Purpose. To compare intraocular pressure (IOP) measurements obtained using the Topocon noncontact tonometer (NCT), the Goldmann applanation tonometer (GAT), and the Corvis ST (CST), a newly developed tonometer with features of visualization and measurement of the corneal deformation response to an air impulse. A secondary objective was to assess the agreement among the devices. Methods. Fifty-nine participants, including glaucoma patients (36 cases) and control volunteers (23 cases), were enrolled. One eye was selected randomly for further study. IOP measurements were obtained with the CST, NCT, and GAT by two experienced clinicians. IOP values were compared. Intraobserver variability and interobserver variability were assessed by the coefficient of variation and intraclass correlation coefficient. Device agreement was calculated by Bland-Altman analysis. Results. Mean IOP for all examined eyes was 18.9 ± 5.8 mm Hg for CST, 21.3 ± 6.8 mm Hg for NCT, and 20.3 ± 5.7 mm Hg for GAT. There was no statistically significant difference in IOP measurements among the tonometers except between the CST and NCT. Correlation analysis showed a high correlation between each pair of devices (all P < 0.001). The CST displayed the best intraobserver variability and interobserver variability. Bland-Altman analysis revealed a bias between CST and GAT, CST and NCT, and GAT and NCT of −1.3, −2.4, and −1.1 mm Hg, with 95% limits of agreement of −6.2 to 3.5 mm Hg, −10.1 to 5.2 mm Hg, and −8.3 to 6.2 mm Hg, respectively. Conclusions. The CST offers an alternative method for measuring IOP. IOP measurements taken with these devices may not be interchangeable.

Purpose: To investigate the applicability of a newly developed corneal topographer in assessing tear film stability. Methods: This is a prospective, case–control study. Forty-four Chinese dry eye patients and 41 normal subjects were recruited. Noninvasive tear film break-up time (NI-BUT) was measured using a new method based on a corneal topographer equipped with modified scan software. The reliability of the measurements was determined. Then, the correlations between the NI-BUT and the traditional fluorescein tear film break-up time, Schirmer I test values, and inferior tear meniscus height measurements were determined. The receiver operating characteristic curve technique was used to evaluate the NI-BUT examination in the diagnosis of dry eye. Results: In total, a significant difference between the NI-BUT and the fluorescein tear film break-up time was found (3.2 ± 2.3 seconds vs. 5.2 ± 3.4 seconds; P < 0.001). The coefficient of variation and intraclass correlation coefficient values of NI-BUT were 12.8% and 0.93, respectively, for NI-BUT for intraobserver repeatability and 15.4% and 0.88, respectively for interobserver repeatability. The NI-BUT showed a good correlation with other dry eye examinations (all P < 0.05). In addition, the NI-BUT was significantly shorter in dry eye patients (2.0 ± 0.2 seconds) than in normal subjects (4.3 ± 0.3 seconds; P < 0.001). When the cutoff value was set at <2.65 seconds, the sensitivity and specificity of the test were 84.1% and 75.6%, respectively. Conclusions: Measurements of NI-BUT obtained with the newly developed corneal topographer may provide a simple, noninvasive screening test for dry eyes with acceptable sensitivity, specificity, and repeatability.

To demonstrate the identification of corneal diseases using a novel deep learning algorithm. A novel hierarchical deep learning network, which is composed of a family of multi-task multi-label learning classifiers representing different levels of eye diseases derived from a predefined hierarchical eye disease taxonomy was designed. Next, we proposed a multi-level eye disease-guided loss function to learn the fine-grained variability of eye diseases features. The proposed algorithm was trained end-to-end directly using 5,325 ocular surface images from a retrospective dataset. Finally, the algorithm’s performance was tested against 10 ophthalmologists in a prospective clinic-based dataset with 510 outpatients newly enrolled with diseases of infectious keratitis, non-infectious keratitis, corneal dystrophy or degeneration, and corneal neoplasm. The area under the ROC curve of the algorithm for each corneal disease type was over 0.910 and in general it had sensitivity and specificity similar to or better than the average values of all ophthalmologists. Confusion matrices revealed similarities in misclassification between human experts and the algorithm. In addition, our algorithm outperformed over all four previous reported methods in identified corneal diseases. The proposed algorithm may be useful for computer-assisted corneal disease diagnosis.