Glaucomatous patterns in Frequency Doubling Technology (FDT) perimetry data identified by unsupervised machine learning classifiers

Christopher Bowd, Robert N. Weinreb, Madhusudhanan Balasubramanian, Intae Lee, Giljin Jang, Siamak Yousefi, Linda M. Zangwill, Felipe A. Medeiros, Christopher A. Girkin, Jeffrey M. Liebmann, Michael H. Goldbaum

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

Purpose: The variational Bayesian independent component analysis-mixture model (VIM), an unsupervised machine-learning classifier, was used to automatically separate Matrix Frequency Doubling Technology (FDT) perimetry data into clusters of healthy and glaucomatous eyes, and to identify axes representing statistically independent patterns of defect in the glaucoma clusters. Methods: FDT measurements were obtained from 1,190 eyes with normal FDT results and 786 eyes with abnormal FDT results from the UCSD-based Diagnostic Innovations in Glaucoma Study (DIGS) and African Descent and Glaucoma Evaluation Study (ADAGES). For all eyes, VIM input was 52 threshold test points from the 24-2 test pattern, plus age. Results: FDT mean deviation was -1.00 dB (S.D. = 2.80 dB) and -5.57 dB (S.D. = 5.09 dB) in FDT-normal eyes and FDT-abnormal eyes, respectively (p<0.001). VIM identified meaningful clusters of FDT data and positioned a set of statistically independent axes through the mean of each cluster. The optimal VIM model separated the FDT fields into 3 clusters. Cluster N contained primarily normal fields (1109/1190, specificity 93.1%) and clusters G1 and G2 combined, contained primarily abnormal fields (651/786, sensitivity 82.8%). For clusters G 1 and G2 the optimal number of axes were 2 and 5, respectively. Patterns automatically generated along axes within the glaucoma clusters were similar to those known to be indicative of glaucoma. Fields located farther from the normal mean on each glaucoma axis showed increasing field defect severity. Conclusions: VIM successfully separated FDT fields from healthy and glaucoma eyes without a priori information about class membership, and identified familiar glaucomatous patterns of loss.

Original languageEnglish (US)
Article numbere85941
JournalPloS one
Volume9
Issue number1
DOIs
StatePublished - Jan 30 2014

Fingerprint

Visual Field Tests
artificial intelligence
Learning systems
glaucoma
Classifiers
Technology
Glaucoma
eyes
Unsupervised Machine Learning
Defects
Independent component analysis
Innovation
testing

All Science Journal Classification (ASJC) codes

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)

Cite this

Glaucomatous patterns in Frequency Doubling Technology (FDT) perimetry data identified by unsupervised machine learning classifiers. / Bowd, Christopher; Weinreb, Robert N.; Balasubramanian, Madhusudhanan; Lee, Intae; Jang, Giljin; Yousefi, Siamak; Zangwill, Linda M.; Medeiros, Felipe A.; Girkin, Christopher A.; Liebmann, Jeffrey M.; Goldbaum, Michael H.

In: PloS one, Vol. 9, No. 1, e85941, 30.01.2014.

Research output: Contribution to journalArticle

Bowd, C, Weinreb, RN, Balasubramanian, M, Lee, I, Jang, G, Yousefi, S, Zangwill, LM, Medeiros, FA, Girkin, CA, Liebmann, JM & Goldbaum, MH 2014, 'Glaucomatous patterns in Frequency Doubling Technology (FDT) perimetry data identified by unsupervised machine learning classifiers', PloS one, vol. 9, no. 1, e85941. https://doi.org/10.1371/journal.pone.0085941
Bowd, Christopher ; Weinreb, Robert N. ; Balasubramanian, Madhusudhanan ; Lee, Intae ; Jang, Giljin ; Yousefi, Siamak ; Zangwill, Linda M. ; Medeiros, Felipe A. ; Girkin, Christopher A. ; Liebmann, Jeffrey M. ; Goldbaum, Michael H. / Glaucomatous patterns in Frequency Doubling Technology (FDT) perimetry data identified by unsupervised machine learning classifiers. In: PloS one. 2014 ; Vol. 9, No. 1.
@article{4da15061f921475da17e4fa3c1b96902,
title = "Glaucomatous patterns in Frequency Doubling Technology (FDT) perimetry data identified by unsupervised machine learning classifiers",
abstract = "Purpose: The variational Bayesian independent component analysis-mixture model (VIM), an unsupervised machine-learning classifier, was used to automatically separate Matrix Frequency Doubling Technology (FDT) perimetry data into clusters of healthy and glaucomatous eyes, and to identify axes representing statistically independent patterns of defect in the glaucoma clusters. Methods: FDT measurements were obtained from 1,190 eyes with normal FDT results and 786 eyes with abnormal FDT results from the UCSD-based Diagnostic Innovations in Glaucoma Study (DIGS) and African Descent and Glaucoma Evaluation Study (ADAGES). For all eyes, VIM input was 52 threshold test points from the 24-2 test pattern, plus age. Results: FDT mean deviation was -1.00 dB (S.D. = 2.80 dB) and -5.57 dB (S.D. = 5.09 dB) in FDT-normal eyes and FDT-abnormal eyes, respectively (p<0.001). VIM identified meaningful clusters of FDT data and positioned a set of statistically independent axes through the mean of each cluster. The optimal VIM model separated the FDT fields into 3 clusters. Cluster N contained primarily normal fields (1109/1190, specificity 93.1{\%}) and clusters G1 and G2 combined, contained primarily abnormal fields (651/786, sensitivity 82.8{\%}). For clusters G 1 and G2 the optimal number of axes were 2 and 5, respectively. Patterns automatically generated along axes within the glaucoma clusters were similar to those known to be indicative of glaucoma. Fields located farther from the normal mean on each glaucoma axis showed increasing field defect severity. Conclusions: VIM successfully separated FDT fields from healthy and glaucoma eyes without a priori information about class membership, and identified familiar glaucomatous patterns of loss.",
author = "Christopher Bowd and Weinreb, {Robert N.} and Madhusudhanan Balasubramanian and Intae Lee and Giljin Jang and Siamak Yousefi and Zangwill, {Linda M.} and Medeiros, {Felipe A.} and Girkin, {Christopher A.} and Liebmann, {Jeffrey M.} and Goldbaum, {Michael H.}",
year = "2014",
month = "1",
day = "30",
doi = "10.1371/journal.pone.0085941",
language = "English (US)",
volume = "9",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "1",

}

TY - JOUR

T1 - Glaucomatous patterns in Frequency Doubling Technology (FDT) perimetry data identified by unsupervised machine learning classifiers

AU - Bowd, Christopher

AU - Weinreb, Robert N.

AU - Balasubramanian, Madhusudhanan

AU - Lee, Intae

AU - Jang, Giljin

AU - Yousefi, Siamak

AU - Zangwill, Linda M.

AU - Medeiros, Felipe A.

AU - Girkin, Christopher A.

AU - Liebmann, Jeffrey M.

AU - Goldbaum, Michael H.

PY - 2014/1/30

Y1 - 2014/1/30

N2 - Purpose: The variational Bayesian independent component analysis-mixture model (VIM), an unsupervised machine-learning classifier, was used to automatically separate Matrix Frequency Doubling Technology (FDT) perimetry data into clusters of healthy and glaucomatous eyes, and to identify axes representing statistically independent patterns of defect in the glaucoma clusters. Methods: FDT measurements were obtained from 1,190 eyes with normal FDT results and 786 eyes with abnormal FDT results from the UCSD-based Diagnostic Innovations in Glaucoma Study (DIGS) and African Descent and Glaucoma Evaluation Study (ADAGES). For all eyes, VIM input was 52 threshold test points from the 24-2 test pattern, plus age. Results: FDT mean deviation was -1.00 dB (S.D. = 2.80 dB) and -5.57 dB (S.D. = 5.09 dB) in FDT-normal eyes and FDT-abnormal eyes, respectively (p<0.001). VIM identified meaningful clusters of FDT data and positioned a set of statistically independent axes through the mean of each cluster. The optimal VIM model separated the FDT fields into 3 clusters. Cluster N contained primarily normal fields (1109/1190, specificity 93.1%) and clusters G1 and G2 combined, contained primarily abnormal fields (651/786, sensitivity 82.8%). For clusters G 1 and G2 the optimal number of axes were 2 and 5, respectively. Patterns automatically generated along axes within the glaucoma clusters were similar to those known to be indicative of glaucoma. Fields located farther from the normal mean on each glaucoma axis showed increasing field defect severity. Conclusions: VIM successfully separated FDT fields from healthy and glaucoma eyes without a priori information about class membership, and identified familiar glaucomatous patterns of loss.

AB - Purpose: The variational Bayesian independent component analysis-mixture model (VIM), an unsupervised machine-learning classifier, was used to automatically separate Matrix Frequency Doubling Technology (FDT) perimetry data into clusters of healthy and glaucomatous eyes, and to identify axes representing statistically independent patterns of defect in the glaucoma clusters. Methods: FDT measurements were obtained from 1,190 eyes with normal FDT results and 786 eyes with abnormal FDT results from the UCSD-based Diagnostic Innovations in Glaucoma Study (DIGS) and African Descent and Glaucoma Evaluation Study (ADAGES). For all eyes, VIM input was 52 threshold test points from the 24-2 test pattern, plus age. Results: FDT mean deviation was -1.00 dB (S.D. = 2.80 dB) and -5.57 dB (S.D. = 5.09 dB) in FDT-normal eyes and FDT-abnormal eyes, respectively (p<0.001). VIM identified meaningful clusters of FDT data and positioned a set of statistically independent axes through the mean of each cluster. The optimal VIM model separated the FDT fields into 3 clusters. Cluster N contained primarily normal fields (1109/1190, specificity 93.1%) and clusters G1 and G2 combined, contained primarily abnormal fields (651/786, sensitivity 82.8%). For clusters G 1 and G2 the optimal number of axes were 2 and 5, respectively. Patterns automatically generated along axes within the glaucoma clusters were similar to those known to be indicative of glaucoma. Fields located farther from the normal mean on each glaucoma axis showed increasing field defect severity. Conclusions: VIM successfully separated FDT fields from healthy and glaucoma eyes without a priori information about class membership, and identified familiar glaucomatous patterns of loss.

UR - http://www.scopus.com/inward/record.url?scp=84900432666&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84900432666&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0085941

DO - 10.1371/journal.pone.0085941

M3 - Article

VL - 9

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 1

M1 - e85941

ER -