Purpose: Fewer than 40% of children in the crucial younger-than-4 age group
are evaluated for visual problems by pediatricians. This is due to impract
icality from either a clinical or practice efficiency standpoint. Current p
hotoscreening methods require trained readers and suffer from significant s
ubjectivity and interobserver variability. We report a cross-sectional, dou
ble-masked study using new digital imaging with objective, automated, compu
terized image analysis.
Methods: Two-hundred six children aged 9 months to 16 years were prospectiv
ely studied in a University-based pediatric ophthalmology practice. images
were taken by volunteers with a modified digital camera which, when downloa
ded, were analyzed within 35 seconds by new image analysis software. The an
alysis was compared to a masked review of a complete pediatric ophthalmic e
xam.
Results: Overall agreement between physician and the objective computerized
analysis was 86.9%. Positive predictive value was 91%, sensitivity was 89%
, and specificity was 83%.
Conclusions: This automated digital imaging screening system eliminates hum
an bias and provides accurate and immediate results. The system requires no
special expertise.