Validating DevCheck: a semi-automated computer algorithm for measuring foveal immaturity in premature infants

Foveal immaturity, frequently associated with premature birth, is an established biomarker for retinopathy of prematurity (ROP) diagnoses. With further research, this biomarker may contribute to the development of an optical coherence tomography (OCT) ROP screening tool. Semi-automated foveal development analysis was validated on handheld swept source OCT images; however, this MATLAB program requires heavy user input and has not been optimized for commercially available spectral domain OCT (SD-OCT). The purpose of this study is to validate a newer program called "DevCheck", which has been optimized for handheld SD-OCT imaging performed on a cohort of awake premature infants at the time of routine ROP examinations.

Imaging was obtained at ROP screening sessions between 2015 and 2018 using the Envisu C2300 handheld SD-OCT. Two independent trained graders executed the novel Python algorithm, "DevCheck", at 36 weeks (34-38 weeks) postmenstrual age. One grader performed the previously validated MATLAB program on a subset of 10 images. The intraclass correlation coefficients (ICCs) were calculated to assess agreement between graders and between programs.

A total of 71 images from 47 awake premature infants (46.8% male; mean gestational age 27.99±2.71 weeks; mean birthweight 974.30±279.54 grams) were included. For intergrader agreement (n=71), the ICCs for inner and outer retinal thicknesses at the fovea were 0.91 [95% confidence interval (CI): 0.86-0.95] and 0.75 (95% CI: 0.60-0.84), respectively. At the parafovea, ICCs were 0.83 (95% CI: 0.72-0.89) and 0.81 (95% CI: 0.70-0.88), respectively. The ICC for foveal angle was 0.82 (95% CI: 0.71-0.89), indicating strong intergrader agreement across most parameters. For agreement between computer programs (n=10), the ICCs for inner and outer retinal thicknesses at the fovea were 0.98 (95% CI: 0.90-0.99) and 0.83 (95% CI: 0.29-0.96), respectively. At the parafovea they were 0.93 (95% CI: 0.73-0.98) and 0.48 (95% CI: -0.43 to 0.86), respectively. The foveal angle ICC was 0.95 (95% CI: 0.80-0.99). Using final DevCheck results, a secondary analysis for this study found older gestational age was significantly associated with decreasing inner retinal fovea/parafovea ratio (P=0.004). No other significant relationships were observed (P>0.10).

We developed DevCheck, a user-friendly Python algorithm for measuring foveal development from handheld SD-OCT. The tool demonstrated good to excellent agreement between graders and comparable results to a prior validated MATLAB program, except for outer retinal thickness at the parafovea. These findings support the feasibility of DevCheck for future research and clinical applications.