Retinal Image Quality Research Articles

Computational analysis of retinal image quality with different contact lens designs in keratoconus.

To determine 1) the relative differences in optical quality of keratoconic eyes fitted with four routinely used CL designs and 2) the Zernike coefficients in the residual wavefront aberration map that may be responsible for differences in the optical quality of keratoconic eyes fitted with these CLs. Wavefront aberrations over a 3-mm pupil diameter were measured without and with Kerasoft IC®, Rose K2®, conventional spherical Rigid Gas Permeable (RGP), and Scleral CLs in 15 mild to moderate keratoconic eyes (20 - 28years) and under unaided viewing in 10 age-similar non-contact lens wearing controls. The resultant through-focus curves constructed for the logarithm of Neural Sharpness (logNS) Image Quality (IQ) metric were quantified in terms of peak value, best focus, and depth of focus. Sensitivity analyses determined the impact of the residual Zernike coefficients of keratoconic eyes fitted with CLs on the IQ of controls at emmetropic refraction. The peak IQ and depth of focus were similar with Rose K2®, conventional RGP, and Scleral CLs (p>0.05, for all) but significantly better than Kerasoft IC® CLs (p<0.01 for all). Best focus was similar across all four CLs (p>0.2 for all). However, the IQ parameters of all the lenses remained significantly poorer than the controls (p<0.01, for all). The IQ of the controls dropped to keratoconic levels with induced residual lower-order Zernike terms and 3rd-order coma across all lenses in the sensitivity analysis (p<0.001). IQ of keratoconic eyes remain suboptimal with routinely dispensed CL designs, largely due to residual lower-order aberrations and coma, all relative to the controls. The performance drop appears greater for the Kerasoft IC® CL relative to the other CL designs. These results may provide the optical basis for psychophysical spatial visual performance reported earlier across these four CL designs for keratoconus.

How Reliable Is Pyramidal Wavefront-Based Sensor Aberrometry in Measuring the In Vivo Optical Behaviour of Multifocal IOLs?

Cataract or refractive lens surgery, along with the implantation of multifocal intraocular lenses (MF-IOL), enables a complete range of functional far, near and intermediate vision. Refractive, diffractive and extended depth of focus (EDoF) or combination of these principles represent the technology used to obtain this multifocality. Aberrometry makes it possible to study the aberrations induced by MF-IOLs. Among the different optical principles available to measure ocular aberrations, pyramidal wavefront-based sensor (PWS) aberrometry shows the highest resolution with MF-IOLs. Retinal image quality measured by a PWS aberrometer differed significantly according to the technology of the implanted lens. Monofocal and diffractive lenses showed the highest values of far-distance retinal image quality, followed by refractive and EDoF lenses; however, retinal image quality analysed in diffractive lenses appears to be more dependent on residual refractive error. Considering this limitation, PWS-aberrometry could be used to compare diffractive lenses. Nevertheless, further studies are needed to provide additional information about the clinical retinal image quality of MF-IOLs and to help surgeons in the important preoperative selection of IOLs.

Smartphone-based fundus imaging for evaluation of Retinopathy of Prematurity in a low-income country: A pilot study.

To evaluate the feasibility of a novel and simple smart phone-based Retinopathy of Prematurity (ROP) screening approach in a resource-constrained setting. This cross-sectional validation study was conducted at the Department of Ophthalmology and Neonatal Intensive Care Unit (NICU) of The Aga Khan University Hospital, Pakistan, from January 2022 to April 2022. A total of 63 images of eyes with active ROP (stage-1, 2, 3, 4 and/or plus or pre-plus disease) were included in this study. The stage of ROP was documented by the principal investigator using an indirect ophthalmoscope and retinal images were obtained using this novel technique. These images were shared with two masked ROP experts who rated the image quality and determined the stage of ROP and presence of plus disease. Their reports were compared with the initial findings reported by principal investigator using indirect ophthalmoscope. We reviewed 63 images for image quality, stage of ROP and presence of plus disease. There was significant agreement between the gold standard and the Rater-1 and 2 for the presence of plus disease (Cohen's kappa was 0.84 and 1.0) and the stage of the disease (Cohen's kappa 0.65 and 1.0). There was significant agreement between the Rater for presence of plus disease and any stage of ROP (Cohen's κ: 0.84 and 0.65 for plus disease and any stage of the ROP, respectively). Rater-1 and 2 rated 96.83% and 98.41% images as excellent / acceptable respectively. High quality retinal images can be captured with a smartphone and 28D lens without using any additional adapter equipment. This approach of ROP screening can form basis of telemedicine for ROP in resource constrained areas.

A deep retinal image quality assessment network with salient structure priors

Retinal image quality assessment is an essential prerequisite for diagnosis of retinal diseases. Its goal is to identify high quality retinal images in which anatomic structures and lesions attracting ophthalmologists’attention most are exhibited clearly and definitely while reject poor quality images. Motivated by this, we mimic the way that ophthalmologists assess the quality of retinal images and propose a method termed SalStructuIQA. First, two salient structures are detected for retinal quality assessment. One is large-size salient structures including optic disc and exudates in large-size. The other is tiny-size salient structures mainly including vessels. Then the proposed two salient structure priors are incorporated with deep convolutional neural network (CNN) to enforce CNN pay attention to these salient structures. Accordingly, two CNN architectures named Dual-branch SalStructIQA and Single-branch SalStructIQA are designed for the incorporation, respectively. Experimental results show that the Fscore of our proposed Dual-branch SalStructIQA and Single-branch SalStructIQA are 0.8723 and 0.8662 respectively on the public Eye-Quality dataset, which demonstrates the effectiveness of our methods.

A study on disability glare vision in young adult subjects

The full assessment of the visual system must include the evaluation of the optical quality of the eye and neural visual functions. The objective evaluation of the retinal image quality is often carried out by computing the point spread function (PSF) of the eye. The central part of the PSF is associated with optical aberrations and the peripheral areas with scattering contributions. In that sense, visual acuity and contrast sensitivity function tests can be considered the perceptual neural response to those contributions characterizing the eye’s PSF. However, in natural viewing conditions, visual acuity tests may provide good vision while contrast sensitivity tests can reveal visual impairment in glare vision conditions, such as exposure to bright light sources or night driving conditions. Here we present an optical instrument for the study of disability glare vision under extended Maxwellian illumination to assess the contrast sensitivity function under glare conditions. The limit of the Total Disability Glare threshold, tolerance, and glare adaptation will be investigated as a function of the angular size of the glare source (GA) and the contrast sensitivity function in young adult subjects.

Intraocular composition of higher order aberrations in non-myopic children.

This study examined anterior corneal, internal ocular, and total ocular higher order aberrations (HOA's), and retinal image quality in a non-myopic, paediatric cohort. Anterior corneal aberrations were derived from corneal topography data captured using a Placido disk videokeratoscope (E300, Medmont International), and whole eye HOA's were measured using a Hartmann-Shack wavefront sensor (COAS-HD, Wavefront Sciences). The associations between HOA's and age, sex, refractive error, and axial length were explored using correlation analyses. Data for 84 children aged between 5 and 12 years (mean ± standard deviation spherical equivalent refraction (SER), +0.63 ± 0.35 D; range 0.00 to +1.75 D) were included, and an eighth order Zernike polynomial was fit for 4 and 6 mm pupil diameters for both the anterior corneal and total ocular HOA's, from which internal ocular HOA's were calculated via subtraction following alignment to a common reference axis (pupil centre). Internal ocular HOA's were of greater magnitude than previous studies of adolescents and adults, however partial internal "compensation" of HOA's was observed, which resulted in reduced levels of HOA's and excellent retinal image quality. Few significant associations were observed between HOA's and age, SER, and axial length (all correlations, p > 0.001), and there were minimal sex-based differences (all comparisons, p > 0.005). Coefficients for vertical coma ( and ) and spherical aberration ( and ), were most strongly associated with the visual Strehl ratio based on the optical transfer function (VSOTF), which indicated that the absolute magnitudes of these Zernike coefficients have the greatest impact on retinal image quality in this paediatric cohort. These findings provide an improved understanding of the optics and retinal image quality of children's eyes.

Early Impact of Laser Vision Correction (LVC) on the Stability and Quality of the Retinal Image

This retrospective comparative study analyzes the early postoperative impact of laser vision correction for myopia on the optical quality and stability of functional vision using a double-pass aberrometer. Retinal image quality and visual function stability were assessed preoperatively, one and three months after myopic laser in situ keratomileuses (LASIK) and photorefractive keratectomy (PRK) using double-pass aberrometry (HD Analyzer, Visiometrics S.L, Terrassa, Spain). The parameters analyzed included vision break-up time (VBUT), objective scattering index (OSI), modulation transfer function (MTF), and Strehl ratio (SR). The study included 141 eyes of 141 patients, of whom 89 underwent PRK and 52 underwent LASIK. No statistically significant differences were noted between the two techniques in any analyzed parameters at three months postoperatively. However, a significant drop was observed in all parameters one month after PRK. Only the OSI and VBUT remained significantly altered from baseline at the three months follow-up visit, with an increased OSI by 0.14 +/- 0.36 (p < 0.01) and a shortened VBUT by 0.57 +/- 2.3 s (p < 0.01). No correlation was found between the changes in optical and visual quality parameters and age, ablation depth, or postoperative spherical equivalent. The stability and quality of the retinal images were similar between LASIK and PRK at three months postoperatively. However, significant degradation in all parameters was found one month after PRK.

Clinical Retinal Image Quality of a Non-diffractive Wavefront-Shaping Extended Depth of Focus (Vivity) Intraocular Lens.

To evaluate clinical retinal optical image quality following implantation of an extended depth of focus intraocular lens (EDOF IOL) (Vivity; Alcon Laboratories, Inc), and to compare it with a monofocal and a trifocal IOL. This prospective, comparative, case-control study included 88 eyes implanted with: (1) 19 monofocal IOLs (AcrySof SA60AT; Alcon Laboratories, Inc); (2) 38 EDOF IOLs (AcrySof IQ Vivity); and (3) 31 trifocal IOLs (AT LISA tri 839MP; Carl Zeiss Meditec AG). Total root mean square, ocular lower (LOA) and higher (HOA) order aberrations, point spread function (PSF) Strehl ratio (PSF with LOA), and PSF Strehl ratio excluding LOA (PSF without LOA) were analyzed using a Pyramidal WaveFront-based sensor aberrometer Osiris (Costruzione Strumenti Oftalmici) at two different pupil sizes (3 and 4 mm). The trifocal IOL showed the highest PSF without LOA at both pupil sizes (0.52 ± 0.12 and 0.31 ± 0.07, respectively), followed by the AcrySof SA60AT (0.39 ± 0.10 and 0.27 ± 0.07) and AcrySof IQ Vivity (0.34 ± 0.11 and 0.24 ± 0.09) (P < .001). The AcrySof IQ Vivity and monofocal IOLs were comparable (P > .05). Despite the comparable postoperative low spherical equivalent among the IOL groups, the AT LISA tri 839MP retinal image quality (PSF with LOA) was the most severely affected by such residual refractive errors (dropped to 0.26 ± 0.06 at 3 mm; P < .001) compared to the monofocal AcrySof SA60AT (0.24 ± 0.07 at 3 mm) and EDOF Acrysof IQ Vivity (0.23 ± 0.06 at 3 mm) groups. The PSF with LOA was comparable (P > .05) among the three groups at both the 3-and 4-mm pupil size. Although trifocal IOLs provided significantly better retinal image quality if influence of LOA is excluded, they also demonstrated to be the most sensitive to residual refractive errors. Both the EDOF Acrysof IQ Vivity and mono-focal AcrySof SA60AT IOLs showed a comparable retinal image quality, and they are also comparable with trifocal IOLs when considering the clinically real PSF (PSF with LOA). [J Refract Surg. 2023;39(2):103-110.].

Influence of Interocular Differences and Alcohol Consumption on Binocular Visual Performance.

The purpose of this study was to assess the influence of a moderate breath-alcohol content (BrAC of 0.40 mg/L) on binocular visual performance for different visual functions after inducing different levels of interocular differences with the use of filters. A total of 26 healthy young subjects were enrolled. The participants participated in two sessions: one without alcohol consumption and another after alcohol consumption. In each session and for the different filter conditions (subjects were wearing Bangerter foil of 0.8 and BPM2 fog filter on the dominant eye), monocular and binocular visual function was evaluated by measuring visual acuity, contrast sensitivity, visual discrimination capacity (and successively by calculating their corresponding binocular summations) and stereopsis (near and distance stereoacuity). In addition, interocular differences were calculated for different retinal-image quality and straylight parameters. All monocular and binocular visual functions were analyzed and stereopsis was significantly impaired by alcohol and filters (p < 0.05). Interocular differences for different ocular parameters and binocular summations for visual parameters were negatively affected by filters but not alcohol. Significant correlations (averaging all the experimental conditions analyzed) were found, highlighting: the higher the interocular differences, the lower the binocular summation and the poorer the stereopsis and, therefore, the worse the binocular visual performance.

OTRE: Where Optimal Transport Guided Unpaired Image-to-Image Translation Meets Regularization by Enhancing.

Non-mydriatic retinal color fundus photography (CFP) is widely available due to the advantage of not requiring pupillary dilation, however, is prone to poor quality due to operators, systemic imperfections, or patient-related causes. Optimal retinal image quality is mandated for accurate medical diagnoses and automated analyses. Herein, we leveraged the Optimal Transport (OT) theory to propose an unpaired image-to-image translation scheme for mapping low-quality retinal CFPs to high-quality counterparts. Furthermore, to improve the flexibility, robustness, and applicability of our image enhancement pipeline in the clinical practice, we generalized a state-of-the-art model-based image reconstruction method, regularization by denoising, by plugging in priors learned by our OT-guided image-to-image translation network. We named it as regularization by enhancing (RE). We validated the integrated framework, OTRE, on three publicly available retinal image datasets by assessing the quality after enhancement and their performance on various downstream tasks, including diabetic retinopathy grading, vessel segmentation, and diabetic lesion segmentation. The experimental results demonstrated the superiority of our proposed framework over some state-of-the-art unsupervised competitors and a state-of-the-art supervised method.

Latest Development in Extended Depth-of-Focus Intraocular Lenses: An Update.

In recent years, there has been an overwhelming influx of different types of intraocular lenses (IOLs) as treatment for presbyopia. The extended depth-of-focus (EDOF) technology creates a single elongated focal point to enhance depth of focus, in contrast to the multiple foci of multifocal (MF) lenses. In this way, the EDOF lenses aim to reduce photic phenomena, glare, and halos, which have been reported in MF IOLs. A potential disadvantage of this is a blur due to decreased retinal image quality when the amount of the aberrations is increased excessively. Multifocality and EDOF characteristics are not exclusive of each other. Frequently, EDOF IOLs are combined with MF optical designs, a bifocal IOL may exhibit EDOF characteristics, likewise an aspheric monofocal IOL or a diffractive or refractive trifocal IOL. Thus, EDOF lenses are commonly subjected to confusion. A wide range of different types of EDOF lenses are available on the market to surgeons. In this practical update, we aim to clarify what is a true EDOF lens, classify the different types of the EDOF lenses based on their optical principle and review their recently reported outcomes. Comprehensive patient examination and selection, combined with knowledge of the most updated options and adequate patient counseling, can avoid dissatisfaction and yield the desired outcomes.

Automated malarial retinopathy detection using transfer learning and multi-camera retinal images

Cerebral malaria (CM) is a fatal syndrome found commonly in children less than 5 years old in Sub-saharan Africa and Asia. The retinal signs associated with CM are known as malarial retinopathy (MR), and they include highly specific retinal lesions such as whitening and hemorrhages. Detecting these lesions allows the detection of CM with high specificity. Up to 23% of CM, patients are over-diagnosed due to the presence of clinical symptoms also related to pneumonia, meningitis, or others. Therefore, patients go untreated for these pathologies, resulting in death or neurological disability. It is essential to have a low-cost and high-specificity diagnostic technique for CM detection, for which We developed a method based on transfer learning (TL). Models pre-trained with TL select the good quality retinal images, which are fed into another TL model to detect CM. This approach shows a 96% specificity with low-cost retinal cameras.

Effect of accommodation on coma at central and peripheral retina

Coma is one of the most common ocular higher order aberrations and highly affects the quality of image. It is assumed that corneal aberrations are balanced by internal (lenticular) aberrations so that retinal image quality may not have great impact. However, during accommodation, the shape, position, and curvature of the crystalline lens changes which might disrupt this balance between internal and corneal aberration. This study aimed to investigate the effect of accommodation on primary coma ( and ) and secondary coma ( and ) in relaxed and accommodated eyes. Zernike coefficients were measured in 53 subjects with Hartmann-Shack aberrometer both at the central and peripheral retina up to 30° off-axis in horizontal and vertical meridians. The process was repeated with 2.50 D accommodation stimulus and comas were compared with and without accommodation. Root-mean-square of total coma was also assessed. With accommodation, vertical comas changed to more negative value and horizontal comas changed to more positive values in most of the off-axis positions. In contrast, the secondary vertical comas became less negative and secondary horizontal comas became more negative with accommodation in most of the off-axis fixations. Thus, the results showed that accommodation affects coma which depends up on position of the fixation.

Comparison of ocular aberrations and retinal image quality provided by a validated aberrometer and a new open field aberrometer for VEMoS project

AbstractPurpose: The aim of this observational study was to compare the retinal image quality and ocular aberrations provided by a validated aberrometer and by a new open field aberrometer.Methods: This is an observational and comparative study composed by healthy eyes not presenting cataract or ocular pathologies affecting media transparence or visual function. Patients underwent a measurement of ocular aberrations by a closed field aberrometer (Osiris, CSO) and by an open field aberrometer developed for VEMoS European project (OFA, CSO). Both measurements were performed by a qualified optometrist and the pupil size selected for the analysis was 4 mm. SPSS program was applied for the statistical analysis, particularly Wilcoxon test was used to compare both devices considering a significance of 0.05.Results: The study included 20 eyes of 20 patients with a mean age of 36.2 ± 9.7 years. The sphere was significantly more negative with Osiris in comparison with OFA (−0.33 ± 2.17 diopters vs. 0.29 ± 2.17 diopters, p &lt; 0.001), the spherical aberration was also slightly lower with Osiris (0.02 ± 0.03 microns vs. 0.03 ± 0.03 microns, p: 0.045). The rest of aberrometric coefficients including the cylinder, coma, trefoil, quadrefoil and secondary astigmatism showed no significant differences between devices (p &gt; 0.05). Moreover, the point spread function did not presented statistical differences, but a trend to obtain higher values was observed in Osiris (0.219 ± 0.127 vs. 0.189 ± 0.126, p: 0.370).Conclusions: The sphere provided by an open field aberrometer was significantly more positive comparing with a close field aberrometer probably associated to an accommodative response induced in close field aberrometers. Similarly, the spherical aberration assumed small changes between devices, which would be generated also by accommodation. The rest of parameters showed no differences between devices indicating a suitable agreement between devices for these coefficients.

A new approach for sensitivity improvement of retinal blood vessel segmentation in high-resolution fundus images based on phase stretch transform

The eye-fundus photograph is widely used for eye examinations. Accurate identification of retinal blood vessels could reveal information that is helpful for clinical diagnoses of many health disorders. Although several methods have been proposed to segment images of retinal blood vessels, the sensitivity of these methods is plausible to be improved. The algorithm’s sensitivity refers to the algorithm’s ability to identify retinal vessel pixels correctly. Furthermore, the resolution and quality of retinal images are improving rapidly. Consequently, new segmentation methods are in demand to overcome issues from high-resolution images. This study presented improved performance of retinal vessel segmentation using a novel edge detection scheme based on the phase stretch transform (PST) function as its kernel. Before applying the edge detection stage, the input retinal images were pre-processed. During the pre-processing step, non-local means filtering on the green channel image, followed by contrast limited adaptive histogram equalization (CLAHE) and median filtering, were applied to enhance the retinal image. After applying the edge detection stage, the post-processing steps, including the CLAHE, median filtering, thresholding, morphological opening, and closing, were implemented to obtain the segmented image. The proposed method was evaluated using images from the high-resolution fundus (HRF) public database and yielded promising results for sensitivity improvement of retinal blood vessel detection. The proposed approach contributes to a better segmentation performance with an average sensitivity of 0.813, representing a clear improvement over several benchmark techniques

Learning for retinal image quality assessment with label regularization

Background and Objective: The assessment of the image quality is crucial before the computer-aided diagnosis of fundus images. This task is very challenging. Firstly, the subjective judgments of graders on image quality lead to ambiguous labels. Secondly, despite being treated as classification in existing works, grading has regression properties that cannot be ignored. Solving the ambiguity problem and regression problem in the label space, and extracting discriminative features, have become the keys to quality assessment.Methods: In this paper, we proposed a framework that can assess the quality of fundus images accurately and reasonably based on deep convolutional neural networks. Drawing on the experience of human graders, a dual-path convolutional neural network with attention blocks is designed to better extract discriminative features and present the bases of decision. Label smoothing and cost-sensitive regularization are designed to solve the label ambiguity problem and the potential regression problem respectively. Besides, a large number of images are annotated by us to further improve the results.Results: We conducted our experiments on the largest retinal image quality assessment dataset with 28,792 retinal images. Our approach achieves 0.8868 precision, 0.8786 recall, 0.8820 F1, and 0.9138 Kappa score. Results show that our approach outperforms state-of-the-art methods.Conclusions: The promising performances reveal that our methods are beneficial to retinal image quality assessment and have potential in other grading tasks.

Optical modelling of an accommodative light field display system and prediction of human eye responses.

The spatio-angular resolution of a light field (LF) display is a crucial factor for delivering adequate spatial image quality and eliciting an accommodation response. Previous studies have modelled retinal image formation with an LF display and evaluated whether accommodation would be evoked correctly. The models were mostly based on ray-tracing and a schematic eye model, which pose computational complexity and inaccurately represent the human eye population's behaviour. We propose an efficient wave-optics-based framework to model the human eye and a general LF display. With the model, we simulated the retinal point spread function (PSF) of a point rendered by an LF display at various depths to characterise the retinal image quality. Additionally, accommodation responses to the rendered point were estimated by computing the visual Strehl ratio based on the optical transfer function (VSOTF) from the PSFs. We assumed an ideal LF display that had an infinite spatial resolution and was free from optical aberrations in the simulation. We tested points rendered at 0-4 dioptres of depths having angular resolutions of up to 4x4 viewpoints within a pupil. The simulation predicted small and constant accommodation errors, which contradict the findings of previous studies. An evaluation of the optical resolution on the retina suggested a trade-off between the maximum achievable resolution and the depth range of a rendered point where in-focus resolution is kept high. The proposed framework can be used to evaluate the upper bound of the optical performance of an LF display for realistically aberrated eyes, which may help to find an optimal spatio-angular resolution required to render a high quality 3D scene.

Predicting the effects of defocus blur on contrast sensitivity with a model-based metric of retinal image quality.

We measure the effect of defocus blur on contrast sensitivity with Sloan letters in the 0.75-2.00 arc min range of letter gaps. We compare our results with the prediction of the Dalimier and Dainty model [J. Opt. Soc. Am. A25, 2078 (2008)JOAOD60740-323210.1364/JOSAA.25.002078] and propose a new metric of retinal image quality that we define as the model limit for very small letters. The contrast sensitivity is measured for computationally blurred Sloan letters (0, 0.25, and 0.50 diopters for a 3 mm pupil) of different sizes (20/40 to 20/15 visual acuity), and subjects look through a small (2 mm) diaphragm to limit the impact of their own aberration on measurements. Measurements and model predictions, which are normalized by the blur-free condition, weakly depend on letter size and are in good agreement with our metric of retinal image quality. Our metric relates two approaches of modeling visual performance: complete modeling of the optotype classification task and calculation of retinal image quality with a descriptive metric.

Optimising image quality with EyeMax Mono lens in dry age-related macular degeneration.

To investigate clinical outcomes in patients with dry age-related macular degeneration (AMD) after intracapsular implantation of a novel EyeMax Mono macular lens. In this study, 22 phakic eyes of 19 moderate to advanced dry AMD patients with macular disciform scar and/or macular atrophy who were followed up for ≥ 3months after surgery were studied. A thorough pre-operative ophthalmological examination was performed, including measurement of corrected distance visual acuity in logMAR and ETDR. Following phacoemulsification, the EyeMax Mono lens was implanted intracapsularly via a 2.2-mm clear corneal incision to improve retinal image quality in all areas of the macula ≤ 10° from the central fovea. Main outcome measures included optimisation of corrected distance visual acuity and surgical safety. Male-to-female ratio was 13:6. Mean age at surgery was 68.37 ± 10.23years. The mean duration of post-operative follow-up was 7.91 ± 3.42months. The mean post-operative refractive spherical equivalent improved to + 2.31 ± 1.56 D with significant visual improvement as early as 3months post-operatively. Post-operative corrected distance visual acuity improved significantly from 1.05 ± 0.45 to 0.72 ± 0.43 logMAR (P < 0.001), equivalent to mean ETDRS of 49.55 ± 20.05 (P < 0.001). There were no major surgical complications, either intra- or post-operatively, except in two patients who experienced intra-operative haptic rupture. Extended macular vision lenses appear to have a comparable safety profile as standard IOLs in the short to medium term. It could be the preferred lens for improving and preserving visual acuity in moderate to advanced dry AMD patients.

A GAN-based deep enhancer for quality enhancement of retinal images photographed by a handheld fundus camera

ObjectiveDue to limited imaging conditions, the quality of fundus images is often unsatisfactory, especially for images photographed by handheld fundus cameras. Here, we have developed an automated method based on combining two mirror-symmetric generative adversarial networks (GANs) for image enhancement. MethodsA total of 1047 retinal images were included. The raw images were enhanced by a GAN-based deep enhancer and another methods based on luminosity and contrast adjustment. All raw images and enhanced images were anonymously assessed and classified into 6 levels of quality classification by three experienced ophthalmologists. The quality classification and quality change of images were compared. In addition, image-detailed reading results for the number of dubiously pathological fundi were also compared. ResultsAfter GAN enhancement, 42.9% of images increased their quality, 37.5% remained stable, and 19.6% decreased. After excluding the images at the highest level (level 0) before enhancement, a large number (75.6%) of images showed an increase in quality classification, and only a minority (9.3%) showed a decrease. The GAN-enhanced method was superior for quality improvement over a luminosity and contrast adjustment method (P＜0.001). In terms of image reading results, the consistency rate fluctuated from 86.6% to 95.6%, and for the specific disease subtypes, both discrepancy number and discrepancy rate were less than 15 and 15%, for two ophthalmologists. ConclusionsLearning the style of high-quality retinal images based on the proposed deep enhancer may be an effective way to improve the quality of retinal images photographed by handheld fundus cameras.