Abstract
Purpose :
While the decision to treat retinopathy of prematurity (ROP) usually depends on the presence of plus disease, its diagnosis is subjective even among experts. ROPtool is a semi-automated computer program that objectively measures vascular characteristics in retinal images. A previous study found that ROPtool tortuosity indices can help predict which infants need treatment for ROP, and that mean tortuosity had the highest predictive value at 38 weeks postmenstrual age (PMA). The purpose of this retrospective study was to analyze retinal images and quantitatively describe longitudinal changes in retinal vascular characteristics in infants treated compared to those not treated for ROP.
Methods :
We analyzed previously collected retinal images that were prospectively taken to evaluate the feasibility and accuracy of using narrow-field imaging for ROP screening. We used ROPtool to analyze ≤2 blood vessels per quadrant in the images. Including only left eyes, we compared the following ROPtool indices by PMA (weeks) between infants treated versus not treated for ROP: tortuosity index (TI), dilation index (DI), combination dilation/tortuosity indices: sum of adjusted indices (SAI) and tortuosity-weighted plus (TWP). Analysis was performed using the highest blood vessel and quadrant, and overall eye values.
Results :
Using ROPtool, we analyzed images of left eyes from 436 imaging sessions (99 infants; 8 (8.1%) treated); average number of imaging sessions per infant=4.4 (range: 1-16); 397 (91.1%) had 3-4 analyzable quadrants. The following indices were statistically significantly higher (p<0.04) for treated versus untreated infants: At 32 weeks PMA, TWP on the eye and blood vessel levels and SAI on the quadrant and blood vessel levels; For PMA=33-37, all 4 ROPtool indices (TI, DI, SAI, TWP) on all levels (eye, quadrant, blood vessel) (Figure 1). By 3-4 weeks prior to treatment, all 4 ROPtool indices were statistically significantly higher in infants destined to require ROP treatment than those who never needed treatment.
Conclusions :
We found that ROPtool analysis of both tortuosity and dilation measurements can help predict eventual need for treatment, at as early as 32 weeks PMA.
This is a 2020 ARVO Annual Meeting abstract.