Diabetic macular edema can be objectively detected by retinal imaging

Key Points

Waterloo, Ontario-Retinal imaging technology (Retina Module, Heidelberg Engineering) created for use with a retina tomograph (Heidelberg Retina Tomograph, Heidelberg Engineering) is beneficial in detecting diabetic macular edema (DME) and in particular could be useful in general ophthalmology practices that lack highly specialized imaging instruments, according to Christopher Hudson, PhD, McOptom, FAAO, professor, School of Optometry, University of Waterloo, Ontario.

Dr. Hudson and colleagues set out to evaluate the ability of edema index values generated by the retina tomograph to differentiate between people with DME and those without. The sample consisted of four groups who were followed prospectively. Group one (n = 48) was made up of nondiabetic controls; group two (n = 66), patients with diabetes but no clinically visible diabetic retinopathy; group three (n = 65), patients with hard exudates and/or microaneurysms within the macula but no clinically visible DME; and group four (n = 52), patients with pretreatable DME, defined using stereofundus biomicroscopy. The mean age of patients in each group was 52.7, 54.2, 56.7, and 58.6 years, respectively.

Image evaluation

"It was only by looking at the image itself that we would be able to distinguish if it represented an age-matched control subject or if it was from a patient, and thereby could differentiate whether it was in group one, two, three, or four," said Dr. Hudson, who also is a professor in the Department of Ophthalmology and Vision Science, University of Toronto, Ontario.

In addition, the images were analyzed objectively. After normal data limits were established based on the nondiabetic, age-matched control group, the retinal imaging technology software of the retina tomograph classified each image as normal or abnormal. For single zone criteria, an image was classified as abnormal when a single sector edema index value exceeded a specified confidence interval. Center-to-zone two criteria called for classification as abnormal when the difference between the center and any of the R2 zones exceeded a specified confidence interval.

Researchers found that edema index values were elevated significantly in patients with diabetes compared with the control group.

"That applies to all three groups, so it suggests that there is something occurring even in group two, who would be considered low risk," Dr. Hudson said.

The researchers also observed a progressive increase in the edema index with worsening retinopathy. The average edema index scores and standard deviation for the entire area covered by the ETDRS grid:

Results of the subjective analysis showed excellent results, with nearly 90% sensitivity and specificity of about 95%. For the objective performance, sensitivity was moderate, around 65%, and specificity was 94%. A subsequent analysis performed after optimizing the criteria improved the sensitivity of the objective results to about 80%.

Discussing the clinical relevance of these findings, Dr. Hudson suggested that techniques such as spectral-domain optical coherence tomography (OCT) will become the mainstay for diagnosing retinal edema in most hospitals and retinal practices. Many general ophthalmologists own retina tomograph devices for purposes such as glaucoma diagnosis, however, and may be more likely to add retinal imaging technology to them than to invest in a new OCT device, he added.

Because those clinicians undoubtedly will see numerous patients who have diabetes as well as an ocular condition, the retinal imaging technology would give them another tool to help detect edema or confirm a diagnosis suggested by other findings, Dr. Hudson said.