Proof-of-Principle Study

Discover how innovative digital photography and sclera color analysis formed the basis of our app. Our proof-of-principle study at UCLH demonstrated the potential of this non-invasive, accessible technique to improve early detection and treatment outcomes for newborns.


‘Yellowness’ chart being compared against the newborn.

  • Image Capture: Photographs of newborns’ eyes were taken using a Nikon D3200 camera with a 60 mm macro lens. Manual focus and auto exposure settings were used and the ISO was set to 1600. White balance was maintained for consistency.
  • Calibration: A customised color chart was used to ensure accurate color readings. Images were captured in RAW format to preserve tonal information.
  • Subjects: Of the 133 newborns photographed, 110 samples were used for analysis due to consistent lighting conditions. The sample included 67 white and 43 non-white infants.

Results

We found that using sclera color to predict TSB levels provided reasonably high correlations (r = 0.75, ρ = 0.72). The ROC curve analysis showed an area under the curve (AUC) of 0.87, indicating the method’s potential as a screening tool.

  • Ambient Lighting: We found that ambient lighting significantly affects prediction accuracy. Using images taken in controlled lighting conditions improved correlation results.
  • Comparisons: We compared our results with those from other techniques like transcutaneous bilirubinometers (TcB), showing that while correlations were slightly lower, the method’s non-contact nature and potential for smartphone integration offer significant advantages.

This study demonstrated that sclera color analysis using digital photography is a promising technique for neonatal jaundice screening. With further improvements in ambient lighting control and algorithm refinement, this method could become a low-cost, accessible tool for early jaundice detection, especially in low-resource settings.