1
views
0
recommends
+1 Recommend
0 collections
    0
    shares
      • Record: found
      • Abstract: found
      • Article: found
      Is Open Access

      Automated algorithms combining structure and function outperform general ophthalmologists in diagnosing glaucoma

      research-article

      Read this article at

      Bookmark
          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

          Abstract

          Purpose

          To test the ability of machine learning classifiers (MLCs) using optical coherence tomography (OCT) and standard automated perimetry (SAP) parameters to discriminate between healthy and glaucomatous individuals, and to compare it to the diagnostic ability of the combined structure-function index (CSFI), general ophthalmologists and glaucoma specialists.

          Design

          Cross-sectional prospective study.

          Methods

          Fifty eight eyes of 58 patients with early to moderate glaucoma (median value of the mean deviation = −3.44 dB; interquartile range, -6.0 to -2.4 dB) and 66 eyes of 66 healthy individuals underwent OCT and SAP tests. The diagnostic accuracy (area under the ROC curve—AUC) of 10 MLCs was compared to those obtained with the CSFI, 3 general ophthalmologists and 3 glaucoma specialists exposed to the same OCT and SAP data.

          Results

          The AUCs obtained with MLCs ranged from 0.805 (Classification Tree) to 0.931 (Radial Basis Function Network, RBF). The sensitivity at 90% specificity ranged from 51.6% (Classification Tree) to 82.8% (Bagging, Multilayer Perceptron and Support Vector Machine Gaussian). The CSFI had a sensitivity of 79.3% at 90% specificity, and the highest AUC (0.948). General ophthalmologists and glaucoma specialists’ grading had sensitivities of 66.2% and 83.8% at 90% specificity, and AUCs of 0.879 and 0.921, respectively. RBF (the best MLC), the CSFI, and glaucoma specialists showed significantly higher AUCs than that obtained by general ophthalmologists (P<0.05). However, there were no significant differences between the AUCs obtained by RBF, the CSFI, and glaucoma specialists (P>0.25).

          Conclusion

          Our findings suggest that both MLCs and the CSFI can be helpful in clinical practice and effectively improve glaucoma diagnosis in the primary eye care setting, when there is no glaucoma specialist available.

          Related collections

          Most cited references37

          • Record: found
          • Abstract: found
          • Article: not found

          LIBSVM: A library for support vector machines

          LIBSVM is a library for Support Vector Machines (SVMs). We have been actively developing this package since the year 2000. The goal is to help users to easily apply SVM to their applications. LIBSVM has gained wide popularity in machine learning and many other areas. In this article, we present all implementation details of LIBSVM. Issues such as solving SVM optimization problems theoretical convergence multiclass classification probability estimates and parameter selection are discussed in detail.
            Bookmark
            • Record: found
            • Abstract: found
            • Article: not found

            Primary Open-Angle Glaucoma Preferred Practice Pattern(®) Guidelines.

            PRIMARY OPEN-ANGLE GLAUCOMA PREFERRED PRACTICE PATTERN®
              Bookmark
              • Record: found
              • Abstract: found
              • Article: not found

              A framework for comparing structural and functional measures of glaucomatous damage.

              While it is often said that structural damage due to glaucoma precedes functional damage, it is not always clear what this statement means. This review has two purposes: first, to show that a simple linear relationship describes the data relating a particular functional test (standard automated perimetry (SAP)) to a particular structural test (optical coherence tomography (OCT)); and, second, to propose a general framework for relating structural and functional damage, and for evaluating if one precedes the other. The specific functional and structural tests employed are described in Section 2. To compare SAP sensitivity loss to loss of the retinal nerve fiber layer (RNFL) requires a map that relates local field regions to local regions of the optic disc as described in Section 3. When RNFL thickness in the superior and inferior arcuate sectors of the disc are plotted against SAP sensitivity loss (dB units) in the corresponding arcuate regions of the visual field, RNFL thickness becomes asymptotic for sensitivity losses greater than about 10dB. These data are well described by a simple linear model presented in Section 4. The model assumes that the RNFL thickness measured with OCT has two components. One component is the axons of the retinal ganglion cells and the other, the residual, is everything else (e.g. glial cells, blood vessels). The axon portion is assumed to decrease in a linear fashion with losses in SAP sensitivity (in linear units); the residual portion is assumed to remain constant. Based upon severe SAP losses in anterior ischemic optic neuropathy (AION), the residual RNFL thickness in the arcuate regions is, on average, about one-third of the premorbid (normal) thickness of that region. The model also predicts that, to a first approximation, SAP sensitivity in control subjects does not depend upon RNFL thickness. The data (Section 6) are, in general, consistent with this prediction showing a very weak correlation between RNFL thickness and SAP sensitivity. In Section 7, the model is used to estimate the proportion of patients showing statistical abnormalities (worse than the 5th percentile) on the OCT RNFL test before they show abnormalities on the 24-2 SAP field test. Ignoring measurement error, the patients with a relatively thick RNFL, when healthy, will be more likely to show significant SAP sensitivity loss before statistically significant OCT RNFL loss, while the reverse will be true for those who start with an average or a relatively thin RNFL when healthy. Thus, it is important to understand the implications of the wide variation in RNFL thickness among control subjects. Section 8 describes two of the factors contributing to this variation, variations in the position of blood vessels and variations in the mapping of field regions to disc sectors. Finally, in Sections 7 and 9, the findings are related to the general debate in the literature about the relationship between structural and functional glaucomatous damage and a framework is proposed for understanding what is meant by the question, 'Does structural damage precede functional damage in glaucoma?' An emphasis is placed upon the need to distinguish between "statistical" and "relational" meanings of this question.
                Bookmark

                Author and article information

                Contributors
                Role: ConceptualizationRole: Data curationRole: Formal analysisRole: Funding acquisitionRole: InvestigationRole: MethodologyRole: Project administrationRole: ResourcesRole: SoftwareRole: SupervisionRole: ValidationRole: VisualizationRole: Writing – original draftRole: Writing – review & editing
                Role: ConceptualizationRole: Data curationRole: Investigation
                Role: ConceptualizationRole: Data curationRole: Investigation
                Role: Formal analysisRole: InvestigationRole: VisualizationRole: Writing – original draftRole: Writing – review & editing
                Role: Formal analysisRole: InvestigationRole: SoftwareRole: Validation
                Role: ConceptualizationRole: Formal analysisRole: InvestigationRole: MethodologyRole: SoftwareRole: ValidationRole: Visualization
                Role: ConceptualizationRole: Formal analysisRole: InvestigationRole: SoftwareRole: ValidationRole: Writing – review & editing
                Role: Formal analysisRole: InvestigationRole: MethodologyRole: Software
                Role: InvestigationRole: MethodologyRole: ValidationRole: Writing – review & editing
                Role: ConceptualizationRole: Data curationRole: Formal analysisRole: Funding acquisitionRole: InvestigationRole: MethodologyRole: Project administrationRole: ResourcesRole: SoftwareRole: SupervisionRole: ValidationRole: VisualizationRole: Writing – original draftRole: Writing – review & editing
                Role: Editor
                Journal
                PLoS One
                PLoS ONE
                plos
                plosone
                PLoS ONE
                Public Library of Science (San Francisco, CA USA )
                1932-6203
                5 December 2018
                2018
                : 13
                : 12
                : e0207784
                Affiliations
                [1 ] Glaucoma Service, Department of Ophthalmology, University of Campinas, Campinas, São Paulo, Brazil
                [2 ] Department of Computer Engineering, Polytechnic School, University of São Paulo, São Paulo, São Paulo, Brazil
                [3 ] Department of Ophthalmology and Visual Sciences, Dalhousie University, Halifax, Nova Scotia, Canada
                [4 ] Duke University Eye Center, Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina, United States of America
                Texas A&M University, UNITED STATES
                Author notes

                Competing Interests: VPC received support from Novartis, Alcon Laboratories, União Quimica, Allergan, and Glaukos; and FAM received support from Novartis, Alcon Laboratories, Bausch & Lomb, Carl Zeiss Meditec, Heidelberg Engineering, Merck, Allergan, Sensimed, Topcon, Reichert, National Eye Institute and Patent - RGC index with royalties paid. There are no products in development or marketed products to declare. This does not alter our adherence to all the PLOS ONE policies on sharing data and materials.

                Author information
                http://orcid.org/0000-0001-7077-8086
                http://orcid.org/0000-0001-5302-7950
                Article
                PONE-D-18-12058
                10.1371/journal.pone.0207784
                6281287
                30517157
                6c458194-78a3-4b5c-aa05-4d634bdb4f04
                © 2018 Shigueoka 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 author and source are credited.

                History
                : 21 April 2018
                : 5 November 2018
                Page count
                Figures: 1, Tables: 3, Pages: 13
                Funding
                Funded by: funder-id http://dx.doi.org/10.13039/501100001807, Fundação de Amparo à Pesquisa do Estado de São Paulo;
                Award ID: 07/51281-9
                This work was supported by São Paulo Research Foundation, FAPESP (07/51281-9). Additional support provided by Novartis, Alcon Laboratories, Bausch & Lomb, Carl Zeiss Meditec, Heidelberg Engineering, Merck, Allergan, Sensimed, Topcon, Reichert, National Eye Institute and Patent - RGC index with royalties to FAM; and from Novartis, Alcon Laboratories, União Química, Allergan and Glaukos to VPC. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
                Categories
                Research Article
                Medicine and Health Sciences
                Ophthalmology
                Eye Diseases
                Glaucoma
                Biology and Life Sciences
                Anatomy
                Head
                Eyes
                Medicine and Health Sciences
                Anatomy
                Head
                Eyes
                Biology and Life Sciences
                Anatomy
                Ocular System
                Eyes
                Medicine and Health Sciences
                Anatomy
                Ocular System
                Eyes
                Medicine and Health Sciences
                Diagnostic Medicine
                Diagnostic Radiology
                Tomography
                Research and Analysis Methods
                Imaging Techniques
                Diagnostic Radiology
                Tomography
                Medicine and Health Sciences
                Radiology and Imaging
                Diagnostic Radiology
                Tomography
                Biology and Life Sciences
                Cell Biology
                Cellular Types
                Animal Cells
                Neurons
                Afferent Neurons
                Retinal Ganglion Cells
                Biology and Life Sciences
                Neuroscience
                Cellular Neuroscience
                Neurons
                Afferent Neurons
                Retinal Ganglion Cells
                Biology and Life Sciences
                Cell Biology
                Cellular Types
                Animal Cells
                Neurons
                Ganglion Cells
                Retinal Ganglion Cells
                Biology and Life Sciences
                Neuroscience
                Cellular Neuroscience
                Neurons
                Ganglion Cells
                Retinal Ganglion Cells
                Physical Sciences
                Mathematics
                Applied Mathematics
                Algorithms
                Machine Learning Algorithms
                Research and Analysis Methods
                Simulation and Modeling
                Algorithms
                Machine Learning Algorithms
                Computer and Information Sciences
                Artificial Intelligence
                Machine Learning
                Machine Learning Algorithms
                Physical Sciences
                Mathematics
                Applied Mathematics
                Algorithms
                Research and Analysis Methods
                Simulation and Modeling
                Algorithms
                Computer and Information Sciences
                Artificial Intelligence
                Machine Learning
                Support Vector Machines
                Computer and Information Sciences
                Artificial Intelligence
                Machine Learning
                Custom metadata
                There are no ethical or legal restrictions on sharing the de-identified data set from our research. Data does not contain any potentially identifying or sensitive patient information. The anonymized data set necessary to replicate our study findings was uploaded as a Supporting Information file.

                Uncategorized
                Uncategorized

                Comments

                Comment on this article