Telescreening for Diabetic Retinopathy

Published on 09/03/2015 by admin

Filed under Opthalmology

Last modified 22/04/2025

Print this page

This article have been viewed 1249 times

Chapter 50 Telescreening for Diabetic Retinopathy

Rajiv Raman, Aditi Gupta, Tarun Sharma

Introduction

The criteria for human diseases amenable to screening approaches were defined by the World Health Organization in 1968 ¹ and diabetic retinopathy fulfills all of these. Visual impairment due to diabetic retinopathy is a significant health problem; however, it has a recognizable presymptomatic stage.² The DCCT and the UKPDS established that intensive diabetes management to obtain near-euglycemic control can prevent and delay the progression of diabetic retinopathy in patients with diabetes.^3,⁴ Timely laser photocoagulation therapy can also prevent loss of vision in a large proportion of patients with sight-threatening diabetic retinopathy.⁵ Screening for diabetic retinopathy saves vision at a relatively low cost, which has been demonstrated in various studies.^6,⁷ The American Academy of Ophthalmology recommends annual dilated eye examinations beginning at the time of diagnosis for patients with type II diabetes.² For those with type I diabetes, the recommendation is retinal examination 3–5 years after diagnosis, with annual exams thereafter.² The barriers for successful screening are numerous and include the high cost of care, poor awareness levels, lack of symptoms in the early stages of disease, socioeconomic factors and poor geographical access to care.⁸ Current screening programs for diabetic retinopathy are either ophthalmologist-based (with actual presence of the ophthalmologist at the site of screening) or ophthalmologist-led (no ophthalmologist at the site of screening). Table 50.1 summarizes the key differences between the two models. Telemedicine for retinopathy screening is an ophthalmologist-led screening model, which may be a logical potential alternative for patients who have been noncompliant with the traditional face-to-face examination by an ophthalmologist. Telemedicine is the exchange of medical data by electronic telecommunications technology allowing a patient’s medical problems to be evaluated, monitored, and possibly treated while the patient and physician are located at sites physically remote from each other.⁹ Ophthalmology is uniquely suited for telemedicine as it is a highly visual and image intensive specialty and digital imagery is easily transmitted by electronic means. Remote assessment for diabetic retinopathy provides an ideal model for telehealth screening initiatives and, in fact, has become one of the most common uses for telemedicine in ophthalmology.

Table 50.1 Differences between the ophthalmologist-led and ophthalmologist-based models for screening for diabetic retinopathy

	Ophthalmologist-led model (Telescreening)	Ophthalmologist-based model
Brief description	Paramedical staff acquire data/images, which are then transferred for interpretation by ophthalmologist	Screening is performed by ophthalmologist
Feasibility	Yes, with less human resources	Needs trained expert
Maintenance	Required	Not required
Capital expenditure	More	Less
Revenue expenditure	Less	More
Interobserver bias	Less	More
Digital photo archiving	Yes	No
Acceptance by community	Yes	Yes

Guidelines for telescreening program

American Telemedicine Association telehealth practice recommendations for diabetic retinopathy

American Telemedicine Association (ATA), Ocular Telehealth Special Interest Group, and the National Institutes of Standards and Technology Working Group, established the telescreening guidelines for diabetic retinopathy.¹⁰ The ATA recommends that telehealth programs for diabetic retinopathy should demonstrate an ability to compare favorably with ETDRS film or digital photography. For screening programs with low thresholds for referral, the International Clinical Diabetic Retinopathy Disease Severity Scale may be used in place of ETDRS scales, however, protocols should state the reference standard used for validation and relevant datasets used for comparison.

The ATA recognizes four categories of telescreening programs, of which higher categories (3 and 4) are generally not necessary for screening programs which do not involve actual management of diabetic eye disease:

Category 1: The program allows identification of patients who have no or minimal diabetic retinopathy and distinguishes them from those who have more than minimal diabetic retinopathy.

Category 2: The program allows identification of patients who do not have sight-threatening diabetic retinopathy and distinguishes them from those who have potentially sight-threatening diabetic retinopathy.

Steps of telescreening

The flow of steps of the telescreening process is diagrammed in Fig. 50.1. In brief, patient enrollment is performed after defining the data to be collected. Since ocular telescreening services for diabetic retinopathy satisfy the criteria of low risk telehealth procedures and are within commonly accepted standards of practice, signature consent may not be required. However, practitioners should provide patients with information about the telescreening program they would reasonably want to know, including differences between care delivered using ocular telehealth approaches versus traditional face-to-face encounters, and a description of what is to be done at the patient’s site and the remote site. The data collected includes fundus images, along with patient examination findings (identification, demographic, and medical information) and some morphological information that is used to make a clinical decision. Fundus images of both eyes of the patient are acquired under a fixed, predetermined imaging protocol. These images are taken by a trained technician using a fundus camera. Due to various factors, the quality of the acquired images may be below the grading standard, thus not providing any meaningful information for examination by the reader. This can be addressed by employing an automatic image quality assessment module. An automatic image quality assessment module will ensure that the images transmitted for diagnosis conform to prescribed gradability standards. During the quality assurance process, the gradable images are selected for compression, whereas the identification of poor quality images can trigger reimaging by the technician. The patient data comprising the clinical data and the fundus images are compressed to make them suitable for low-bandwidth network connectivity. The patient data are transmitted to the servers via the Internet or satellite. At the reading center, the images are graded for presence of retinal lesions and the determination of a diabetic retinopathy level; referred to “next level” graders if necessary; and a retinopathy structured report is generated. Only qualified readers should perform retinal image grading and interpretation. If a reader is not a licensed eye care provider, specific training is required. A licensed, qualified eye care provider with expertise in diabetic retinopathy and familiarity with telescreening program technology should supervise the readers. An adjudicating reader (an ophthalmologist with special qualifications in diabetic retinopathy by training or experience) may resolve discrepant interpretations. Image processing algorithms should undergo rigorous clinical validation. A report comprising the findings, the results and the medical advice given by the expert is made available to the patient and the care team at the remote site through an accessible interface.

Fig. 50.1 Flowchart representing the steps of telescreening. This figure shows the sequential steps of telescreening carried out at the imaging center and the reading center.

Technical considerations

Image acquisition

The gold standard for telescreening is the ETDRS 7 mydriatic standard field 35 mm stereoscopic color fundus photographs.¹¹ However, more practical alternatives, such as digital fundus photography^12,¹³ and nonmydriatic fundus photography,^14,¹⁵ have been evaluated. Digital imaging has the advantage of faster and easier acquisition, transmission, and storage. Several investigators have reported a high level of correlation between stereoscopic digital imaging and slide film for the identification of most features of diabetic retinopathy.^12,¹³

Regarding nonmydriatic fundus photography, a higher rate of unreadable photographs has been reported through undilated versus dilated pupils.^14,¹⁵ Diabetic persons often have smaller pupils and a greater incidence of cataracts, which may limit image quality if the procedure is performed through an undilated pupil. Pupil dilation using 0.5% tropicamide is associated with a minimal risk of angle closure glaucoma. Programs using pupil dilation should have a defined protocol to recognize and address this potential complication.

The unsatisfactory performance of nonmydriatic photography has led to the concept of “targeted mydriasis,” offering mydriasis only to a preselected group of patients, in whom undilated photography is known to produce dismal results.¹⁵ However, the exact “target” remains to be defined. Based on ROC curve analysis, Raman et al.¹⁶ predetermined the cutoff values for “target mydriasis” groups as vision <6/12 (20/40 Snellen equivalent) and age >59 years. Staged mydriasis is another option.¹⁷ In this model, a nonmydriatic single digital photograph for screening is taken. If an unsatisfactory nonmydriatic photograph is obtained, the patient undergoes immediate pupillary dilation with 1% tropicamide and the photograph is then repeated. Using this protocol, 75–80% of patients do not require mydriasis.

According to ATA recommendations, image acquisition personnel (“imagers”) should possess the knowledge and skills for independent imaging or with assistance and consultation by telephone, since a licensed eye care professional may not be physically available at all times during a telehealth session.

After acquisition, the transfer of the images can be “real-time” or by a “store-and-forward” technique. In real-time transfer, the captured images and associated data are immediately (“simultaneously”) seen by the remote ophthalmologist. Whereas in a store-and-forward technique, captured images and data are compressed, stored, and then forwarded for retrieval by a remote ophthalmologist later.

Compression

Data and image compression facilitates transmission and storage of retinal images. The time needed for transmission can also be dramatically reduced by image compression.

Compression may be used if algorithms have undergone clinical validation. Image data can be compressed using a variety of standards, including JPEG, JPEG Lossless, JPEG 2000, and Run-length encoding (RLE). The International Standards Organization (ISO/IEC JTC1/SC2/WG10) has prepared an International Standard, ISO/IS-15444–1 (JPEG 2000 Part 1), for the digital compression and coding of continuous-tone still images. This standard is known as the JPEG 2000 Standard. Digital Imaging and Communication in Medicine (DICOM) recognizes JPEG and JPEG 2000 for lossy compression of medical images.¹⁸ ATA recommends that the compression types and ratios should be periodically reviewed to ensure appropriate clinical image quality and diagnostic accuracy. Some studies have attempted to look at the effect of various levels of compression on the quality of the image with both subjective and objective parameters.^19,²⁰ The level of acceptable compression ranges from 1 : 28 to 1 : 52.^19,²⁰

Data transfer, archiving, and retrieval

The described telemedicine models reported earlier used the Internet to transmit images.^21,²² In rural areas and mobile clinics, satellite transmission is a more preferred option because of poor infrastructure. A variety of technologies are available for data communication and transfer. Telescreening programs should determine specifications for transmission technologies best suited to their needs.

The images and reports are transmitted digitally via electronic picture archiving and communication systems (PACS); this eliminates the need for manual file transfer or retrieval. A PACS consists of four major components: the imaging instrumentation, a secured network for the transmission of patient information, workstations for interpreting and reviewing images, and archives for the storage and retrieval of images and reports. The universal format for PACS image storage and transfer is DICOM. To minimize errors, data communications should be compliant with DICOM standards.

Telescreening systems should provide storage capacities in compliance with facility, state, and federal medical record retention regulations. Digital images obtained by telescreening are typically stored locally on a PACS for rapid retrieval. Past images and reports should also be available for retrieval. It is important and is required in the United States by the Security Rule’s Administrative Safeguards section of the Health Insurance Portability and Accountability Act (HIPAA) that facilities have a means of recovering images in the event of an error or disaster.

Security and documentation

Transmission of retinal imaging studies and study results should conform to HIPAA privacy and security requirements. Ocular telehealth systems should have defined network and software security protocols so as to protect patient confidentiality and identification of image data. Protective measures should be taken to safeguard data integrity against intentional or unintentional data corruption. If using the Internet, the Security Rule’s Technical Safeguards section of HIPAA requires that the images be encrypted during transmission. Privacy should be ensured through a minimum 128-bit encryption and two-factor authentication technology.

The ATA recommends that reports should be based on Health Level 7 (HL7) and DICOM standards software forms and should meet interoperability standards. Medical nomenclature should conform to Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT®) standards.

Operational considerations

Detection of diabetic retinopathy and macular edema

As per ATA recommendation, both the ETDRS and International Clinical Diabetic Retinopathy Disease Severity Scales may be used for classifying diabetic retinopathy and macular edema. Nonstereo imaging methods can detect diabetic retinopathy quite well but diabetic macular thickening cannot be detected as reliably. Hence, all nonstereo imaging methods need additional information about retinal thickness for an accurate assessment of macular edema. The retinal thickness analyzer (RTA), which combines two imaging modalities – a wide-angle, red-free black-on-white fundus photograph and a detailed map of retinal thickness at the posterior pole – had mean 93% sensitivity for diagnosis of proliferative diabetic retinopathy and 100% for macular edema when compared with clinical examination.²³

Role of the reading center to grade retinal images

Reading personnel

The gold standard of image reading by ophthalmologists is impractical in rural areas. Ruamviboonsuk et al.²⁵ evaluated the interobserver differences among the nonphysician personnel (local ophthalmic photographers and certified ophthalmic nurses, who attended an intensive instruction course for this screening program) and ophthalmologists (retina specialists and general ophthalmologists without additional training), in the interpretation of single-field digital fundus images for diabetic retinopathy screening. Retinal specialists had the best agreement among the groups. Photographers were more reliable than the nurses. The authors concluded that the retina specialists should be effective interpreters without additional training, general ophthalmologists may need more training, but nonphysician personnel must have comprehensive training.

Handling of ungradable images

ATA guidelines recommend that the inability to obtain or read images should be considered a positive finding and patients with unobtainable or unreadable images should be promptly re-imaged or referred for evaluation by an eye care specialist. Authors from the Joslin Vision Network (JVN) reported that of those images that were judged ungradable, a large proportion had pathology that required referral for comprehensive examination.²⁶ In the Gloucestershire Study, 3.7% of patients had unassessable images (including those with cataract), of whom 10.3% had referable retinopathy.¹⁴

The United Kingdom National Screening Committee (UKNSC) recommends that arrangements should be made for patients with ungradable images to be examined either by an ophthalmologist or by a trained and accredited person supervised by an ophthalmologist (as in the Scottish scheme). It may still not be possible to assess a very small number of patients due to a range of disabilities (for example it may not be possible for a patient to hold still in one position either for assessment or for treatment). It should be noted that some patients with ungradable images may be unsuitable for treatment due to a condition that is not going to be improved with treatment in either eye. Clearly great care must be taken before such a decision is made.

While nonmydriatic photography screening is being successfully used, several factors have been reported to result in ungradable images in nonmydriatic retinal photography. Increasing age is an important factor.²⁷ Media opacity such as cataract and small pupil are other major factors. Scanlon et al.²⁷ suggested that a 20% failure rate for nonmydriatic photography might be acceptable. They also supported the use of nonmydriatic photography for the group of individuals <50 years of age who are at the lowest risk of ungradable images, if the screening programs are directed towards the detection of sight-threatening diabetic retinopathy.

Quality assurance

In 2000, the UKNSC stressed the integration of quality assurance as a core feature of telescreening programs for diabetic retinopathy and proposed the criteria and minimal/achievable standards for each quality assurance objective.²⁸ Since then, other ongoing quality assurance programs have published their methods and outcomes.^29,³⁰ Different screening programs have different criteria for image regrading, resulting in different numbers of retinal photographs that need to be regraded for quality assurance (6% to 46%).³¹

There are two categories of quality assurance:³²

(a) Internal quality assurance that is integrated as part of the day-to-day workflow in a screening program measured against national standards.

(b) External quality assurance which has three main functions: the monitoring of ongoing program performance against the quality standards, the organization of peer-review visits, and the administration of an external proficiency testing system for all graders.

There is international consensus that screening programs for diabetic retinopathy should achieve at least 80% sensitivity, 95% specificity, and <5% technical failure rates.³³ The quality assurance group of the National Screening Committee has recommended that quality assurance should involve the second examination of all images initially reported to have any diabetic retinopathy, together with 10% of the negative images, independently as part of the internal quality assurance system.³⁴

The ATA and UKNSC have established the national standards for quality assurance for a diabetic retinopathy screening program. ATA has specified major categories of performance to be evaluated, as applicable to the program, at the level of the originating site and at the reading center. The UKNSC has provided 19 standards or parameters for quality assurance. Several authors have suggested further measures. Leese et al.³¹ recommended the use of automated grading systems running in parallel with manual grading; and concentrating quality assurance in the smaller number of patients with high-risk nonproliferative disease.

Evaluating telescreening programs

Efficacy

Telescreening has been shown to detect diabetic retinopathy and macular edema with a reasonably high sensitivity and specificity.^9,³⁵ Whited et al.³⁵ reviewed the available literature and noted that the sensitivity and specificity values ranged from 50% to 93% for detection of diabetic retinopathy. Similar high efficacy was reported on comparing teleophthalmology for macular edema detection to both gold standards, i.e. slit-lamp biomicroscopy and stereoscopic photography.³⁵

Patient satisfaction

Since telescreening involves remote care without evaluation by the doctor in person, there are concerns regarding lack of satisfaction among patients. Studies have shown, however, that telescreening has equal, if not better, satisfaction than in-person evaluation by a doctor.^36,³⁷ Paul et al. assessed patient satisfaction levels and factors influencing satisfaction during teleophthalmology consultation in India using a patient satisfaction questionnaire. He found that 37.34% of the patients felt that telescreening is more satisfying than an in-person evaluation, and 60% felt that both models are equally satisfying. It was also noted that patients who asked questions during the screening were 2.18 times more likely to be satisfied with teleophthalmology than those who did not.

Cost-effectiveness

Bjorvig et al.,³⁸ in an economic analysis, concluded that telemedicine was a less costly option for screening in places with higher patient workloads. Telemedicine was also proven to be cost-effective in the prison populations by Aoki et al.,³⁹ where it may have special utility due to costs and safety issues associated with transporting prisoners.

Gomez-Ulla et al.⁶ did a comparative cost analysis of diabetic retinopathy telescreening versus standard ophthalmoscopy, from both Public Healthcare System (PHS) and patient perspectives. The authors concluded that from the PHS perspective, direct fundus examination is less costly than telescreening owing to the higher capital costs required for the purchase of digital imaging equipment. From a global perspective, however, the digital imaging alternative is more convenient because the travel cost and loss of income for the patient are lower.

Recently, Jones et al.⁷ reviewed the evidence available on cost-effectiveness. They concluded that telemedicine is cost-effective for retinopathy screening in remote and rural communities and other groups with travel difficulties and the cost-effectiveness increases with an increase in patient workload.

Advances in telescreening

Recent advances resulting in better and faster telecommunication, miniaturization of diagnostic equipment including digital cameras, and automation of retinal image analysis, offer excellent opportunities to expand telescreening services in more remote areas.

Automated retinal image analysis

The shortage of manpower imposes a limitation on the screening capability of telehealth programs serving a steadily growing diabetic population. Therefore, an automated image analysis system able to detect diabetic retinopathy is a vital necessity, especially in the coming years.

Over the past decade several attempts have been made to either semi-automate or fully-automate retinal image analysis. Tools have been developed for analyzing and enhancing the image quality (correction of illumination, increasing image contrast, histogram equalization, vessel segmentation, edge sharpening, and image deconvolution), and for providing automated identification of pathologic retinal lesions (neural networks, region growing, morphological analysis, and classification algorithms).⁴⁰ Automated identification of retinal lesions can identify the absence or presence of diabetic retinopathy based on the detection of microaneurysms and dot hemorrhages (dark lesions),⁴¹ or can detect referable retinopathy based on the detection of exudates (bright lesions) and blot hemorrhages (dark lesions).⁴²

Winder et al.⁴⁰ conducted a structured survey of algorithms for the automatic detection of retinopathy in digital color retinal images. The authors pointed out the need for clear guidelines and goals in image processing research in order to avoid producing results that are difficult to compare in terms of the success of the algorithms or techniques.