Novel image recognition software to improve the identification of trichiasis cases

Primary research question:

• Can a smartphone-based app with an integrated image analysis algorithm be used to increase the accuracy and yield of trachomatous trichiasis (TT) screening, compared to standard TT case finder screening?

Secondary research questions:

• Can a smartphone-based app be used to identify houses or neighbourhoods that TT case finders miss during TT case finding activities? If so, are these the most neglected households?
• Are individuals who live further from the surgery camp site less likely to present for TT surgery?

Trachomatous trichiasis (TT) is on target to be the first blinding condition to be eliminated as a public health problem. Substantial progress has been made, with nine countries already declaring elimination. With progress comes new challenges, both for reaching this goal and for demonstrating sustained elimination. As the number of TT cases declines, those needing services become harder to find.

House-to-house searches in remote communities are quickly becoming the global standard for TT case identification. Most programmes utilise local community members to serve as TT case finders for these searches. However, poor case identification and limited screening time are significant challenges with this approach.

This study focuses on this important public health problem by improving the way in which we can identify people who may need TT surgery and evaluate case finding coverage. The foundation of this research project is a new smartphone-based app that integrates automated analysis of eyelid images to identify possible TT cases and to track their progress in receiving surgery. This app ultimately will have the ability to reach neglected populations most at need for TT screening. The app will build upon the existing TT Tracker app that is utilized in >50 countries worldwide to track patients longitudinally from surgical intervention through follow-up.

If the study confirms that the approach is successful, it will be easily scalable, since it piggybacks on the TT Tracker, which is already deployed globally. The app will be made available through standard app stores, for immediate download worldwide.