Journal of Ophthalmology / 2013 / Article / Tab 3

Clinical Study

Glaucoma Diagnostic Accuracy of Machine Learning Classifiers Using Retinal Nerve Fiber Layer and Optic Nerve Data from SD-OCT

Table 3

Areas under the ROC curve (aROCs) for each SD-OCT parameter and sensitivities (%) with fixed specificities of 80% and 90%.

SD-OCTaROC (CI)Specificity 80%Specificity 90%

Average thickness0.783 (0.690–0.858)62.251.9
Quadrant
 Temporal 0.641 (0.540–0.733)38.728.0
 Superior 0.747 (0.652–0.828)57.855.4
 Nasal 0.672 (0.573–0.761)41.923.8
 Inferior0.775 (0.682–0.851)63.145.6
Clock hour
 10.690 (0.591–0.777)49.127.3
 20.720 (0.623–0.804)52.945.9
 30.563 (0.462–0.661)*23.118.4
 40.597 (0.495–0.692)*26.412.2
 50.642 (0.542–0.734)28.025.6
 60.711 (0.613–0.796)45.632.6
 70.764 (0.670–0.842)54.740.7
 80.638 (0.537–0.730)44.226.6
 90.564 (0.463–0.662)*31.525.7
 100.670 (0.570–0.759)47.731.9
 110.741 (0.646–0.823)58.632.6
 120.686 (0.587–0.774)42.824.5
Cup/disc area0.846 (0.762–0.910)67.760.0
Average cup/disc 0.843 (0.758–0.907)66.658.2
Vertical cup/disc0.832 (0.746–0.899)70.858.9
Rim area0.828 (0.741–0.895)70.162.4
Cup volume0.786 (0.694–0.860)64.942.1
Disc area0.594 (0.493–0.690)*33.319.3

SD-OCT: spectral domain optical coherence tomography; CI: confidence interval of 95%.
Parameters with aROCs not significantly different from chance.

Article of the Year Award: Outstanding research contributions of 2020, as selected by our Chief Editors. Read the winning articles.