Abstract
Abstract: :
Purpose: The inter– and intra–grader variability among individuals who grade OCT images for clinical trials has not been reported. Herein we determined Duke OCT Reading Center inter– and intra–grader agreement for OCT images produced during conduct of a study of diabetic macular edema (DME). Methods: The fast macular thickness map protocol was used to produce scans (6 radial scans/visit) for 119 visits of 48 patients with DME enrolled in a clinical trial. Experienced OCT graders first independently determined whether or not a scan was gradable. If gradable, each of the 6 scans for each visit was graded for the presence of macular edema, cysts (including number and size) subretinal fluid (SRF), and epiretinal membrane (ERM). To determine inter–grader agreement, the pre–adjudication assessment of whether a scan was gradable, and the scan grade assigned by each grader, for each of the parameters was compared, respectively. To measure intra–grader variability, a subset of these scans was re–graded on a different date by an individual grader. The results were stratified based on the specific OCT instrument (OCT I and II vs Stratus OCT) to determine effects of lower and higher resolution instruments, respectively. Results: There was excellent inter–grader agreement with respect to the determination of whether a scan was gradable, and the scan grade for each of the evaluated parameters. The agreement for the determination of whether a scan was gradable for macular edema was 96–100%, for cysts was 93–100%, for SRF was 94–100%, and for ERM was 97–100%. Of the gradable scans, inter–grader agreement for macular edema was 82–95%, for cysts was 71–86%, for SRF was 94–99% and for ERM was 94–100%. Likewise, there was excellent intra–grader agreement both with respect to the determination of whether a scan was gradable, and the scan grade for each of the evaluated parameters. The intra–grader agreement for the determination of whether a scan was gradable for macular edema was 100% for cysts was 91–98%, for SRF was 88–98% and for ERM was 77–96%. Of the gradable scans, intra–grader agreement for macular edema was 91–98%, for cysts was 86–98%, for SRF was 92–98% and for ERM was 94–100%. There was no significant difference in the inter–grader or intra–grader agreement for the OCT I/II and Stratus OCT. Conclusions: OCT graders in an OCT Reading Center setting are able to independently grade multiple morphological parameters of OCT scans obtained from eyes with DME with a high level of inter–and intra–grader agreement. These data support the use of an OCT Reading Center to evaluate OCT scans in clinical trials of DME.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • diabetic retinopathy • imaging/image analysis: clinical