Methods: We collected 3794 corneal images from 542 eyes of 280 subjects and developed seven deep learning models based on anterior and posterior eccentricity, anterior and posterior elevation, anterior and posterior sagittal curvature, and corneal thickness maps to extract deep corneal features. An independent subset with 1050 images collected from 150 eyes of 85 subjects from a separate center was used to validate models. We developed a hybrid deep learning model to detect KCN. We visualized deep features of corneal parameters to assess the quality of learning subjectively and computed area under the receiver operating characteristic curve (AUC), confusion matrices, accuracy, and F1 score to evaluate models objectively.
Results: In the development dataset, 204 eyes were normal, 123 eyes were suspected KCN, and 215 eyes had KCN. In the independent validation dataset, 50 eyes were normal, 50 eyes were suspected KCN, and 50 eyes were KCN. Images were annotated by three corneal specialists. The AUC of the models for the two-class and three-class problems based on the development set were 0.99 and 0.93, respectively.
Conclusions: The hybrid deep learning model achieved high accuracy in identifying KCN based on corneal maps and provided a time-efficient framework with low computational complexity.
Translational Relevance: Deep learning can detect KCN from non-invasive corneal images with high accuracy, suggesting potential application in research and clinical practice to identify KCN.
METHODS: This retrospective study comprised 284 right eyes. Patients aged 18 years or older with myopia up to -12.00 D and/or astigmatism up to -6.00 DC and who underwent femtosecond LASIK were recruited. Patients were divided into three subgroups: low myopia (-0.50 to -3.00 D), moderate myopia (>-3.00 to ≤-6.00 D), and high myopia (>-6.00 D), according to their pre-LASIK spherical equivalent (SE). The variables included for analysis were PCC (central 0-3.0 mm, pericentral 3.0-6.0 mm, and peripheral region 6.0-9.0 mm), PCE, PCA, internal anterior chamber depth, intraocular pressure, and central cornea thickness at the pre- and post-LASIK stages.
RESULTS: The central PCC remained unchanged across all three myopia subgroups at 1 month when compared to the pre-LASIK stage and remained stable at 6 months. The pericentral regions became flatter across all myopia subgroups at 1 month postsurgery (P < 0.001) and remained unchanged at 6 months. This trend was not seen in the peripheral cornea regions, which remained unchanged at 1 and 6 months post-LASIK when compared to pre-LASIK mean readings. There were minimal changes in post-LASIK posterior cornea astigmatism throughout follow-up. There was no incidence of post-LASIK surgery ectasia in this study population.
CONCLUSION: Post-LASIK, the different cornea subregions behaved differently. Overall, the posterior cornea surface remained stable post-LASIK across all myopia subgroups throughout follow-up.
MATERIALS AND METHODS: This study is a comparative, crosssectional and hospital-based study. We recruited 83 children with VKC and 83 healthy children as controls. All children underwent complete ocular examinations, CCT measurement using an ultrasound pachymeter and corneal topography using a Placido disc corneal analyser.
RESULTS: There was a statistically significant difference of means CCT and topographic indices in children with VKC compared to controls (p<0.05). The probability keratoconus reached 18% in children with VKC. The mean CCT was observed to be thinnest in the severe-to-very severe groups of VKC compared to the mild-to-moderate (p<0.05). The means simulated-K1 and -K2, apical keratometry, apical gradient curvature, superior-inferior index and keratoconus prediction index were significantly different in severe-tovery severe VKC compared to mild-to-moderate VKC and controls (p<0.05). However, there was no significant difference in mean cylinder value and percent probability keratoconus when comparing different groups of severity of VKC (p=0.912 and 0.070 respectively).
CONCLUSION: Children with VKC have thinner CCT and topographic indices changes compared to healthy children. Similar pattern was observed between groups with VKC. Degree of astigmatism and probability of keratoconus were similar in mild-to-moderate and severe-to-very severe groups.
METHODS: This was a cross-sectional observational study. From 2013 to 2014, we recruited inhabitants aged 50 years or older in Guangzhou, China. Among 1,117 participants in the study, data from 1,015 phakic right eyes were used for analyses. Ocular parameters including axial length (AL), anterior chamber depth (ACD), and corneal curvature (K) were measured using an IOL Master.
RESULTS: The mean AL, ACD, and K were 23.48 mm [95 % confidence interval (CI), 23.40-23.55], 3.03 mm (CI, 3.01-3.05), and 44.20 mm (CI, 44.11-44.29), respectively. A mean reduction in ACD with age was observed (P = 0.002) in male subjects but not in female subjects (P = 0.558). Male subjects had significantly longer ALs (23.68 mm versus 23.23 mm, P