties and their interactions. The measurement of the numerous material parameters that define this material behavior may be key in the refinement of patient-specific models. The goal of this study was to investigate the relative contribution of these parameters to the biomechanical and optical responses of human cornea predicted by a widely accepted anisotropic hyperelastic finite element model, with regional variations in the alignment of fibers. Design of experiments methods were used to quantify the relative importance of material properties including matrix stiffness, fiber stiffness, fiber nonlinearity and fiber dispersion under physiological IOP. Our sensitivity results showed that corneal apical displacement was influenced nearly evenly by matrix stiffness, fiber stiffness and nonlinearity. However, the variations in corneal optical aberrations (refractive power and spherical aberration) were primarily dependent on the value of the matrix stiffness. The optical aberrations predicted by variations in this material parameter were sufficiently large to predict clinically important changes in retinal image quality. Therefore, well-characterized individual variations in matrix stiffness could be critical in cornea modeling in order to reliably predict optical behavior under different IOPs or after corneal surgery.

Semi-automated quantification of hard exudates in colour fundus photographs diagnosed with diabetic retinopathy

Background

Hard exudates (HEs) are the classical sign of diabetic retinopathy (DR) which is one of the leading causes of blindness, especially in developing countries. Accordingly, disease screening involves examining HEs qualitatively using fundus camera. However, for monitoring the treatment response, quantification of HEs becomes crucial and hence clinicians now seek to measure the area of HEs in the digital colour fundus (CF) photographs. Against this backdrop, we proposed an algorithm to quantify HEs using CF images and compare with previously reported technique using ImageJ.

Methods

CF photographs of 30 eyes (20 patients) with diabetic macular edema were obtained. A robust semi-automated algorithm was developed to quantify area covered by HEs. In particular, the proposed algorithm, a two pronged methodology, involved performing top-hat filtering, second order statistical filtering, and thresholding of the colour fundus images. Subsequently, two masked observers performed HEs measurements using previously reported ImageJ-based protocol and compared with those obtained through proposed method. Intra and inter-observer grading was performed for determining percentage area of HEs identified by the individual algorithm.

Results

Of the 30 subjects, 21 were males and 9 were females with a mean age of the 50.25 ± 7.80 years (range 33–66 years). The correlation between the two measurements of semi-automated and ImageJ were 0.99 and 0.99 respectively. Previously reported method detected only 0–30% of the HEs area in 9 images, 30–60% in 12 images and 60–90% in remaining images, and more than 90% in none. In contrast, proposed method, detected 60–90% of the HEs area in 13 images and 90–100% in remaining 17 images.

Conclusion

Proposed method semi-automated algorithm achieved acceptable accuracy, qualitatively and quantitatively, on a heterogeneous dataset. Further, quantitative analysis performed based on intra- and inter-observer grading showed that proposed methodology detects HEs more accurately than previously reported ImageJ-based technique. In particular, we proposed algorithm detect faint HEs also as opposed to the earlier method.

Semi-automated quantification of retinal IS/OS damage in en-face OCT image

A variety of vision ailments are indicated by structural changes in the retinal substructures of the posterior segment of the eye. In particular, integrity of the inner-segment/outer-segment (IS/OS) junction directly relates to the visual acuity. In the en-face optical coherence tomography (OCT) image, IS/OS damage manifests as a dark spot in the foveal region, and its quantification, usually performed by experts, assumes diagnostic significance. In this context, in view of the general scarcity of experts, it becomes imperative to develop algorithmic methods to reduce expert time and effort. Accordingly, we propose a semi-automated method based on level sets. As the energy function, we adopt mutual information which exploits the difference in statistical properties of the lesion and its surroundings. On a dataset of 27 en-face OCT images, segmentation obtained by the proposed algorithm exhibits close visual agreement with that obtained manually. Importantly, our results also match manual results in various statistical criteria. In particular, we achieve a mean Dice coefficient of 85.69%, desirably close to the corresponding observer repeatability index of 89.45%. Finally, we quantify algorithmic accuracy in terms of two quotient measures, defined relative to observer repeatability, which could be used as bases for comparison with future algorithms, even if the latter are tested on disparate datasets.

An automated algorithm for blood vessel count and area measurement in 2-D choroidal scan images

We present an automated algorithm for the detection of blood vessels in 2-D choroidal scan images followed by a measurement of the area of the vessels. The objective is to identify vessel parameters in the choroidal stroma that are affected by various abnormalities. The algorithm is divided into five stages. In the first stage, the image is denoised to remove sensor noise and facilitate further processing. In the second stage, the image is segmented in order to find the region of interest. In the third stage, three different contour detection methods are applied to address different challenges in vessel contour. In the fourth stage, the outputs of the three contour detection methods are combined to achieve refined vessel contour detection. In the fifth and final stage, the area of these contours are measured. The results have been evaluated by a practicing opthalmologist and performance of the algorithm relative to expert detection is reported.

Adaptive optics scanning laser ophthalmoscopy may support early diagnosis of glaucoma

Purpose:

To compare image characteristics of retinal nerve fiber layer (RNFL) between glaucoma patients and healthy controls using adaptive optics scanning laser ophthalmoscopy (AOSLO).

Methods:

This was a cross-sectional pilot study with two groups: a glaucoma group with patients with moderate or severe glaucoma as per the Hodapp–Parrish–Anderson classification system and a control group with healthy individuals. The optic nerve damage in moderate glaucoma was predominantly located in only one hemisphere; the other hemisphere was un- or minimally affected on optical coherence tomography and automated perimetry and is referred to as early glaucoma. The structure of RNFL bundles and gain (%) in RNFL images with mean pixel values between 15 and 35 were analyzed. Imaging was performed one degree away from the optic disc margin at two and four cardinal clock positions in the glaucoma and control groups, respectively. The field of view was 1.3° at 2.3 m resolution. We studied one eye per participant.

Results:

There were 11 glaucoma patients and 7 healthy controls. Imaging was successful at 88% of the locations in controls and early glaucoma; the reflectivity differed significantly (0.51 and 0.56, respectively, P < 0.001) but not the structure of RNFL bundles (Cohen’s Kappa 0.11) between them. In patients with moderate and severe glaucoma, imaging was successful only at 46% of the locations; RNFL bundles were not discernible, and RNFL reflectivity did not differ from those with early glaucoma (P < 0.11).

Conclusion:

The recorded gain (%) of RNFL images obtained using AOSLO could be an objective indicator of early glaucoma.

Is Ahmed Glaucoma Valve Consistent in Performance?

Purpose

To investigate the mechanism of action and consistency in flow characteristics of the Ahmed glaucoma valve (AGV) under simulated physiological conditions in-vitro and to evaluate whether resistance during priming has any effect on performance of the device.

Methods

Each newly opened AGV device was connected to a digital manometer and was primed with normal saline. The device was then placed in a saline bath and connected to an open manometer, a digital manometer, and an infusion pump. Saline was infused at a rate of 3 μL/min for 24 hours. Digital manometer readings were recorded at 4 Hz.

Results

Data obtained from 9 devices are presented as medians (ranges). The priming pressure was 1130 (835, 1625) mm Hg. Pressure versus time curves showed two distinct phases; transient and steady phases. The transient phase peak pressure was 24 (13, 45) mm Hg. In the steady phase, opening and closing pressures were 13 (10, 17) and 7 (4, 9) mm Hg, respectively; the valve leaflets briefly opened every 73.9 (51, 76.6) minutes and the fluctuation of pressure (difference between opening and closing pressures) was 6 (3, 9) mm Hg. The Spearman correlation coefficient between priming and opening and priming and closing pressure was ρ = −0.13 (P = 0.72) and ρ = −0.36 (P = 0.33), respectively.

Conclusions

The device showed functionality like a valve. The resistance during priming did not affect opening and closing pressures of the AGV. This study showed variable in vitro performance of the AGV.

Translational Relevance

These laboratory findings might, at least partly, explain the variability in the clinical outcome of the device.

Thermal Imaging of the Ocular Surface in Thyroid Eye Disease: A Comparison between Active, Inactive and Healthy Eyes

Aim: To report the temperature differences on the ocular surface using infrared thermal imaging (TI) in thyroid eye disease (TED) and healthy eyesMethods: Prospective, consecutive, comparative case series comparing TI parameters between active and inactive TED with healthy controls. The data captured included baseline demography, activity of TED and the TI parameters. Area under the receiver operating characteristic curves (AUROC) were calculated for the temperature values to determine the sensitivity and specificity in distinguishing active from inactive TED. The Youden index and the predictor cut off values were also reported.Results: The study included 11 eyes with active TED, 46 eyes with inactive TED and 40 eyes healthy patients. Temperatures of pre-determined points on the ocular surface in degrees were compared between the three groups. Temperature at the caruncle, medial and lateral conjunctiva was noted to be significantly higher in the active TED group compared to inactive TED and healthy eyes. The most favorable Youden index (0.7) was noted for the caruncle with an AUROC value of 0.91, a sensitivity of 91% and a specificity of 79%. Correlation coefficient for the caruncular temperature with the corresponding CAS (clinical activity score) was 0.65 (95% C.I. 0.45 to 0.78, p < .0001).Conclusion: Thermal imaging in TED is an objective way of documenting surface inflammation by the surrogate marker of temperature change. It supplements CAS in clinically evident cases and could be useful in staging eyes with subtle clinical signs as well as those where signs have reduced following initiation of treatment.

Automated detection of choroid boundary and vessels in optical coherence tomography images

Structural changes in the choroid, a layer located between the retina and sclera, could indicate various vision impairments. Consequently, ophthalmologists inspect optical coherence tomography (OCT) scans of the posterior section of the eye towards making diagnosis. With a view to assist diagnosis, we propose an automated technique for segmentation of the choroid layer. Specifically, we detect the upper and lower boundaries of the choroid using structural similarity and adaptive Hessian analysis. Subsequently, we detect choroid vessels within those boundaries using a level set method. Experimental results are presented using spectral domain (SD) OCT images.

Investigation of alterations in multifractality in optical coherence tomographic images of in vivo human retina

Optical coherence tomography (OCT) enables us to monitor alterations in the thickness of the retinal layer as disease progresses in the human retina. However, subtle morphological changes in the retinal layers due to early disease progression often may not lead to detectable alterations in the thickness. OCT images encode depth-dependent backscattered intensity distribution arising due to the depth distributions of the refractive index from tissue microstructures. Here, such depth-resolved refractive index variations of different retinal layers were analyzed using multifractal detrended fluctuation analysis, a special class of multiresolution analysis tools. The analysis extracted and quantified microstructural multifractal information encoded in normal as well as diseased human retinal OCT images acquired <italic<in vivo</italic<. Interestingly, different layers of the retina exhibited different degrees of multifractality in a particular retina, and the individual layers displayed consistent multifractal trends in healthy retinas of different human subjects. In the retinal layers of diabetic macular edema (DME) subjects, the change in multifractality manifested prominently near the boundary of the DME as compared to the normal retinal layers. The demonstrated ability to quantify depth-resolved information on multifractality encoded in OCT images appears promising for the early diagnosis of diseases of the human eye, which may also prove useful for detecting other types of tissue abnormalities from OCT images.

Effect of Intraocular Pressure and Anisotropy on the Optical Properties of the Cornea: A Study Using Polarization Sensitive Optical Coherence Tomography

PURPOSE

We hypothesize that because of the anisotropic properties of the cornea, there should be a nonuniform change in birefringence with an increase in intraocular pressure (IOP). In this in vitro study, anisotropic properties, stress distribution within the cornea, and the effect of IOP on changes in stress level were investigated.

DESIGN

Button inflation tests for deformation with polarization sensitive optical coherence tomography were used to demonstrate optical and material anisotropy on ex vivo human corneas.

METHODS

Inflation tests were performed on human donor corneoscleral rims. Using a turntable and hydrostatic column, each corneoscleral rim was subjected to a hydrostatic pressure of 0, 10, 15, and 20 mm Hg. At each pressure step, 4 scans at 0, 45, 90, and 135 degrees were taken by a polarization sensitive optical coherence tomography system, and the birefringence images and normal intensity-based images were recorded; images were later compiled for analysis.

RESULTS

The retardation changed with the axis of orientation (P [T ≤ t] 1-tailed = 0.025) and IOP (P [T ≤ t] 1-tailed = 0.019). Optical thickness of the cornea decreased with increasing IOP.

CONCLUSIONS

The optical properties of the cornea are modified with change in IOP. This is not uniform because of distinct anisotropic properties. Anisotropic properties may unpredictably affect the optical quality of cornea during or after the surgeries. Changes in corneal birefringence can be also used as a tool for measuring the IOP of the eye.

Automated 2D-3D quantitative analysis of corneal graft detachment post DSAEK based on AS-OCT images

BACKGROUND AND OBJECTIVE

In current ophthalmological practices, assessment of graft condition post Descemet's stripping automated endothelial keratoplasty (DSAEK) is performed qualitatively using few (four) anterior-segment optical coherence tomography (AS-OCT) radial B-scans. From those scans, clinicians need to mentally synthesize the graft in 3D, and estimate its overall condition. In contrast, quantitative representation of 360° thickness profile would facilitate better visualization of graft condition, and hence medical decision making. Consequently, clinicians seek to quantify potential detachments in 3D from the aforementioned sparse B-scans. Against this backdrop, aiming to assist doctors in making an accurate postoperative assessment, we attempted at 3D visualization and quantification of donor lenticule separation (DLS) using only four AS-OCT radial B-scans.

METHODS

We developed an automated methodology to model and quantify DLS in 3D using only four AS-OCT B-scans. Firstly, we demonstrated that detachment can be viewed as a tubular vessel-like structure and hence can be detected using Hessian matrix analysis. Secondly, a two-stage interpolation was employed for determining the 3D profile of the detachment. Finally, various clinically significant parameters including type of separation (communicative and non-communicative), volume and 360° thickness profile of the detachment, thickness (central and peripheral) of the recipient cornea and donor graft were also obtained. Accuracy of the proposed algorithm was substantiated via thorough statistical analysis, specifically, vis-á-vis intra- and inter-observer repeatability using Dice coefficient (DC).

RESULTS

On twenty seven eyes of 27 patients (male and female), four radial OCT B-scans with 45° angular separation taken per eye, proposed algorithm found that donor graft detached completely in 3 eyes and detached one side (communicative) in remaining 24 eyes which is in agreement with expert's opinion. Quantitatively, proposed algorithm achieves a mean DC of 81.35% with manual reference which is close to the corresponding inter-observer repeatability value of 86.77%. Volume estimation of the detachment indicates that 11 eyes had 0-25%, 9 had 25-50%, 5 had 50-75% and 2 had 75-100% detachment.

CONCLUSIONS

Accuracy of the proposed methodology was corroborated vis-à-vis observer delineation. This model of image analysis may aid in prognostication of graft outcome in patients with graft detachment after DSAEK.

Foveal cone count reduction in resolved endophthalmitis: an adaptive optics scanning laser ophthalmoscopy (AO-SLO)-based prospective pilot study

AIM

To report the foveal cone count in eyes with resolved endophthalmitis vis-à-vis normal fellow eyes using an indigenous adaptive optics scanning laser ophthalmoscopy (AO-SLO).

METHODS

In a prospective cross-sectional comparative pilot study, we recruited patients with resolved endophthalmitis in one eye (study eye) and a normal fellow eye (control eye). Collected data included measurement of the best-corrected visual acuity (BCVA), spectral-domain optical coherence tomography (OCT) imaging and AO-SLO imaging and cone counting at the fovea in both eyes.

RESULTS

The study included 12 eyes of 6 patients. The mean age was 51.66±11.97 years (median 56 years). BCVA in all control eyes was 20/20 (logarithm of the minimum angle of resolution (LogMAR) 0), and in the study, eyes was 0.21±0.13 (median 0.19, Snellen 20/30; p=0.001; 95% CI -0.39 to -0.09). The follow-up was 18.66±12.32 (median 20 months). The cone count at the fovea in the control eye was 4356.33±1993.93 (median 4498), and in the study eye, it was 2357.16±1541.17 (median 2187.5; p=0.03; 95% CI -3556 to -1082).

CONCLUSIONS

Eyes with resolved endophthalmitis with near-normal vision have reduced number of foveal cones even in absence of OCT-detected gross structural changes.

TRANSLATIONAL RELEVANCE

The current work describes the application of cellular-level imaging technique called adaptive optics scanning laser ophthalmoscopy (AO-SLO) to the clinical condition of resolved endophthalmitis. The study of retinal cell biology at the cellular level is possible using the emerging technology of AO-SLO. This new investigative modality that has the potential to image the retina at the cellular level until the photoreceptors is more likely to unravel the pathophysiology of a variety of retinal diseases.

Diagnostic Quality Assessment of Ocular Fundus Photographs: Efficacy of Structure-Preserving ScatNet Features

Various ophthalmic procedures critically depend on high-quality images. For instance, efficiency of teleophthalmology, a framework to bring advanced eye care to remote regions, is determined by the capability of assessing diagnostic quality of ocular fundus photographs (FPs), and rejecting poor-quality ones at the source. In this context, we study algorithmic methods of classifying high- and low-quality FPs. Crucially, diagnostic quality (DQ) - determined by clinically, but not necessarily perceptually, significant structures - is not synonymous with perceptual appeal. Yet, traditional methods handpick features individually (or in small subsets) to meet certain ad hoc perceptual requirements. In contrast, we investigate the efficacy of a comprehensive set of structure-preserving features, systematically generated by a deep scattering network (ScatNet). Specifically, we consider three advanced machine learning classifiers, train each using ScatNet as well as traditional features separately, and demonstrate that the former ensure significantly superior performance for each classifier under multiple criteria including classification accuracy.

Blebophthalmos: proptosis due to a massive bleb following implant surgery in refractory paediatric glaucoma

A 5-year-old boy with failed combined trabeculotomy and trabeculectomy underwent Ahmed glaucoma valve (AGV) implantation and presented 5 weeks postoperatively, with sudden onset proptosis downward displacement of the globe, limited extraocular movements and displacement of tube from anterior chamber. CT scan of the orbits showed a large cystic bleb with posterior dislocation of the implant and was explanted. In vitro testing of implant for fluid flow characteristics revealed a dysfunctional valve with anterior leak, which could have possibly lead to anterior and inferior fluid accumulation resulting in displacement of implant, a probable cause for this rare complication. The case was successfully managed later with the Aurolab aqueous drainage device implantation. This case report cautions the surgeon for the possibility of a large cystic bleb following abnormal leak in AGV resulting in orbital complications like proptosis, motility limitation and impending optic nerve compression.

Innovative bulls eye drop applicator for self-instillation of eye drops

PURPOSE

To report the role of Bulls eye drop applicator device in self-instillation of eye drops and in prevention of wastage of medicine.

METHODS

A small pocket sized device "Bulls eye drop applicator" was developed, which is a simple, handy, mirror-based attachment which can be attached to any eye drop bottle to help in accurate self-instillation of the drops in the eye. The prototype of this device was tried in ten volunteer subjects (Group 1) and 15 patients (Group 2); they were asked to use lubricating eye drops (5 ml bottle) in one eye without the device (N) and in the other eye with the device (M). The numbers of attempts for the application of eye drops were noted and the residual eye drops in the returned bottles were measured for quantitative assessment.

RESULTS

Ten volunteers and 11/15 patients completed the study. At the completion of the study, there were a total of 232 applications of eye drops in the Group 1 N (without device) and 1 M (with device). To achieve these 232 applications, there were 330 attempts without the device and 266 attempts of instillation were needed with the device (P < 0.0001). In Group 2, there were a total of 544 applications of eye drops; to achieve this, there were 879 attempts in Group 2 N and 685 attempts of instillation in Group 2 M (p < 0.0001). The cumulative quantity of residual drops in the returned bottles collected from Group 2 N was 5.1 ml and it was 19.7 ml in Group 2 M (p = 0.001).

CONCLUSIONS

The number of attempts of instillation reduced significantly with the use of the eye drop applicator device. The use of eye drops applicator device reduced the wastage of drops from 42.2% to 14.6% in Group 1 N Vs Group 1 M and saving of about 35.7 % in Group 2 M compared to Group 2 N.