Abstract
Purpose:
To establish the agreement between these two assessment modalities for detecting presence/absence of hard exudates (HE) and its interaction with the graders experience.
Methods:
One hundred eyes of 51 consecutive patients (mean age 59.5, min 28 max 87 years old, SD ± 13,26) with non proliferative R1 diabetic retinopathy (ETDRS level 20-35), referred from the English National Diabetic Screening Programme for M1 maculopahty were imaged with both non-stereo Fundus Photographs (FP) (Zeiss Visucam Pro NM) and Multicolor Spectralis (McS) (Heidelberg Engeneering) in an hospital based OCT virtual clinic. FP were graded as sharp, unsharp or ungradable. Ungradable FP were excluded. A retinal photographer (MC) and ophthalmologist (EC) graded in a masked fashion the presence of exudates on FP and McS. Cohen’s Kappa statistic was used to ascertain the level of agreement between the two imaging modalities and between observers. Fundus photographs and the ophthalmologist’s observations were set as gold standard for analysis purposes. A senior ophthalmologist (SP) more familiar with McS reviewed the disagreements and graded HE as being significant on FP if HE were ≥ than ETDRS Standard Photo 3.
Results:
McS and FP matched detection of HE in 83 eyes for the ophthalmologist observations. The mismatch were due in 14 cases to McS failing to identify HE and in 3 cases to McS detecting HE not visible on unsharp, lower quality FP. Interobserver agreement was highest for detection of any HE on fundus photos with K=0.71. Agreement between McS and FP achieved K=0.61 for the ophthalmologist and K=0.45 for the photographer. Senior ophthalmologist arbitration reduced the McS and FP disagreement to 4 cases, of which only 2 had significant HE. Confounding factors causing disagreement were questionable or dot HE, abnormal reflectance from ERM and unsharp FP.
Conclusions:
McS seems to need more clinically experienced observers to match detection of HE on FP. Conversely, assessment of HE on fundus photos seems to be less influenced by operator expertise. Correctly interpreting green reflectance helped in confirming presence of HE on McS. In some cases McS achieved sharper fundus images than FP. The use of McS as a single equipment to assess HE in the absence of fundus photographs in M1 maculopathy OCT virtual clinics needs further evaluation and validation.
Keywords: 499 diabetic retinopathy •
550 imaging/image analysis: clinical •
585 macula/fovea