Abstract
Purpose: :
To determine if mydriatic ultrawide-field fundus imaging identifies diabetic retinopathy (DR) lesions not observed w/ ETDRS film fundus photography (ETDRS photos) or dilated fundus exam (DFE).
Methods: :
Mydriatic Optos® 200 degree (D200), ETDRS photos and masked retina specialist DFE were obtained. Two masked readers graded images for presence/extent of DR lesions and clinical DR severity. Distribution of H/MA, venous beading (VB), IRMA, and NVE on D200 were compared to ETDRS photos. A 3rd masked retina specialist adjudicated discrepancies. Unweighted (K) and weighted (KW) kappa statistics (linear scale) assessed agreement.
Results: :
206 eyes (103 diabetic patients) were evaluated. By ETDRS photos there was no DR in 23 eyes (12%), mild NPDR in 36 (18%), moderate NPDR in 72 (36%), severe/very severe NPDR in 17 (9%), & PDR in 51 (26%). DR severity between D200 & ETDRS photos matched in 67% (KW=0.74, K=0.57) and was within 1 level in 91%. D200 & DFE matched in 58% (KW=0.67, K=0.45), and was within 1 level in 87%. When D200 did not match DFE (N=86), higher agreement occurred between D200 & ETDRS photos (KW=0.70, K=0.50) than DFE & ETDRS photos (KW=0.33, K=0.08). Retinal lesions distribution on D200 in relation to ETDRS photo coverage (within ETDRS fields; outside ETDRS fields; both) was HMA (68%, 30%, 2%), VB (61%, 13%, 26%), IRMA (69%, 27%, 4%), NVE (66%, 34%, 0%). NVEs were seen in 29 eyes by ETDRS photos or DFE. NVEs in 12 (41%) eyes were seen by both ETDRS photos & DFE, 11 (38%) by ETDRS photos and 6 (21%) by DFE. All NVE seen on ETDRS photos were seen on D200. NVE in 11 (38%) eyes were seen on ETDRS & D200 and not on DFE. NVE in 5 (17%) eyes seen by DFE were not seen in either D200 or ETDRS photos. NVE in 9 (31%) additional eyes were seen on D200 that were not observed on either ETDRS photos or DFE. Of these, 4 (44%) were within ETDRS fields and did not increase DR severity. The 5 additional D200 detected NVEs outside ETDRS fields led to PDR detection in 3 of 9 (33%) eyes. Considering all retinal lesions detected outside ETDRS fields, 29 eyes (14.5%) would have been graded as more severe DR than ETDRS photos and none graded as less severe.
Conclusions: :
D200 images had substantial agreement with ETDRS photos and DFE in determining DR severity (>91% within 1 step). Agreement was less for DFE, but validity of DFE findings by ETDRS photos only occurred in 26%. Nearly 1/3 of DR lesions were located outside the ETDRS image area. Ultrawide imaging increased NVE identification >30% and DR severity grade in >14%. Ultrawide field imaging may identify more DR lesions than ETDRS photos or DFE (and perhaps more accurately than DFE), with potential implications for disease progression.
Keywords: diabetic retinopathy • imaging/image analysis: clinical • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)