OCT readers agree on retinal disease features in scans

Optical coherence tomography (OCT) has become increasingly popular for evaluating and managing many retinal diseases, such as age-related macular degeneration (AMD). The OCT images of AMD obtained during conduct of clinical trials usually are interpreted by certified readers to see what kinds of abnormalities might be present. However, not much is known about the reliability of one reader as compared with other readers in detecting and measuring morphologic features in the images. Another area of interest is the reproducibility of results from one reader when evaluating the same scan twice.

Dr. Nanfei Zhang and Dr. Glenn J. Jaffe from Duke University Medical Center in Durham, N.C., and their colleagues tested the inter- and intrareader agreement of three certified OCT readers at the OCT Reading Center at Duke using images from an interventional neovascular AMD clinical trial.

Researchers tested the inter- and the intrareader agreement of OCT scan readers with images from an age-related macular degeneration clinical trial. The readers evaluated and graded for each patient a 7-mm line scan offset by 5° (seen here) and six radial scans. Reprinted with permission of the American Journal of Ophthalmology.

As reported in the July issue of the American Journal of Ophthalmology, an OCT scanner from Carl Zeiss Meditec of Dublin, Calif., was used to scan 132 eyes of 110 patients. Six radial scans and a 7-mm line scan offset by 5° were obtained for each patient. The three readers first independently evaluated whether each scan was gradable, and then they evaluated the gradable scans for multiple morphologic characteristics. Three reader pairs for each scan were then measured for inter-reader agreement.

The researchers discovered that for all three reader pairs, there was 96 to 100 percent agreement on scan gradability and 84 to 96 percent agreement on the actual scan grades assigned for each of the morphologic characteristics.

They also found that, when the readers regraded randomly selected scans about one year later, the intrareader agreement of scan gradability ranged from 96 to 100 percent and that the agreement for scan grades ranged from 79 to 98 percent. The researchers believe that their results indicate that well-trained OCT readers can grade independently with a high level of inter- and intrareader agreement.