Expert vs. non expert grading of ROP from digital video images
Reviewed by Fiona Rowe

The primary purpose of this study was to determine whether digital video images of the retina obtained using an indirect ophthalmoscope imaging system could be accurately graded for zone and stage of retinopathy of prematurity (ROP) and the presence of plus or pre-plus disease. The secondary aim was to determine whether these digital retinal video images could be accurately graded to detect the presence of disease (type I ROP) requiring treatment by comparing two predefined criteria for referral. The first criterion for referral was the presence of pre-threshold disease, pre-plus disease or plus disease. The second criterion was the presence of only pre-plus or plus disease. The study included 114 infants with median gestation age of 26 weeks (23-33) and median birth weight of 840g (450-2300). Experts judged 60% of images to have fair or good image quality versus 55% of non-experts. Of these images, the experts and non-experts correctly identified zone (75 vs. 74%), stage of ROP (75 vs. 40%) and presence of pre-plus or plus disease (79%). Sensitivity of grading images for presence of pre-threshold, pre-plus or plus disease was 100% for experts and non-experts. Specificity was 77 and 82%. The non-experts had a more difficult time identifying the stage of ROP, particularly stage 3. Neither grader misclassified zone by more than one zone. The authors conclude that high quality retinal video images can be read with high sensitivity and specificity to screen for type I ROP. 

Evaluation of the accuracy of grading indirect ophthalmoscope video images for retinopathy of prematurity screening.
Prakalapakorn S, Wallace DK , Dollard RS, Freedman SF.
JOURNAL OF PEDIATRIC OPHTHALMOLOGY AND STRABISMUS
2015;52:85-92.