ABSTRACT
BACKGROUND/AIMS: Trachoma programs base treatment decisions on the community prevalence of the clinical signs of trachoma, assessed by direct examination of the conjunctiva. Automated assessment could be more standardized and more cost-effective. We tested the hypothesis that an automated algorithm could classify eyelid photographs better than chance. METHODS: A total of 1,656 field-collected conjunctival images were obtained from clinical trial participants in Niger and Ethiopia. Images were scored for trachomatous inflammation-follicular (TF) and trachomatous inflammation-intense (TI) according to the simplified World Health Organization grading system by expert raters. We developed an automated procedure for image enhancement followed by application of a convolutional neural net classifier for TF and separately for TI. One hundred images were selected for testing TF and TI, and these images were not used for training. RESULTS: The agreement score for TF and TI tasks for the automated algorithm relative to expert graders was κ = 0.44 (95% CI: 0.26 to 0.62, P < 0.001) and κ = 0.69 (95% CI: 0.55 to 0.84, P < 0.001), respectively. DISCUSSION: For assessing the clinical signs of trachoma, a convolutional neural net performed well above chance when tested against expert consensus. Further improvements in specificity may render this method suitable for field use.
Subject(s)
Conjunctiva/diagnostic imaging , Eyelids/diagnostic imaging , Image Processing, Computer-Assisted/methods , Neural Networks, Computer , Photography , Trachoma/diagnostic imaging , Algorithms , Artificial Intelligence , Ethiopia/epidemiology , Humans , Niger/epidemiology , Photography/methods , Prevalence , Trachoma/epidemiologyABSTRACT
AbstractWe evaluated a new trachoma scarring ranking system with potential use in clinical research. The upper right tarsal conjunctivas of 427 individuals from Ethiopian villages with hyperendemic trachoma were photographed. An expert grader first assigned a scar grade to each photograph using the 1981 World Health Organization (WHO) grading system. Then, all photographs were ranked from least (rank = 1) to most scarring (rank = 427). Photographic grading found 79 (18.5%) conjunctivae without scarring (C0), 191 (44.7%) with minimal scarring (C1), 105 (24.6%) with moderate scarring (C2), and 52 (12.2%) with severe scarring (C3). The ranking method demonstrated good internal validity, exhibiting a monotonic increase in the median rank across the levels of the 1981 WHO grading system. Intrarater repeatability was better for the ranking method (intraclass correlation coefficient = 0.84, 95% CI = 0.74-0.94). Exhibiting better internal and external validity, this ranking method may be useful for evaluating the difference in scarring between groups of individuals.
