Glaucoma is a progressive optic neuropathy with characteristic structural changes in the optic nerve head reflected in the visual field. The visual-field sensitivity test is commonly used in a ...clinical setting to evaluate glaucoma. Standard automated perimetry (SAP) is a common computerized visual-field test whose output is amenable to machine learning. We compared the performance of a number of machine learning algorithms with STATPAC indexes mean deviation, pattern standard deviation, and corrected pattern standard deviation. The machine learning algorithms studied included multilayer perceptron (MLP), support vector machine (SVM), and linear (LDA) and quadratic discriminant analysis (QDA), Parzen window, mixture of Gaussian (MOG), and mixture of generalized Gaussian (MGG). MLP and SVM are classifiers that work directly on the decision boundary and fall under the discriminative paradigm. Generative classifiers, which first model the data probability density and then perform classification via Bayes' rule, usually give deeper insight into the structure of the data space. We have applied MOG, MGG, LDA, QDA, and Parzen window to the classification of glaucoma from SAP. Performance of the various classifiers was compared by the areas under their receiver operating characteristic curves and by sensitivities (true-positive rates) at chosen specificities (true-negative rates). The machine-learning-type classifiers showed improved performance over the best indexes from STATPAC. Forward-selection and backward-elimination methodology further improved the classification rate and also has the potential to reduce testing time by diminishing the number of visual-field location measurements.
To determine whether neural network techniques can improve differentiation between glaucomatous and nonglaucomatous eyes, using the optic disc topography parameters of the Heidelberg Retina Tomograph ...(HRT; Heidelberg Engineering, Heidelberg, Germany).
With the HRT, one eye was imaged from each of 108 patients with glaucoma (defined as having repeatable visual field defects with standard automated perimetry) and 189 subjects without glaucoma (no visual field defects with healthy-appearing optic disc and retinal nerve fiber layer on clinical examination) and the optic nerve topography was defined by 17 global and 66 regional HRT parameters. With all the HRT parameters used as input, receiver operating characteristic (ROC) curves were generated for the classification of eyes, by three neural network techniques: linear and Gaussian support vector machines (SVM linear and SVM Gaussian, respectively) and a multilayer perceptron (MLP), as well as four previously proposed linear discriminant functions (LDFs) and one LDF developed on the current data with all HRT parameters used as input.
The areas under the ROC curves for SVM linear and SVM Gaussian were 0.938 and 0.945, respectively; for MLP, 0.941; for the current LDF, 0.906; and for the best previously proposed LDF, 0.890. With the use of forward selection and backward elimination optimization techniques, the areas under the ROC curves for SVM Gaussian and the current LDF were increased to approximately 0.96.
Trained neural networks, with global and regional HRT parameters used as input, improve on previously proposed HRT parameter-based LDFs for discriminating between glaucomatous and nonglaucomatous eyes. The performance of both neural networks and LDFs can be improved with optimization of the features in the input. Neural network analyses show promise for increasing diagnostic accuracy of tests for glaucoma.
To determine which machine learning classifier learns best to interpret standard automated perimetry (SAP) and to compare the best of the machine classifiers with the global indices of STATPAC 2 and ...with experts in glaucoma.
Multilayer perceptrons (MLP), support vector machines (SVM), mixture of Gaussian (MoG), and mixture of generalized Gaussian (MGG) classifiers were trained and tested by cross validation on the numerical plot of absolute sensitivity plus age of 189 normal eyes and 156 glaucomatous eyes, designated as such by the appearance of the optic nerve. The authors compared performance of these classifiers with the global indices of STATPAC, using the area under the ROC curve. Two human experts were judged against the machine classifiers and the global indices by plotting their sensitivity-specificity pairs.
MoG had the greatest area under the ROC curve of the machine classifiers. Pattern SD (PSD) and corrected PSD (CPSD) had the largest areas under the curve of the global indices. MoG had significantly greater ROC area than PSD and CPSD. Human experts were not better at classifying visual fields than the machine classifiers or the global indices.
MoG, using the entire visual field and age for input, interpreted SAP better than the global indices of STATPAC. Machine classifiers may augment the global indices of STATPAC.
To determine whether topographical measurements of the parapapillary region analyzed by machine learning classifiers can detect early to moderate glaucoma better than similarly processed measurements ...obtained within the disc margin and to improve methods for optimization of machine learning classifier feature selection.
One eye of each of 95 patients with early to moderate glaucomatous visual field damage and of each of 135 normal subjects older than 40 years participating in the longitudinal Diagnostic Innovations in Glaucoma Study (DIGS) were included. Heidelberg Retina Tomograph (HRT; Heidelberg Engineering, Dossenheim, Germany) mean height contour was measured in 36 equal sectors, both along the disc margin and in the parapapillary region (at a mean contour line radius of 1.7 mm). Each sector was evaluated individually and in combination with other sectors. Gaussian support vector machine (SVM) learning classifiers were used to interpret HRT sector measurements along the disc margin and in the parapapillary region, to differentiate between eyes with normal and glaucomatous visual fields and to compare the results with global and regional HRT parameter measurements. The area under the receiver operating characteristic (ROC) curve was used to measure diagnostic performance of the HRT parameters and to evaluate the cross-validation strategies and forward selection and backward elimination optimization techniques that were used to generate the reduced feature sets.
The area under the ROC curve for mean height contour of the 36 sectors along the disc margin was larger than that for the mean height contour in the parapapillary region (0.97 and 0.85, respectively). Of the 36 individual sectors along the disc margin, those in the inferior region between 240 degrees and 300 degrees, had the largest area under the ROC curve (0.85-0.91). With SVM Gaussian techniques, the regional parameters showed the best ability to discriminate between normal eyes and eyes with glaucomatous visual field damage, followed by the global parameters, mean height contour measures along the disc margin, and mean height contour measures in the parapapillary region. The area under the ROC curve was 0.98, 0.94, 0.93, and 0.85, respectively. Cross-validation and optimization techniques demonstrated that good discrimination (99% of peak area under the ROC curve) can be obtained with a reduced number of HRT parameters.
Mean height contour measurements along the disc margin discriminated between normal and glaucomatous eyes better than measurements obtained in the parapapillary region.
We apply a variational method to automatically determine the number of mixtures of independent components in high-dimensional datasets, in which the sources may be nonsymmetrically distributed. The ...data are modeled by clusters where each cluster is described as a linear mixture of independent factors. The variational Bayesian method yields an accurate density model for the observed data without overfitting problems. This allows the dimensionality of the data to be identified for each cluster. The new method was successfully applied to a difficult real-world medical dataset for diagnosing glaucoma.
To determine whether Heidelberg Retina Tomograph (HRT; Heidelberg Engineering, Dossenheim, Germany) classification techniques and investigational support vector machine (SVM) analyses can detect ...optic disc abnormalities in glaucoma-suspect eyes before the development of visual field abnormalities.
Glaucoma-suspect eyes (n = 226) were classified as converts or nonconverts based on the development of repeatable (either two or three consecutive) standard automated perimetry (SAP)-detected abnormalities over the course of the study (mean follow-up, approximately 4.5 years). Hazard ratios for development of SAP abnormalities were calculated based on baseline classification results, follow-up time, and end point status (convert, nonconvert). Classification techniques applied were HRT classification (HRTC), Moorfields Regression Analysis, forward-selection optimized SVM (SVM fwd) and backward elimination-optimized SVM (SVM back) analysis of HRT data, and stereophotograph assessment.
Univariate analyses indicated that all classification techniques were predictors of the development of two repeatable abnormal SAP results, with hazards ratios (95% confidence interval CI) ranging from 1.32 (1.00-1.75) for HRTC to 2.0 (1.48-2.76) for stereophotograph assessment (all P < or = 0.05). Only SVM (SVM fwd and SVM back) analysis of HRT data and stereophotograph assessment were univariate predictors of the development of three repeatable abnormal SAP results, with hazard ratios (95% CI) ranging from 1.73 (1.16-2.82) for SVM fwd to 1.82 (1.19-3.12) for SVM back (both P < 0.007). Multivariate analyses including each classification technique individually in a model with age, baseline SAP pattern standard deviation PSD, and baseline IOP indicated that all classification techniques except HRTC (P = 0.06) were predictors of the development of two repeatable abnormal SAP results with hazards ratios ranging from 1.30 (0.99, 1.73) for HRTC to 1.90 (1.37, 2.69) for stereophotograph assessment. Only SVM (SVM fwd and SVM back) analysis of HRT data and stereophotograph assessment were significant predictors of the development of three repeatable abnormal SAP results in multivariate analyses; hazard ratios of 1.57 (1.03, 2.59) and 1.70 (1.18, 2.51), respectively. SAP PSD was a significant predictor of two repeatable abnormal SAP results in multivariate models with all classification techniques, with hazard ratios ranging from 3.31 (1.39, 7.89) to 4.70 (2.02, 10.93) per 1-dB increase.
HRT classifications techniques and stereophotograph assessment can detect optic disc topography abnormalities in glaucoma-suspect eyes before the development of SAP abnormalities. These data support strongly the importance of optic disc examination for early glaucoma diagnosis.
To compare machine learning classifiers trained on three clustering schemes to determine whether distinguishing healthy eyes from those with glaucomatous optic neuropathy (GON) can be optimized by ...training with clustered data.
Two machine learning classifiers-quadratic discriminant analysis (QDA) and support vector machines with Gaussian kernel (SVMg)-were trained separately using standard perimetry data from the Diagnostic Innovations in Glaucoma Study (DIGS), clustered using three clustering schemes on a training data set (123 eyes/123 glaucoma patients with GON; 135 eyes/135 normal control subjects). Trained classifiers were then applied to an independent data set containing 69 eyes of 69 glaucoma patients with early visual field loss and 83 eyes of 83 normal control subjects. Two control conditions were included: unclustered data and a random assignment of locations to clusters.
Areas under the receiver operating characteristic (ROC) curve ranged from 0.85 (SVMg, thresholds clustered by Glaucoma Hemifield Test sectors) to 0.92 (QDA, thresholds clustered by Garway-Heath mapping) for the training data set. Use of clustered data showed no significant optimization of sensitivity over use of unclustered data, and no single clustering method resulted in significantly higher performance in the independent data set. Sensitivities tended to be higher with QDA than with SVMg, regardless of specificity cutoff and clustering
QDA performed better with the early glaucoma data set than did the SVMg. Clustering may be advantageous when data-dimension reduction is needed-for example, when combining field results with other high-dimensional data (e.g., structural imaging data)-but it is not necessary for visual field data alone.
Missing data are common in real-world data sets and are a problem for many estimation techniques. We have developed a variational Bayesian method to perform independent component analysis (ICA) on ...high-dimensional data containing missing entries. Missing data are handled naturally in the Bayesian framework by integrating the generative density model. Modeling the distributions of the independent sources with mixture of gaussians allows sources to be estimated with different kurtosis and skewness. Unlike the maximum likelihood approach, the variational Bayesian method automatically determines the dimensionality of the data and yields an accurate density model for the observed data without overfitting problems. The technique is also extended to the clusters of ICA and supervised classification framework.
Clustering by unsupervised learning with machine learning classifiers was shown to segment clusters of patterns in standard automated perimetry (SAP) for glaucoma in previous publications. In this ...study, unsupervised learning by independent component analysis decomposed SAP field patterns into axes, and the information represented by these axes was evaluated.
SAP fields were used that were obtained with the Humphrey Visual Field Analyzer (Carl Zeiss Meditec, Dublin, CA) from 189 normal eyes and 156 eyes with glaucomatous optic neuropathy (GON) determined by masked review with stereoscopic optic disc photographs. The variational Bayesian independent component analysis mixture model (vB-ICA-mm) partitioned the SAP fields into the most informative number of clusters. Simultaneously, the model learned an optimal number of maximally independent axes for each cluster.
The most informative number of clusters in the SAP set was two. vB-ICA-mm placed 68.6% of the eyes with GON in a cluster labeled G and 98.4% of the eyes with normal optic discs in a cluster labeled N. Cluster G optimally contained six axes. Post hoc analysis of patterns generated at -1 SD and +2 SD from the cluster G mean on the six axes revealed defects similar to those identified by experts as indicative of glaucoma. SAP fields associated with an axis showed increasing severity, as they were located farther in the positive direction from the cluster G mean.
vB-ICA-mm represented the SAP fields with patterns that were meaningful for glaucoma experts. This process also captured severity in the patterns uncovered. These findings should validate vB-ICA-mm as a data-mining technique for new and unfamiliar complex tests.
To determine whether an unsupervised machine learning classifier can identify patterns of visual field loss in standard visual fields consistent with typical patterns learned by decades of human ...experience.
Standard perimetry thresholds for 52 locations plus age from one eye of each of 156 patients with glaucomatous optic neuropathy (GON) and 189 eyes of healthy subjects were clustered with an unsupervised machine classifier, variational Bayesian mixture of factor analysis (vbMFA).
The vbMFA formed five distinct clusters. Cluster 5 held 186 of 189 fields from normal eyes plus 46 from eyes with GON. These fields were then judged within normal limits by several traditional methods. Each of the other four clusters could be described by the pattern of loss found within it. Cluster 1 (71 GON + 3 normal optic discs) included early, localized defects. A purely diffuse component was rare. Cluster 2 (26 GON) exhibited primarily deep superior hemifield defects, and cluster 3 (10 GON) held deep inferior hemifield defects only or in combination with lesser superior field defects. Cluster 4 (6 GON) showed deep defects in both hemifields. In other words, visual fields within a given cluster had similar patterns of loss that differed from the predominant pattern found in other clusters. The classifier separated the data based solely on the patterns of loss within the fields, without being guided by the diagnosis, placing 98.4% of the healthy eyes within the same cluster and spreading 70.5% of the eyes with GON across the other four clusters, in good agreement with a glaucoma expert and pattern standard deviation.
Without training-based diagnosis (unsupervised learning), the vbMFA identified four important patterns of field loss in eyes with GON in a manner consistent with years of clinical experience.