Can Commercially Available Handheld Retinal Cameras Effectively Screen Diabetic Retinopathy?

Abstract

Organizations that care for people with diabetes have increasingly adopted telemedicine-based diabetic retinopathy screening (TMDRS) as a way to increase adherence to recommended retinal exams. Recently, handheld retinal cameras have emerged as a low-cost, lightweight alternative to traditional bulky tabletop retinal cameras. Few published clinical trials have been performed on handheld retinal cameras. Peer-reviewed articles about commercially available handheld retinal cameras have concluded that they are a usable alternative for TMDRS, however, the clinical results presented in these articles do not meet criteria published by the United Kingdom Diabetic Eye Screening Programme and the American Academy of Ophthalmology. The future will likely remedy the shortcomings of currently available handheld retinal cameras, and will create more opportunities for preventing diabetic blindness.

Clinicians and administrators have been increasingly motivated to improve rates of retinal exams among their patients with diabetes. Early detection and treatment of sight-threatening diabetic retinal disease can prevent vision loss by as much as 90%, ¹ however, adherence with recommended eye exams has generally been suboptimal. Diabetic retinopathy (DR) remains the leading cause of blindness among working age adults in the United States. In 2014, Lee et al reported that fewer than half of patients with diabetes in the United States receive eye exams at recommended intervals, which are usually once per year. ² This rate of adherence with retinal exams is much lower in rural and socioeconomically disadvantaged communities. Many primary care and endocrinology clinics are looking for alternative methods to provide retinal exams for their patients with diabetes and interest in telemedicine-based diabetic retinopathy screening (TMDRS) has grown considerably. With TMDRS systems, patients are photographed with digital retinal cameras on site with their diabetes care provider and the images are transmitted to remote eye care providers who then return a consult, usually indicating the level of retinopathy and follow up recommendations. ³ TMDRS has been shown to increase uptake of retinal exams among patients with diabetes, ⁴ however, clear guidelines about the minimum quality of retinal photography necessary to serve as a surrogate for face-to-face retinal exams remains an open question.

Handheld Retinal Camera Image Quality

Zhang et al ⁵ performed retinal photography on 56 subjects with a Pictor hand held retinal camera and concluded “Pictor can capture retinal images of sufficient quality to screen for DR with and without [pupil] dilation.” In this study, sensitivity and specificity of interpretation of images with undilated pupils averaged 79% and 77% respectively when 5 reviewers detecting referable DR were compared to a dilated clinical exam. Undilated retinal exams is the predominant way that retinal images are acquired in retinopathy screening programs outside of eye care practices. In this study, 1 out of 5 patients with referable DR will not be referred and may wait another year before the next test. Conversely, more than 1 out of 5 patients will be referred for ophthalmological care unnecessarily. This performance may be lower in actual screening settings when busy medical assistants perform the photography rather than the medical residents who performed this trial. The acceptability of this level of referral accuracy requires examination.

Retinal Image Quality

Over a decade has passed since comprehensive meta-analyses of studies showed that retinal imaging with standard digital retinal cameras was superior to ophthalmoscopic exams for detection of diabetic retinal lesions.^6-8 Since then, standard tabletop digital retinal cameras are still the predominant imaging devices used in TMDRS throughout the world, although smaller, low-cost, lightweight, handheld retinal cameras are beginning to emerge as a convenient alternative to the traditional bulky cameras. These smaller devices fit better with the constraints of space and mobility that are often encountered in diabetes care outside of the eye care practice. Clinical validation studies are now being published, and the results should be reviewed carefully to determine if these new imaging devices provide sufficient image quality to effectively screen for DR. Handheld retinal cameras require more eye-hand coordination and alignment. Patients with diabetes are often less steady and have multiple conditions that affect the view of the retinas, such as cataracts, smaller and more sluggish pupils, and inadequate tear film. What is the minimum image quality necessary to accurately triage patients with diabetes?

Retinal Image Quality Standards

The American Academy of Ophthalmology (AAO) recognizes that screening for diabetic retinopathy using validated digital imaging can be a sensitive and effective detection method, stating that validation is by comparison to the gold standard for the detection and classification of diabetic retinopathy, which is stereoscopic color fundus photographs in 7 standard fields, as defined by the Early Treatment Diabetic Retinopathy Study (ETDRS) group. ⁹ The AAO Quality Secretariat further stated that “caution should be exercised in strictly applying the test characteristics from the reported studies; most tests perform less well in the real-world setting.” ¹⁰ Under these conditions there is no commercially available handheld retinal camera that meets these standards.

The American Diabetic Association has not established a minimum recommended image quality for a screening technique to be considered adequate, and their 2016 position statement only requires that images be of “high quality.” ¹¹ The British Diabetic Association has stated that for any diabetic retinopathy screening program to be viable, it must have at least 80% sensitivity and 95% specificity. ¹² The United Kingdom’s Diabetic Eye Screening is the largest program worldwide, screening over 2 million individuals per year. This program evaluates new retinal cameras every 6 months where “an expert team tests all fundus cameras to ensure they can detect subtle diabetic pathology, are robust and easy to use. . . . Cameras that provide inadequate quality images can introduce risk to a programme.” As of May 2016 no handheld retinal cameras appear on their approved list of 20 non-mydriatic retinal cameras for retinal screening. ¹³

Validation of Handheld Devices

Only a few other studies exist that attempt to validate handheld retinal cameras. For example, Quellec et al ¹⁴ compared the handheld Horus DEC 200 by MIIS to the tabletop AFC-330 by Nidek. They found that the handheld camera required significantly more time to capture images, and that the image quality was significantly worse than the table-top camera. They concluded that the handheld camera was not a cost-effective solution, but could be employed for retinopathy screening. Other studies were related to smartphone devices and products that are still in development with similar findings.

The choice of retinal camera for TMDRS is evolving. Factors that affect the quality of retinal images fall within 3 general categories: optical properties (illumination, light scatter, pupil size requirements, retinal field of view, and depth of focus, and optical artifacts), image file properties (resolution, color depth, and image file type and size), and user properties (alignment, computer-user interface, information system interface, and ergonomics). Currently available handheld retinal cameras are superior to tabletop models in some of these factors but significantly inferior in others. Whether or not to use a handheld retinal camera depends on the goals of the program. Simply documenting that a retinal exam has been performed in order to meet quality measures currently requires lower image quality than a service that effectively detects sight-threatening disease. In some communities with limited access to eye care, TMDRS is often the only chance for patients to be evaluated, and image quality requirements are higher. A few studies^15-17 have found low adherence to referral recommendations for ophthalmological care among patients discovered to have sight-threatening disease, regardless of availability and cost of treatment resources. Arguably, this finding would mean that patient engagement in future treatment and diabetes control needs to begin earlier, before severe retinal disease is present. This indicates the need for higher image quality in TMDRS to detect subtle changes in the retina that herald the beginning of microvascular complications.

Conclusion

TMDRS is gaining popularity, but caution is needed in selecting the appropriate retinal camera for the task. Handheld cameras are a practical, convenient alternative to bulky tabletop cameras, but currently don’t provide sufficient image quality for validated screening programs. The future will likely remedy the shortcomings of handheld cameras, and will create more opportunities for preventing diabetic blindness.