Using Deep Learning to Automate Goldmann Applanation Tonometry Readings.
Journal
Ophthalmology
ISSN: 1549-4713
Titre abrégé: Ophthalmology
Pays: United States
ID NLM: 7802443
Informations de publication
Date de publication:
11 2020
11 2020
Historique:
received:
11
02
2020
revised:
16
04
2020
accepted:
20
04
2020
pubmed:
29
4
2020
medline:
23
12
2020
entrez:
29
4
2020
Statut:
ppublish
Résumé
To develop an objective and automated method for measuring intraocular pressure using deep learning and fixed-force Goldmann applanation tonometry (GAT) techniques. Prospective cross-sectional study. Patients from an academic glaucoma practice. Intraocular pressure was estimated by analyzing videos recorded using a standard slit-lamp microscope and fixed-force GAT. Video frames were labeled to identify the outline of the reference tonometer and the applanation mires. A deep learning model was trained to localize and segment the tonometer and mires. Intraocular pressure values were calculated from the deep learning-predicted tonometer and mire diameters using the Imbert-Fick formula. A separate test set was collected prospectively in which standard and automated GAT measurements were collected in random order by 2 independent masked observers to assess the deep learning model as well as interobserver variability. Intraocular pressure measurements between standard and automated methods were compared. Two hundred sixty-three eyes of 135 patients were included in the training and validation videos. For the test set, 50 eyes from 25 participants were included. Each eye was measured by 2 observers, resulting in 100 videos. Within the test set, the mean difference between automated and standard GAT results was -0.9 mmHg (95% limits of agreement [LoA], -5.4 to 3.6 mmHg). Mean difference between the 2 observers using standard GAT was 0.09 mmHg (LoA,-3.8 to 4.0 mmHg). Mean difference between the 2 observers using automated GAT videos was -0.3 mmHg (LoA, -4.1 to 3.5 mmHg). The coefficients of repeatability for automated and standard GAT were 3.8 and 3.9 mmHg, respectively. The bias for even-numbered measurements was reduced when using automated GAT. Preliminary measurements using deep learning to automate GAT demonstrate results comparable with those of standard GAT. Automated GAT has the potential to improve on our current GAT measurement standards significantly by reducing bias and improving repeatability. In addition, ocular pulse amplitudes could be observed using this technique.
Identifiants
pubmed: 32344074
pii: S0161-6420(20)30412-7
doi: 10.1016/j.ophtha.2020.04.033
pmc: PMC7606368
mid: NIHMS1587910
pii:
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
1498-1506Subventions
Organisme : NEI NIH HHS
ID : K23 EY029246
Pays : United States
Informations de copyright
Copyright © 2020 American Academy of Ophthalmology. All rights reserved.
Références
Ophthalmology. 2012 Aug;119(8):1552-7
pubmed: 22578443
J Glaucoma. 2019 Apr;28(4):302-306
pubmed: 30946710
Acta Ophthalmol (Copenh). 1960;38:548-61
pubmed: 13787537
Br J Ophthalmol. 2020 Oct;104(10):1394-1398
pubmed: 31871048
Control Clin Trials. 1994 Aug;15(4):299-325
pubmed: 7956270
Graefes Arch Clin Exp Ophthalmol. 2008 Apr;246(4):559-65
pubmed: 18193267
Bull Mem Soc Fr Ophtalmol. 1954;67:474-7; discussion, 477-8
pubmed: 13284610
Trans Am Acad Ophthalmol Otolaryngol. 1968 Jan-Feb;72(1):112-4
pubmed: 5637509
J Glaucoma. 2013 Dec;22(9):736-9
pubmed: 23708422
Br J Ophthalmol. 2005 Dec;89(12):1572-5
pubmed: 16299132
Clin Exp Optom. 2008 Nov;91(6):524-9
pubmed: 18651843
Ophthalmol Glaucoma. 2019 Nov - Dec;2(6):429-434
pubmed: 32672576
Eye (Lond). 2020 Feb;34(2):360-365
pubmed: 31399703
Curr Eye Res. 2011 Aug;36(8):727-32
pubmed: 21780922
J Glaucoma. 2020 Feb;29(2):124-126
pubmed: 31764580
Ophthalmology. 1999 Apr;106(4):653-62
pubmed: 10201583
Arch Ophthalmol. 1999 May;117(5):573-83
pubmed: 10326953
Acta Ophthalmol. 2014 Jun;92(4):e280-5
pubmed: 24456194