Sensitivity and specificity of machine learning classifiers for glaucoma diagnosis using Spectral Domain OCT and standard automated perimetry.

PURPOSE: To evaluate the sensitivity and specificity of machine learning classifiers (MLCs) for glaucoma diagnosis using Spectral Domain OCT (SD-OCT) and standard automated perimetry (SAP). METHODS: Observational cross-sectional study. Sixty two glaucoma patients and 48 healthy individuals were included. All patients underwent a complete ophthalmologic examination, achromatic standard automated perimetry (SAP) and retinal nerve fiber layer (RNFL) imaging with SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec Inc., Dublin, California). Receiver operating characteristic (ROC) curves were obtained for all SD-OCT parameters and global indices of SAP. Subsequently, the following MLCs were tested using parameters from the SD-OCT and SAP: Bagging (BAG), Naive-Bayes (NB), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Random Forest (RAN), Ensemble Selection (ENS), Classification Tree (CTREE), Ada Boost M1(ADA),Support Vector Machine Linear (SVML) and Support Vector Machine Gaussian (SVMG). Areas under the receiver operating characteristic curves (aROC) obtained for isolated SAP and OCT parameters were compared with MLCs using OCT+SAP data. RESULTS: Combining OCT and SAP data, MLCs' aROCs varied from 0.777(CTREE) to 0.946 (RAN).The best OCT+SAP aROC obtained with RAN (0.946) was significantly larger the best single OCT parameter (p<0.05), but was not significantly different from the aROC obtained with the best single SAP parameter (p=0.19). CONCLUSION: Machine learning classifiers trained on OCT and SAP data can successfully discriminate between healthy and glaucomatous eyes. The combination of OCT and SAP measurements improved the diagnostic accuracy compared with OCT data alone.

Sensitivity and specificity of machine learning classifiers for glaucoma diagnosis using Spectral Domain OCT and standard automated perimetry / Sensibilidade e especificidade dos classificadores de aprendizagem de máquina para o diagnóstico de glaucoma usando Spectral Domain OCT e perimetria automatizada acromática

PURPOSE: To evaluate the sensitivity and specificity of machine learning classifiers (MLCs) for glaucoma diagnosis using Spectral Domain OCT (SD-OCT) and standard automated perimetry (SAP). METHODS: Observational cross-sectional study. Sixty two glaucoma patients and 48 healthy individuals were included. All patients underwent a complete ophthalmologic examination, achromatic standard automated perimetry (SAP) and retinal nerve fiber layer (RNFL) imaging with SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec Inc., Dublin, California). Receiver operating characteristic (ROC) curves were obtained for all SD-OCT parameters and global indices of SAP. Subsequently, the following MLCs were tested using parameters from the SD-OCT and SAP: Bagging (BAG), Naive-Bayes (NB), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Random Forest (RAN), Ensemble Selection (ENS), Classification Tree (CTREE), Ada Boost M1(ADA),Support Vector Machine Linear (SVML) and Support Vector Machine Gaussian (SVMG). Areas under the receiver operating characteristic curves (aROC) obtained for isolated SAP and OCT parameters were compared with MLCs using OCT+SAP data. RESULTS: Combining OCT and SAP data, MLCs' aROCs varied from 0.777(CTREE) to 0.946 (RAN).The best OCT+SAP aROC obtained with RAN (0.946) was significantly larger the best single OCT parameter (p<0.05), but was not significantly different from the aROC obtained with the best single SAP parameter (p=0.19). CONCLUSION: Machine learning classifiers trained on OCT and SAP data can successfully discriminate between healthy and glaucomatous eyes. The combination of OCT and SAP measurements improved the diagnostic accuracy compared with OCT data alone.

OBJETIVO: Avaliar a sensibilidade e especificidade dos classificadores de aprendizagem de máquina no diagnóstico de glaucoma usando Spectral Domain OCT (SD-OCT) e perimetria automatizada acromática (PAA). MÉTODOS: Estudo transversal observacional. Sessenta e dois pacientes com glaucoma e 48 indivíduos normais foram incluídos. Todos os pacientes foram submetidos a exame oftalmológico completo, e perimetria automatizada acromática (24-2 SITA; Humphrey Field Analyzer II, Carl Zeiss Meditec, Inc., Dublin, CA) e exame de imagem da camada de fibras nervosas utilizando SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec Inc., Dublin, California). Curvas ROC (Receiver operator characteristic) foram obtidas para todos os parâmetros do SD-OCT e índices globais do campo visual (MD, PSD, GHT). Subsequentemente, os seguintes classificadores de aprendizagem de máquina (CAMs) foram testados usando parâmetros do OCT e CV: Bagging (BAG), Naive-Bayes (NB), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Random Forest (RAN), Ensemble Selection (ENS), Classification Tree (CTREE), Ada Boost M1(ADA), Support Vector Machine Linear (SVML) e Support Vector Machine Gaussian (SVMG). Áreas abaixo da curva ROC (aROC) obtidas com os parâmetros isolados do campo visual (CV) e OCT foram comparados com os CAMs usando dados associados do OCT e CV. RESULTADOS: Combinando os dados do OCT e do CV, aROCs dos CAMs variaram entre 0,777(CTREE) e 0,946 (RAN). A maior aROC dos CAMs OCT+CV obtida com RAN (0,946) foi significativamente maior que o melhor parâmetro do OCT (p<0,05), mas não houve diferença estatística significativa com o melhor parâmetro do CV (p=0,19). CONCLUSÃO: Os classificadores de aprendizagem de máquina treinados com dados do OCT e CV podem discriminar entre olhos normais e glaucomatosos com sucesso. A combinação das medidas do OCT e CV melhoraram a acurácia diagnóstica comparados aos parâmetros do OCT.

Sensitivity and specificity of machine learning classifiers and spectral domain OCT for the diagnosis of glaucoma.

Purpose. To investigate the sensitivity and specificity of machine learning classifiers (MLC) and spectral domain optical coherence tomography (SD-OCT) for the diagnosis of glaucoma. Methods. Sixty-two patients with early to moderate glaucomatous visual field damage and 48 healthy individuals were included. All subjects underwent a complete ophthalmologic examination, achromatic standard automated perimetry, and RNFL imaging with SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec, Inc., Dublin, California, USA). Receiver operating characteristic (ROC) curves were obtained for all SD-OCT parameters. Subsequently, the following MLCs were tested: Classification Tree (CTREE), Random Forest (RAN), Bagging (BAG), AdaBoost M1 (ADA), Ensemble Selection (ENS), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Naive-Bayes (NB), and Support Vector Machine (SVM). Areas under the ROC curves (aROCs) obtained for each parameter and each MLC were compared. Results. The mean age was 57.0±9.2 years for healthy individuals and 59.9±9.0 years for glaucoma patients (p=0.103). Mean deviation values were -4.1±2.4 dB for glaucoma patients and -1.5±1.6 dB for healthy individuals (p<0.001). The SD-OCT parameters with the greater aROCs were inferior quadrant (0.813), average thickness (0.807), 7 o'clock position (0.765), and 6 o'clock position (0.754). The aROCs from classifiers varied from 0.785 (ADA) to 0.818 (BAG). The aROC obtained with BAG was not significantly different from the aROC obtained with the best single SD-OCT parameter (p=0.93). Conclusions. The SD-OCT showed good diagnostic accuracy in a group of patients with early glaucoma. In this series, MLCs did not improve the sensitivity and specificity of SD-OCT for the diagnosis of glaucoma.

Sex-specific hemodynamic and non-hemodynamic determinants of aortic root size in hypertensive subjects with left ventricular hypertrophy.

Aortic root (AoR) dilatation is more frequently observed in hypertensive individuals and is independently associated with left ventricular (LV) hypertrophy. Although the LV structure has sex-specific predictors, it remains unknown whether there are gender-related differences in the determinants of AoR size. We carried out a cross-sectional analysis of clinical, laboratory, anthropometric, funduscopic and echocardiographic features of 438 hypertensive patients with LV hypertrophy (266 women and 172 men). Women with enlarged AoR had higher cardiac output (P=0.0004), decreased peripheral vascular resistance (P=0.009), higher prevalence of mild aortic regurgitation (P=0.02) and increased waist circumference (P=0.04), whereas AoR-dilated men presented with a higher prevalence of concentric LV hypertrophy (P=0.0008) and mild aortic regurgitation (P=0.005) and increased log C-reactive protein levels (P=0.02), compared with sex-matched normal AoR subjects. In women, AoR dilatation associated with cardiac output, mild aortic regurgitation and waist circumference in a multivariate model including age, body surface area, height, homeostasis model assessment index, LV mass index, diastolic blood pressure, menopause status and use of antihypertensive medications as independent variables. Conversely, AoR dilatation associated with LV relative wall thickness, log C-reactive protein and mild aortic regurgitation without contributions from diastolic blood pressure, height, body surface area, LV mass index, peripheral vascular resistance and antihypertensive medications in men. Taken together, these results suggest that both volume overload and abdominal obesity are related to AoR dilatation in hypertensive women, whereas AoR enlargement is associated more with inflammatory and myocardial growth-related parameters in hypertensive men with LV hypertrophy.