Purpose: To assess the use of deep learning (DL) for computer-assisted glaucoma identification, and the impact of training using images selected by an active learning strategy, which minimizes labelling cost. Additionally, this study focuses on the explainability of the glaucoma classifier. Methods: This original investigation pooled 8433 retrospectively collected and anonymized colour optic disc-centred fundus images, in order to develop a deep learning-based classifier for glaucoma diagnosis. The labels of the various deep learning models were compared with the clinical assessment by glaucoma experts. Data were analysed between March and October 2018. Sensitivity, specificity, area under the receiver operating characteristic curve (AUC), and amount of data used for discriminating between glaucomatous and nonglaucomatous fundus images, on both image and patient level. Results: Trained using 2072 colour fundus images, representing 42% of the original training data, the trained DL model achieved an AUC of 0.995, sensitivity and specificity of, respectively, 98.0% (CI 95.5%-99.4%) and 91% (CI 84.0%-96.0%), for glaucoma versus non-glaucoma patient referral. Conclusions: These results demonstrate the benefits of deep learning for automated glaucoma detection based on optic disc-centred fundus images. The combined use of transfer and active learning in the medical community can optimize performance of DL models, while minimizing the labelling cost of domain-specific mavens. Glaucoma experts are able to make use of heat maps generated by the deep learning classifier to assess its decision, which seems to be related to inferior and superior neuroretinal rim (within ONH), and RNFL in superotemporal and inferotemporal zones (outside ONH).

BackgroundTo evaluate the relationships between Reichert Ocular Response Analyzer (ORA) parameters corneal hysteresis (CH) and corneal response factor (CRF) and ocular dimensions, age and intraocular pressure.MethodsTwo hundred and twelve eyes of 212 participants with no ocular pathology had CH and CRF measured with the ORA. Intraocular pressure (IOP) was measured with the Dynamic Contour tonometer and central corneal thickness (CCT) was also evaluated. Partial least squares linear regression (PLSLR) analyses were performed to examine the relationships between each response variable, CH and CRF, and the predictor variables age, corneal curvature (CC), axial length (AL), CCT and IOP.ResultsCH was positively associated with CCT and negatively associated with age (scaled coefficients: CCT 0.62, p < 0.0001; age -0.55, p <0.0001; r2 = 0.25). CRF was positively associated with CCT and DCT IOP and negatively associated with age and AL (scaled coefficients: CCT 0.89, p < 0.0001; DCT IOP 0.46, p < 0.01; age – 0.60, p < 0.0001; AL -0.37, p < 0.01; r2 = 0.43). There was no significant association between CC and CH or CRF.ConclusionsThe study suggests that age and CCT are strongly associated with CH and CRF, and that the latter is also influenced by AL and IOP. However, the variables studied could explain only 25% and 43% of the measured variation in CH and CRF, respectively, suggesting other factors also affect the values of these measurements.

In normal healthy eyes, the ocular pulse amplitude remains stable during normal outpatient office hours and was not correlated with blood pressure or age of patients.

Patients with glaucoma seem to have distinctive corneal biomechanical properties compared to OHT and NL. They may be influenced by many other unknown subparameters.

ABSTRACT.Purpose: To compare corneal hysteresis (CH) and corneal resistance factor (CRF) measured with the Ocular Response Analyzer Ò tonometer (ORA) between (i) African normals and treated primary open-angle glaucoma (POAG) patients and (ii) between normals and treated POAG Caucasians. To analyse the correlation of CH and CRF with visual field (VF) defects in the two groups.Methods: This comparative study included 59 African (29 (POAG), 30 normals) and 55 Caucasians (30 POAG and 25 normals) subjects. Goldmann applanation tonometry (GAT) and ORA measurements were performed in a randomized sequence. Visual field was tested with the Swedish interactive threshold algorithms standard strategy of the Humphrey perimeter. Hoddap classification was used to estimate the severity of VF defects.Results: Primary open-angle glaucoma Africans were younger than POAG Caucasians (p < 0.001). Goldmann applanation tonometry and central corneal thickness (CCT) did not differ significantly between the four subgroups. African normals had lower CH than Caucasian controls (p < 0.001). CH was 9.2 ± 1.1 and 8.3 ± 1.7 mmHg respectively in POAG Caucasians and Africans (p < 0.001). African controls had higher ORA corneal-compensated intraocular pressure (IOPcc) than Caucasian controls (p < 0.001). Primary open-angle glaucoma Africans had higher IOPcc values than Caucasian POAGs (p < 0.001). CH and IOPcc were associated with race (p < 0.001) but not with CCT. Based on mean deviation values (MD), POAG Africans had more severe VF defects. CH was correlated with MD (r = 0.442; p = 0.031) and severity of VF defects only in POAG Africans (r = )0.464; p = 0.013).Conclusions: African normal subjects and POAG patients had an altered CH, which is associated with a significant underestimation of GAT IOP. This may potentially contribute to the earlier development and greater severity of glaucoma damage in Africans compared with Caucasians at diagnosis.