The impact of ear growth on identification rates using an ear biometric system in young infants
Files
Published version
Date
Authors
Simukanga, Ainani
Kobayashi, Misaki
Etter, Lauren
Qin, Wenda
Pieciak, Rachel
Albuquerque, Duarte
Chen, Yu-Jen
Betke, Margrit
MacLeod, William
Phiri, Jackson
Version
Published version
OA Version
Citation
A. Simukanga, M. Kobayashi, L. Etter, W. Qin, R. Pieciak, D. Albuquerque, Y.-.J. Chen, M. Betke, W. MacLeod, J. Phiri, L. Mwananyanda, C.J. Gill. "The impact of ear growth on identification rates using an ear biometric system in young infants." Gates Open Research, Volume 5, pp. 179 - 179. https://doi.org/10.12688/gatesopenres.13459.1
Abstract
BACKGROUND: Accurate patient identification is essential for delivering longitudinal care. Our team developed an ear biometric system (SEARCH) to improve patient identification. To address how ear growth affects matching rates longitudinally, we constructed an infant cohort, obtaining ear image sets monthly to map a 9-month span of observations. This analysis had three main objectives: 1) map trajectory of ear growth during the first 9 months of life; 2) determine the impact of ear growth on matching accuracy; and 3) explore computer vision techniques to counter a loss of accuracy. METHODOLOGY: Infants were enrolled from an urban clinic in Lusaka, Zambia. Roughly half were enrolled at their first vaccination visit and ~half at their last vaccination. Follow-up visits for each patient occurred monthly for 6 months. At each visit, we collected four images of the infant’s ears, and the child’s weight. We analyze ear area versus age and change in ear area versus age. We conduct pair-wise comparisons for all age intervals.
RESULTS: From 227 enrolled infants we acquired age-specific datasets for 6 days through 9 months. Maximal ear growth occurred between 6 days and 14 weeks. Growth was significant until 6 months of age, after which further growth appeared minimal. Examining look-back performance to the 6-month visit, baseline pair-wise comparisons yielded identification rates that ranged 46.9–75%. Concatenating left and right ears per participant improved identification rates to 61.5–100%. Concatenating images captured on adjacent visits further improved identification rates to 90.3–100%. Lastly, combining these two approaches improved identification to 100%. All matching strategies showed the weakest matching rates during periods of maximal growth (i.e., <6 months).
CONCLUSION: By quantifying the effect that ear growth has on performance of the SEARCH platform, we show that ear identification is a feasible solution for patient identification in an infant population 6 months and above.
Description
License
Copyright: © 2021 Simukanga A et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.