Screening for age-related macular degeneration using nonstereo digital fundus photographs

Diagnostic accuracy of community optometrists for age-related macular degeneration using colour fundus photographs: A pilot evaluation

PURPOSE

The accurate diagnosis of age-related macular degeneration (AMD) represents an important step in delaying and preventing vision loss and achieving optimal patient care. Therefore, this pilot study aimed to estimate the diagnostic accuracy of community optometrists for identifying AMD using colour fundus photographs (CFPs) to support sample size calculations for subsequent definitive studies.

METHODS

Five practising community optometrists were invited to classify a total of 1023 CFPs for the (1) presence of AMD, and, if applicable, (2) stage of AMD (early/intermediate/late geographic atrophy/late neovascular AMD). Diagnosis by referral centre clinicians formed the reference standard. Diagnostic accuracy was assessed by the area under the receiver operating characteristic curve (aROC). Sensitivity, specificity, positive and negative predictive values were also calculated.

RESULTS

Of the 1023 CFPs included in the study, 226 images were of AMD and 797 images were of other ocular conditions or no abnormal findings. Participating community optometrists had a mean (SD) age of 30.2 (8.9) years, 60.0% (3/5) were female and the mean number of years practising in primary eye care was 5.4 (5.4) years. Community optometrists demonstrated excellent performance for diagnosing AMD, with an aROC of 0.86 (95% CI 0.83 to 0.89), sensitivity of 84.5% (95% CI 79.1 to 89.0) and specificity of 88.0% (95% CI 85.5 to 90.1). The aROC (95% CI) for diagnosing early, intermediate, late geographic atrophy and late neovascular AMD was 0.82 (0.73 to 0.91), 0.76 (0.72 to 0.81), 0.69 (0.49 to 0.90) and 0.55 (0.34 to 0.75), respectively.

CONCLUSIONS

These results justify the need for an appropriately powered definitive study to assess community clinicians' diagnostic accuracy for AMD.

Sensitivity and specificity of high-resolution wide field fundus imaging for detecting neovascular age-related macular degeneration

PURPOSE

Early detection and treatment are important management strategies for neovascular age-related macular degeneration (AMD). The purpose of this study was to determine the sensitivity and specificity in detecting neovascular AMD using two wide-field imaging systems: ClarusTM (CLARUS 500™, Carl Zeiss Meditec AG, Jena, Germany) and Optos®(Optos California®, Optos PLC, Dunfermline, United Kingdom), compared to conventional digital fundus photographs.

METHODS

We retrospectively analyzed 109 eyes of 73 consecutive patients with neovascular AMD, who underwent standard examination and multimodal imaging, including fundus photography, and optical coherence tomography (OCT). Unmasked graders utilized slit-lamp biomicroscopy and OCT to diagnose neovascular AMD. Masked graders evaluated ClarusTM, Optos®, and digital fundus photograph methods to determine the presence of choroidal neovascularization associated with AMD. Sensitivity and specificity analyses were performed using combined fundoscopy and OCT as the reference standard.

RESULTS

Ninety eyes were diagnosed with neovascular AMD and the remaining 19 eyes were normal based on the reference standard. Of these, neovascular AMD was detected using ClarusTM in 94.4% (85/90). The sensitivities of Optos® and digital fundus photographs were 81.1% (73/90) and 87.8% (79/90), respectively. The specificities using ClarusTM, Optos®, and digital fundus photographs were 89.5% (17/19), 94.7% (18/19), and 89.5% (17/19), respectively.

CONCLUSION

ClarusTM, with its ability to image high-resolution wide field fundus, was considered superior for diagnosing neovascular AMD with high sensitivity and specificity. It may be a useful screening tool for early detection of neovascular AMD, facilitating prompt referral and treatment.

Evaluation of a deep learning system for the joint automated detection of diabetic retinopathy and age-related macular degeneration

PURPOSE

To validate the performance of a commercially available, CE-certified deep learning (DL) system, RetCAD v.1.3.0 (Thirona, Nijmegen, The Netherlands), for the joint automatic detection of diabetic retinopathy (DR) and age-related macular degeneration (AMD) in colour fundus (CF) images on a dataset with mixed presence of eye diseases.

METHODS

Evaluation of joint detection of referable DR and AMD was performed on a DR-AMD dataset with 600 images acquired during routine clinical practice, containing referable and non-referable cases of both diseases. Each image was graded for DR and AMD by an experienced ophthalmologist to establish the reference standard (RS), and by four independent observers for comparison with human performance. Validation was furtherly assessed on Messidor (1200 images) for individual identification of referable DR, and the Age-Related Eye Disease Study (AREDS) dataset (133 821 images) for referable AMD, against the corresponding RS.

RESULTS

Regarding joint validation on the DR-AMD dataset, the system achieved an area under the ROC curve (AUC) of 95.1% for detection of referable DR (SE = 90.1%, SP = 90.6%). For referable AMD, the AUC was 94.9% (SE = 91.8%, SP = 87.5%). Average human performance for DR was SE = 61.5% and SP = 97.8%; for AMD, SE = 76.5% and SP = 96.1%. Regarding detection of referable DR in Messidor, AUC was 97.5% (SE = 92.0%, SP = 92.1%); for referable AMD in AREDS, AUC was 92.7% (SE = 85.8%, SP = 86.0%).

CONCLUSION

The validated system performs comparably to human experts at simultaneous detection of DR and AMD. This shows that DL systems can facilitate access to joint screening of eye diseases and become a quick and reliable support for ophthalmological experts.

Views of ophthalmologists on the genetics of age-related macular degeneration: Results of a qualitative study

Background

Age-related macular degeneration (AMD) is the leading cause of blindness in industrialized countries. It is a multifactorial disease of the retina modified by environmental/individual (e.g. smoking) and genetic factors. 34 independent genomic loci are associated with the risk to develop AMD; an interaction between smoking and genetics is currently investigated. It is unclear how the knowledge on the strong genetic component has entered the knowledge base of practicing ophthalmologists, and how they inform and counsel their (AMD) patients about it. In this study, we explore the ophthalmologists’ view on AMD genetics, and their inclination towards communicating genetic risks to patients.

Methods

We recruited a purposive sample of thirty German ophthalmologists (office based: n = 15, hospital employees: n = 15, f:8/30), who took part in a recorded semi-standardized interview. Transcripts were analyzed using content analysis.

Results

The majority of office-based ophthalmologists claimed to be unfamiliar with genetics of AMD, in contrast to hospital-affiliated ophthalmologists. Both office and hospital ophthalmologists were convinced that genetics lacks practical relevance in everyday patient care. Many withhold information on heritability or genetic background of AMD from patients and their relatives, for fear of unsettling those individuals. The relevance of the genetic component of AMD or an individuals’ high genetic risk for prevention, e.g. screening or lifestyle modifications in persons with adverse genetic profile, was rated low.

Conclusion

Developing genetic educational programs tailored to the routine care of ophthalmologists may be indicated, as well as a better two-way communication between research and practice. Exploring patient views about their expectations to being informed about genetic disease etiology, or about their individual risk, would help inform communication strategies.

The cost-effectiveness of systematic screening for age-related macular degeneration in South Korea

Background

Interventions that can facilitate early diagnosis of age-related macular degeneration (AMD) will facilitate early treatment and improve clinical outcomes but there has been concerns about additional medical costs to the health care system. An examination through a retina fundus photography by a non-specialist has been suggested as a potential cost-effective alternative to a direct examination by a specialist, but limited scientific data exists on the cost-effectiveness of screening strategies for AMD. Our objective is to conduct an economic evaluation of various population-wide screening strategies for AMD among the South Korean population.

Methods and findings

Using a Markov cohort model, we evaluated the cost-effectiveness of four AMD screening strategies (opportunistic examination, opportunistic treatment, systematic photography, and systematic examination) in comparison with status quo (no screening) for South Korean adults. We projected a life time horizon to study a hypothetical cohort of 100,00 persons of age 40 with and without AMD at baseline. The outcome measures were quality-adjusted life-years (QALYs) gained, cost from the societal perspective, and the incremental cost-effectiveness ratio (ICER) of each strategy. Interventions were evaluated at a willingness-to-pay (WTP) threshold of 30,000,000 KRW ($27,538) per QALY gained. Deterministic and probabilistic sensitivity analyses were conducted to address the model uncertainty. Opportunistic examination was strongly dominated because it generated fewer expected QALYs but incurred greater expected cost than the other screening strategies. The mean lifetime expected costs were 289,013 KRW, 363,692 KRW, 9,351,964 KRW, and 12,309,783 KRW, and the mean QALYs gained were 37.73, 37.75, 40.47, 40.68, for no screening, opportunistic treatment, systematic photography, and systematic examination, respectively. The results were most sensitive to the utility weight of mild AMD, the probability of complication from treatment, the cost of being in mild AMD, and the probability of recovery from complication. After eliminating the two weakly dominated strategies, systematic photography was cost-effective at the ICER of 3,310,448 KRW per QALY in comparison to status quo.

Conclusions

Under the WTP threshold of 30,000,000 KRW per QALY, systematic photography is cost-effective for screening AMD in South Korean adults. Systematic examination by ophthalmologists generates more expected QALY and cost compared to systematic photography.

Multi-categorical deep learning neural network to classify retinal images: A pilot study employing small database

Deep learning emerges as a powerful tool for analyzing medical images. Retinal disease detection by using computer-aided diagnosis from fundus image has emerged as a new method. We applied deep learning convolutional neural network by using MatConvNet for an automated detection of multiple retinal diseases with fundus photographs involved in STructured Analysis of the REtina (STARE) database. Dataset was built by expanding data on 10 categories, including normal retina and nine retinal diseases. The optimal outcomes were acquired by using a random forest transfer learning based on VGG-19 architecture. The classification results depended greatly on the number of categories. As the number of categories increased, the performance of deep learning models was diminished. When all 10 categories were included, we obtained results with an accuracy of 30.5%, relative classifier information (RCI) of 0.052, and Cohen's kappa of 0.224. Considering three integrated normal, background diabetic retinopathy, and dry age-related macular degeneration, the multi-categorical classifier showed accuracy of 72.8%, 0.283 RCI, and 0.577 kappa. In addition, several ensemble classifiers enhanced the multi-categorical classification performance. The transfer learning incorporated with ensemble classifier of clustering and voting approach presented the best performance with accuracy of 36.7%, 0.053 RCI, and 0.225 kappa in the 10 retinal diseases classification problem. First, due to the small size of datasets, the deep learning techniques in this study were ineffective to be applied in clinics where numerous patients suffering from various types of retinal disorders visit for diagnosis and treatment. Second, we found that the transfer learning incorporated with ensemble classifiers can improve the classification performance in order to detect multi-categorical retinal diseases. Further studies should confirm the effectiveness of algorithms with large datasets obtained from hospitals.

Reaching the Unreachable: Novel Approaches to Telemedicine Screening of Underserved Populations for Vitreoretinal Disease

Telemedicine involves electronic communication between a physician in one location and a patient in another location to provide remote medical care. Ophthalmologists are increasingly employing telemedicine, particularly in retinal disease screening and monitoring. Telemedicine has been utilized to decrease barriers to care and yield greater patient satisfaction and lower costs, while maintaining high sensitivity and specificity. This review discusses common patient barriers to eye care, innovative approaches to retinal disease screening and monitoring using telemedicine, and eye care policy initiatives needed to enact large-scale telemedicine eye disease screening programs.

Intra- and inter-rater agreement between an ophthalmologist and mid-level ophthalmic personnel to diagnose retinal diseases based on fundus photographs at a primary eye center in Nepal: the Bhaktapur Retina Study

BACKGROUND

Early detection can reduce irreversible blindness from retinal diseases. This study aims to assess the intra- and inter-rater agreement of retinal pathologies observed on fundus photographs between an ophthalmologist and two-mid level ophthalmic personnel (MLOPs).

METHOD

A population-based, cross-sectional study was conducted among subjects 60 years and above in the Bhaktapur district of Nepal. Fundus photographs of 500 eyes of 500 subjects were assessed. The macula-centered 45-degree photographs were graded twice by one ophthalmologist and two MLOPs. Intra-rater and inter-rater agreements were assessed for the ophthalmologist and the MLOPs.

RESULT

Mean age was 70.22 years ± 6.94 (SD). Retinal pathologies were observed in 55.6 % of photographs (age-related macular degeneration: 34.2 %; diabetic retinopathy: 4.2 %; retinal vein occlusion: 3.8 %). Twelve (2.4 %) fundus pictures were non-gradable. The intra-rater agreement for overall retinal pathologies, retinal hemorrhage, and maculopathy were substantial both for the ophthalmologist as well as for the MLOPs. There was moderate inter-rater agreement between the ophthalmologist and the first MLOP on second rating for overall retinal pathologies, [kappa (k); 95 % CI = 0.59 (0.51-0.66)], retinal hemorrhage [k; 95 % CI = 0.60 (0.41-0.78)], and maculopathy [k; 95 % CI = 0.52 (0.43-0.60)]. Inter-rater agreement between the ophthalmologist and the second MLOP for second rating was moderate for overall retinal pathologies [k; 95 % CI = 0.52 (0.44-0.60)], substantial agreement for retinal hemorrhage [k; 95 % CI = 0. 68 (0.52-0.84)], moderate agreement for maculopathy [k; 95 % CI = 0.59 (0.50-0.67)].

CONCLUSION

There is moderate agreement between the MLOPs and the ophthalmologist in grading fundus photographs for retinal hemorrhages and maculopathy.

Teleretinal imaging for detection of referable macular degeneration

PURPOSE

The purpose of this study was to determine the sensitivity and specificity for detection of referable age-related macular degeneration (AMD) using an existing nonmydriatic telemedicine pathway for diabetic retinopathy screening with comparison to same-day face-to-face examination by a retina specialist.

METHODS

Subjects in this study underwent nonmydriatic and mydriatic digital retinal imaging on the same day as stereoscopic dilated examination of the macula by a retina specialist and the level of AMD was recorded for each eye. Images were graded by two trained readers as nonreferable or referable (AREDS [Age-Related Eye Disease Study] grading of level 3 or greater). Sensitivity and specificity were calculated by comparing referral recommendations between each reader and the retina specialist ("gold standard").

RESULTS

There were 47 subjects (94 eyes) enrolled in the study. Sensitivity for nonreferable AMD with nonmydriatic imaging was 1.0 (reader 1) and 1.0 (reader 2), whereas specificity was 0.75 (reader 1) and 0.91 (reader 2). Sensitivity for referable AMD with nonmydriatic imaging was 0.84 (reader 1) and 0.88 (reader 2), whereas specificity was 0.81 (reader 1) and 0.81 (reader 2).

CONCLUSIONS

Our study showed that nonmydriatic digital retinal imaging had excellent sensitivity and specificity in identifying referable and nonreferable AMD using an existing validated telemedicine pathway for diabetic retinopathy screening.

A survey of current and anticipated use of standard and specialist equipment by UK optometrists

PURPOSE

To investigate current and anticipated use of equipment and information technology (IT) in community optometric practice in the UK, and to elicit optometrists' views on adoption of specialist equipment and IT.

METHODS

An anonymous online questionnaire was developed, covering use of standard and specialist diagnostic equipment, and IT. The survey was distributed to a random sample of 1300 UK College of Optometrists members.

RESULTS

Four hundred and thirty-two responses were received (response rate = 35%). Enhanced (locally commissioned) or additional/separately contracted services were provided by 73% of respondents. Services included glaucoma repeat measures (30% of respondents), glaucoma referral refinement (22%), fast-track referral for wet age-related macular degeneration (48%), and direct cataract referral (40%). Most respondents (88%) reported using non-contact/pneumo tonometry for intra-ocular pressure measurement, with 81% using Goldmann or Perkins tonometry. The most widely used item of specialist equipment was the fundus camera (74% of respondents). Optical Coherence Tomography (OCT) was used by 15% of respondents, up from 2% in 2007. Notably, 43% of those anticipating purchasing specialist equipment in the next 12 months planned to buy an OCT. 'Paperless' records were used by 39% of respondents, and almost 80% of practices used an electronic patient record/practice management system. Variations in responses between parts of the UK reflect differences in the provision of the General Ophthalmic Services contract or community enhanced services. There was general agreement that specialised equipment enhances clinical care, permits increased involvement in enhanced services, promotes the practice and can be used as a defence in clinico-legal cases, but initial costs and ongoing maintenance can be a financial burden. Respondents generally agreed that IT facilitates administrative flow and secure exchange of health information, and promotes a state-of-the-art practice image. However, use of IT may not save examination time; its dynamic nature necessitates frequent updates and technical support; the need for adequate training is an issue; and security of data is also a concern.

CONCLUSION

UK optometrists increasingly employ modern equipment and IT services to enhance patient care and for practice management. While the clinical benefits of specialist equipment and IT are appreciated, questions remain as to whether the investment is cost-effective, and how specialist equipment and IT may be used to best advantage in community optometric practice.

Feasibility of telemedicine in detecting diabetic retinopathy and age-related macular degeneration

Age-related macular degeneration and diabetic retinopathy are important causes of visual impairment and blindness in the world. Because of recent advances and newly available treatment modalities along with the devastating consequences associated with late stages of these diseases, much attention has been paid to the importance of early detection and improving patient access to specialist care. Telemedicine or, more specifically, digital retinal imaging utilizing telemedical technology has been proposed as an important alternative screening and management strategy to help meet this demand. In this paper, we perform a literature review and analysis that evaluates the validity and feasibility of telemedicine in detecting diabetic retinopathy and age-related macular degeneration. Understanding both the progress and barriers to progress that have been demonstrated in these two areas is important for future telemedicine research projects and innovations in telemedicine technology.

The retinal disease screening study: prospective comparison of nonmydriatic fundus photography and optical coherence tomography for detection of retinal irregularities

PURPOSE

To compare the sensitivity of volume scanning with optical coherence tomography (OCT) to nonmydriatic color fundus photography (FP) for the detection of retinal irregularities in asymptomatic populations.

METHODS

Asymptomatic subjects without known ocular disease were recruited over a 6-month period. For each eye, two undilated 45° fundus images and four undilated volume OCT image sets covering the macula and optic nerve were obtained. Color images were evaluated for irregularities both inside and outside the area covered by OCT. OCT image sets were evaluated for internal limiting membrane irregularities, abnormal retinal thickness, hyper/hyporeflective features, and photoreceptor/retinal pigment epithelium (RPE) irregularities. Detection sensitivities were compared and false-negative cases were analyzed.

RESULTS

A total of 284 eyes (144 subjects) were included, with a mean age of 38.1 years (range 18-77). Among 253 eyes (135 subjects) with gradable images from both FP and OCTs, the detection sensitivities for OCT were higher (96.2% infield and 85.7% in full field) than for FP (19.9% infield and 43.8% in full field) for all irregularities evaluated in the study (including epiretinal irregularities, abnormal retinal thickness, intraretinal hyperreflective/hyporeflective features, and photoreceptor/RPE irregularities). Overall, the presence of definite irregularities on either fundus imaging or OCT by eye in this asymptomatic population was 42.6% (121/284), with 39.4% (112/284) of eyes having RPE irregularities such as drusen.

CONCLUSIONS

For detection of a variety of retinal irregularities evaluated in the current study, volume OCT scanning was more sensitive than nonmydriatic retinal photography in our asymptomatic individuals. OCT detected clinically relevant disease features, such as subretinal fluid, that were missed by FP, and had a lower ungradable image rate. It is likely that OCT will be added to photography screening in the near future for chorioretinal disease.

Quality of nonmydriatic digital fundus photography obtained by nurse practitioners in the emergency department: the FOTO-ED study

OBJECTIVE

Nonmydriatic fundus photography by non-ophthalmic-trained personnel has recently been shown to be a potential alternative to direct ophthalmoscopy in the emergency department (ED). We evaluated the reliability of a novel quality rating scale and applied this scale to nonmydriatic fundus photographs taken during routine ED patient encounters to determine factors associated with diminished photograph quality.

DESIGN

Prospective, cross-sectional study.

PARTICIPANTS

We included 350 patients enrolled in the Fundus photography versus Ophthalmoscopy Trials Outcomes in the Emergency Department study who were photographed by nurse practitioners after <30 minutes of training followed by supervision.

METHODS

Photographs of both eyes were graded for quality on 2 occasions by 2 neuro-ophthalmologists. Four regions were independently evaluated for quality: Optic disc, macula, and superior and inferior vascular arcades. Quality as a function of the number of photographs taken was evaluated by Kaplan-Meier analysis. Mixed effects ordinal logistic regression was used to evaluate for predictors of image quality while accounting for the repeated measures design.

MAIN OUTCOME MEASURES

Overall photographic quality (1-5 scale; 5 best).

RESULTS

We evaluated 1734 photographs. Inter- and intraobserver agreements between neuro-ophthalmologists were very good (weighted kappa, 0.84-0.87). Quality of the optic disc area was better than those of other retinal areas (P<0.002). Kaplan-Meier analysis showed that if a high-quality photograph of an eye was not obtained by the third attempt, it was unlikely that one would be obtained at all. A 10-second increase in the interphotograph interval before a total of 40 seconds increased the odds of a 1-unit higher quality rating by 1.81 times (95% confidence interval [CI], 1.68-1.98), and a 10-year increase in age decreased the odds by 0.76 times (95% CI, 0.69-0.85). Black patients had 0.42 times (95% CI, 0.28-0.63) the odds of a 1-unit higher quality rating compared with whites.

CONCLUSIONS

Our 5-point scale is a reliable measure of nonmydriatic photograph quality. The region of interest, interphotograph interval, and patient age and race are significant predictors of image quality for nonmydriatic photographs taken by nurse practitioners in the ED. Addressing these factors may have a direct impact on the successful implementation of nonmydriatic fundus photography into the ED.

Interobserver and intraobserver reliability of detecting age-related macular degeneration using a nonmydriatic digital camera

PURPOSE

To evaluate the interobserver and intraobserver reliability of detecting early and late age-related macular degeneration (AMD) using a nonmydriatic digital camera in two distinct groups of older people.

DESIGN

Prospective study.

METHODS

The two groups consisted of a series of patients older than 70 years hospitalized in a geriatric unit and a younger series of people older than 55 years. In both groups, nonmydriatic color fundus photographs were obtained and graded independently by two ophthalmologists (V.L. and M.S.). No ophthalmic examination was performed. Main outcome measures were frequencies of early and late AMD and interobserver and intraobserver agreement.

RESULTS

Among 233 patients in group 1 (mean age, 84.6 years), only 119 patients (51%) could undergo photography because of associated multiple morbidities. Mean age of group 2 was 63.8 years. In group 1, 35 (14.5%) of 238 pictures were ungradable. In series 2, 65 (9.1%) of 716 pictures were ungradable. Frequencies of early and late AMD were 30.3% and 5.9% vs 12.6% and 2.6% in series 1 and 2, respectively. Interobserver and intraobserver agreement was good or excellent (kappa > 0.6) in both groups.

CONCLUSIONS

In the entire geriatric cohort, 43% of the patients had gradable pictures allowing a diagnosis. These patients would otherwise have had no access to any form of funduscopy. In the younger population, nonmydriatic pictures permitted a diagnosis in 90% of the individuals. Detection of AMD with a nonmydriatic digital camera may lead to large-scale screening and specific management.

Impact of direct electronic optometric referral with ocular imaging to a hospital eye service

AIMS

A study to assess the feasibility, safety, and clinical effectiveness of electronic referral--with and without images--of patients directly from optometrists in primary care to the hospital eye service (HES) in contrast to the traditional paper-based referral, through the general practitioner (GP).

METHODS

Three optometry practices sent consecutive referrals with images through the NHS Net to the HES. The standard General Ophthalmic Service form was electronically redesigned with additional information on patient choice for advice, appointment, or surgery. All paper referrals to the HES from the same three optometry practices before the study period were analysed (control group A) as were all paper referrals from the remaining optometrists in Fife (control group B).

RESULTS

A total of 346 electronic referrals were received over 18 months. 218 (63%) were classified as requiring and 128 (37%) as not requiring a HES appointment. The latter were subsequently examined with unexpected pathology found in three cases (glaucoma, macular pigment epithelial detachment, and possible peripheral retinal tear). In both groups, the major pathologies reported were macular degeneration, cataract, glaucoma, diabetic retinopathy, and abnormal retinal appearances. A total of 17 (15%) patients in group A and 26 (8.4%) patients in group B were classified as not requiring HES appointment. These control groups indicate that approximately 10-15% of paper referrals are not seen in the HES. To summarise, therefore, 63% of people referred by the optometrist directly using electronic referral (with or without images) were given a HES appointment compared to 85% of people referred through the traditional paper method (without images) through their GP.

CONCLUSION

Electronic referral with images to the HES is safe, speedy, efficient, and clinically accurate given some limitations and avoids unnecessary consultation in 37% of referrals.