Preliminary assessment of three protocols for the screening of amblyopia through Monte Carlo simulation

Shand, Keely and Boribalburephan, Atirut and Giardini, Mario Ettore; (2024) Preliminary assessment of three protocols for the screening of amblyopia through Monte Carlo simulation. In: Proceedings of the IEEE International Instrumentation and Measurement Technology Conference (I2MTC) 2024. IEEE, GBR. (In Press)

[thumbnail of Shand-etal-I2MTC-2024-Preliminary-assessment-of-three-protocols-for-the-screening-of-amblyopia-through-Monte-Carlo-simulation] Text. Filename: Shand-etal-I2MTC-2024-Preliminary-assessment-of-three-protocols-for-the-screening-of-amblyopia-through-Monte-Carlo-simulation.pdf
Accepted Author Manuscript
Restricted to Repository staff only until 1 January 2099.
License: Strathprints license 1.0

Download (1MB) | Request a copy

Abstract

Amblyopia is a neurodevelopmental disorder that causes irreversible vision loss in one eye. In order to be effective, treatment needs to start as early as possible, ideally in early infancy. The screening for amblyopia assesses the disparity of the child’s visual acuity (VA). In pre-verbal children, this screening is performed using a preferential looking test. These testing protocols rely on a subjective estimate by the test operator of the child’s attention, and this estimate is subject to errors. Little is understood of the quantitative impact of error rates on screening performance. In this paper, a Monte Carlo simulation to compare the clinical performance of three preferential-looking test protocols for paediatric VA screening tests for amblyopia is described. The inter-protocol differences of “Cardiff Acuity Test” (CAT), “Keeler Acuity Cards for Infants” (KACI) and “Teller Acuity Cards” (TAC) are assessed by iteratively executing through a simulation loop of an examiner testing a test subject using the three set protocols. The measured VA from each protocol and the actual VA have been compared using Bland-Altman statistics. It was determined that CAT and KACI both have a systematic bias, whereby they measure the VA at a greater logMAR value than the smallest testing resolution. KACI bias is greater due to the greater step size used.