Noninvasive Machine Learning Screening Model for Dacryocystitis Based on Ocular Surface Indicators.
Journal
The Journal of craniofacial surgery
ISSN: 1536-3732
Titre abrégé: J Craniofac Surg
Pays: United States
ID NLM: 9010410
Informations de publication
Date de publication:
Historique:
pubmed:
17
7
2021
medline:
4
1
2022
entrez:
16
7
2021
Statut:
ppublish
Résumé
Dacryocystitis is an orbital disease that can be easily misdiagnosed. The most common diagnostic tools for dacryocystitis are computed tomography, lacrimal duct angiography, and lacrimal tract irrigation. Yet, those are invasive methods, which are not conducive to extensive screening. To explore the significance of ocular surface indicators and demographic data in the screening of dacryocystitis. Data were prospectively collected from 56 patients with dacryocystitis (56 eyes) and 56 healthy individuals. Collected indicators included demographic information (gender, age), ocular surface data of tear meniscus height, objective scatter index (OSI), and clinical diagnosis. The model features were screened out by machine learning to establish a dacryocystitis screening model. Tear meniscus height, OSI_maximum Lyapunov exponent, basic OSI, median of OSI, mean of OSI, slope coefficient of OSI linear regression, coefficient of variation in OSI, interquartile range of OSI, and other 8 parameters were used as model parameters to establish a dacryocystitis screening model with an overall detection accuracy of 85.71%. This new screening model that is based on ocular surface indicators provides a new option for noninvasive screening of dacryocystitis.
Sections du résumé
BACKGROUND
BACKGROUND
Dacryocystitis is an orbital disease that can be easily misdiagnosed. The most common diagnostic tools for dacryocystitis are computed tomography, lacrimal duct angiography, and lacrimal tract irrigation. Yet, those are invasive methods, which are not conducive to extensive screening.
OBJECTIVE
OBJECTIVE
To explore the significance of ocular surface indicators and demographic data in the screening of dacryocystitis.
MATERIALS AND METHODS
METHODS
Data were prospectively collected from 56 patients with dacryocystitis (56 eyes) and 56 healthy individuals. Collected indicators included demographic information (gender, age), ocular surface data of tear meniscus height, objective scatter index (OSI), and clinical diagnosis. The model features were screened out by machine learning to establish a dacryocystitis screening model.
RESULTS
RESULTS
Tear meniscus height, OSI_maximum Lyapunov exponent, basic OSI, median of OSI, mean of OSI, slope coefficient of OSI linear regression, coefficient of variation in OSI, interquartile range of OSI, and other 8 parameters were used as model parameters to establish a dacryocystitis screening model with an overall detection accuracy of 85.71%.
CONCLUSIONS
CONCLUSIONS
This new screening model that is based on ocular surface indicators provides a new option for noninvasive screening of dacryocystitis.
Identifiants
pubmed: 34267140
doi: 10.1097/SCS.0000000000007863
pii: 00001665-900000000-92418
doi:
Types de publication
Journal Article
Langues
eng
Pagination
e23-e28Informations de copyright
Copyright © 2021 by Mutaz B. Habal, MD.
Déclaration de conflit d'intérêts
The authors report no conflicts of interest.
Références
Chung SY, Rafailov L, Turbin RE, et al. The microbiologic profile of dacryocystitis. Orbit 2019; 38:72–78.
Li EY, Wong ES, Wong AC, et al. Primary vs secondary endoscopic dacryocystorhinostomy for acute dacryocystitis with lacrimal sac abscess formation: a randomized clinical trial. JAMA Ophthalmol 2017; 135:1361–1366.
Baek JS, Jeong SH, Lee JH, et al. Cause and management of patients with failed endonasal dacryocystorhinostomy. Clin Exp Otorhinolaryngol 2017; 10:85–90.
Asheim J, Spickler E. CT demonstration of dacryolithiasis complicated by dacryocystitis. AJNR Am J Neuroradiol 2005; 26:2640–2641.
Ali MJ, Joshi SD, Naik MN, et al. Clinical profile and management outcome of acute dacryocystitis: two decades of experience in a tertiary eye care center. Semin Ophthalmol 2015; 30:118–123.
Kawakita T. Regeneration of lacrimal gland function to maintain the health of the ocular surface. Invest Ophthalmol Vis Sci 2018; 59:DES169–DES173.
Garreis F, Jahn J, Wild K, et al. Expression and regulation of S100 fused-type protein hornerin at the ocular surface and lacrimal apparatus. Invest Ophthalmol Vis Sci 2017; 58:5968–5977.
Ben Saida A. Noisy chaos in intraday financial data: evidence from the American index. Appl Math Comput 2014; 226:258–265.
Ben Saida A, Litimi H. High level chaos in the exchange and index markets. Chaos Solitons Fract 2013; 54:90–95.
Galliot F, Patel SR, Cochener B. Objective scatter index: working toward a new quantification of cataract? Refract Surg 2016; 32:96–102.
Song X, Langenbucher A, Gatzioufas Z, et al. Effect of biometric characteristics on the change of biomechanical properties of the human cornea due to cataract surgery. Biomed Res Int 2014; 2014:628019.
Gouvea L, Waring GO 4th, Brundrett A, et al. Objective assessment of optical quality in dry eye disease using a double-pass imaging system. Clin Ophthalmol 2019; 13:1991–1996.
Pan AP, Wang QM, Huang F, et al. Correlation among lens opacities classification system III grading, visual function index-14, pentacam nucleus staging, and objective scatter index for cataract assessment. Am J Ophthalmol 2015; 159:241.e2–247.e2.
Sun L, Kong X, Xu J, et al. A hybrid gene selection method based on ReliefF and ant colony optimization algorithm for tumor classification. Sci Rep 2019; 9:8978.
Mookiah MR, Acharya UR, Koh JE, et al. Automated diagnosis of age-related macular degeneration using greyscale features from digital fundus images. Comput Biol Med 2014; 53:55–64.
Singh A, Dutta MK, ParthaSarathi M, et al. Image processing based automatic diagnosis of glaucoma using wavelet features of segmented optic disc from fundus image. Comput Methods Programs Biomed 2016; 124:108–120.
Chen W, Wang Y, Cao G, et al. A random forest model based classification scheme for neonatal amplitude-integrated EEG. Biomed Eng Online 2014; 13: (Suppl 2): S4.