Classifying glaucoma exclusively with OCT: comparison of three clustering algorithms derived from machine learning.

BACKGROUND/AIMS: To objectively classify eyes as either healthy or glaucoma based exclusively on data provided by peripapillary retinal nerve fiber layer (pRNFL) and ganglion cell-inner plexiform (GCIPL) measurements derived from spectral-domain optical coherence tomography (SD-OCT) using machine learning algorithms. METHODS: Three clustering methods (k-means, hierarchical cluster analysis -HCA- and model-based clustering-MBC-) were used separately to classify a training sample of 109 eyes as either healthy or glaucomatous using solely 13 SD-OCT parameters: pRNFL average and sector thicknesses and GCIPL average and minimum values together with the six macular wedge-shaped regions. Then, the best-performing algorithm was applied to an independent test sample of 102 eyes to derive close estimates of its actual performance (external validation). RESULTS: In the training sample, accuracy was 91.7% for MBC, 81.7% for k-means and 78.9% for HCA (p value = 0.02). The best MBC model was that in which subgroups were allowed to have variable volume and shape and equal orientation. The MBC algorithm in the independent test sample correctly classified 98 out of 102 cases for an overall accuracy of 96.1% (95% CI, 92.3-99.8%), with a sensitivity of 94.3 and 100% specificity. The accuracy for pRNFL was 92.2% (95% CI, 86.9-97.4%) and for GCIPL 98.0% (95% CI, 95.3-100%). CONCLUSIONS: Clustering algorithms in general (and MBC in particular) seem promising methods to help discriminate between healthy and glaucomatous eyes using exclusively SD-OCT-derived parameters. Understanding the relative merits of one method over others may also provide insights into the nature of the disease.

External Validation and Clinical Applicability of Two Optical Coherence Tomography-Based Risk Calculators for Detecting Glaucoma.

Purpose: To clinically validate the diagnostic ability of two optical coherence tomography (OCT)-based glaucoma diagnostic calculators (GDCs). Methods: We conducted a retrospective, consecutive sampling of 76 patients with primary open-angle glaucoma, 107 glaucoma suspects, and 67 controls. Demographics, reliable visual field testing, and macular and optic disc OCT were collected. The reference diagnosis was compared against the probability of having glaucoma obtained from two GDCs derived from multivariate logistic regressions using quantitative and qualitative (GDC1) or only quantitative (GDC2) OCT data. The discrimination (area under the curve [AUC]) and calibration (calibration plots) were compared for both calculators and the best OCT parameters. Results: GDC2 was able to identify 46.9% more suspects and 14.7% more glaucomatous eyes than GDC1. Both GDCs obtained the highest discriminative ability in glaucomatous eyes (GDC1 AUC = 0.949; GDC2 = 0.943 vs inferior peripapillary retinal nerve fiber layer [pRNFL] = 0.931; P = 0.43). The discriminating ability was not as good for glaucoma suspects, but the GDCs were not inferior to pRNFL (GDC 1 AUC = 0.739; GDC2 = 0.730; inferior pRNFL = 0.760; P = 0.54) and GDC2 was still able to correctly identify up to 30.8% more cases than the conventional OCT classification. Calibration showed risk underestimation for both groups and calculators, but it was better in GDC2 and in patients with glaucoma. Conclusions: OCT-based calculators showed an excellent diagnostic performance in glaucomatous eyes. GDC2 was able to identify approximately 30% more cases than the conventional pRNFL inferior OCT classification in both groups, suggesting a potential role of these composite scores in clinical practice. Translational Relevance: These OCT-based calculators may improve glaucoma diagnosis in clinical care.