Our purpose is to demonstrate the changes in cornea nerve parameters and symptoms and signs in dry eye disease (DED) patients after oral vitamin B1 and mecobalamin treatment. In this randomized double-blind controlled trial, DED patients were randomly assigned to either the treatment group (oral vitamin B1 and mecobalamin, artificial tears) or the control group (artificial tears). Corneal nerve parameters via in vivo confocal microscopy (IVCM), DED symptoms, and signs were assessed at baseline and 1 and 3 months post-treatment. In total, 398 eyes from 199 patients were included. In the treatment group, there were significant improvements in corneal nerve length, width, and neuromas, the sign of conjunctival congestion score (CCS), symptoms of dryness, pain, photophobia, blurred vision, total symptom score, and OSDI (OSDI) at 1/3 months post-treatment (all p < 0.05). Patients who received vitamin B1 and mecobalamin showed greater improvement in CCS, dryness scores at 1 month (p < 0.05), corneal fluorescein staining (CFS) (p = 0.012), photophobia (p = 0.032), total symptom scores (p = 0.041), and OSDI (p = 0.029) at 3 months. Greater continuous improvement in CFS (p = 0.045), dryness (p = 0.033), blurred vision (p = 0.031) and total symptom scores (p = 0.023) was demonstrated at 3 months than at 1 month post-treatment in the treatment group. We found that oral vitamin B1 and mecobalamin can improve corneal nerve length, width, reflectivity and the number of neuromas in IVCM, thereby repairing epithelial cells and alleviating some ocular symptoms. Thus, vitamin B1 and mecobalamin are potential treatment options for patients with DED.

To develop a deep learning model for automatically segmenting tarsus and meibomian gland areas on meibography, we included 1087 meibography images from dry eye patients. The contour of the tarsus and each meibomian gland was labeled manually by human experts. The dataset was divided into training, validation, and test sets. We built a convolutional neural network-based U-net and trained the model to segment the tarsus and meibomian gland area. Accuracy, sensitivity, specificity, and receiver operating characteristic curve (ROC) were calculated to evaluate the model. The area under the curve (AUC) values for models segmenting the tarsus and meibomian gland area were 0.985 and 0.938, respectively. The deep learning model achieved a sensitivity and specificity of 0.975 and 0.99, respectively, with an accuracy of 0.985 for segmenting the tarsus area. For meibomian gland area segmentation, the model obtained a high specificity of 0.96, with high accuracy of 0.937 and a moderate sensitivity of 0.751. The present research trained a deep learning model to automatically segment tarsus and the meibomian gland area from infrared meibography, and the model demonstrated outstanding accuracy in segmentation. With further improvement, the model could potentially be applied to assess the meibomian gland that facilitates dry eye evaluation in various clinical and research scenarios.

Purpose: The purpose of this study was to investigate the short- and long-term learning effect of dynamic visual acuity (DVA) tests. Methods: Participants between 18 and 30 years with corrected to normal visual acuity were enrolled in this study. Three repeated sessions were performed, with 15 minutes and 15 days’ intervals between sessions. Each session included 9 DVA tests of horizontal, vertical, and diagonal motions of E optotypes at 20, 40, and 80 degrees per second (dps). The short- and long-term learning effects were analyzed from repeated DVA tests. Results: Of the 58 enrolled participants, the mean age was 23.1 ± 2.1 years. DVA significantly varied among motion types and velocities (P < 0.05, respectively). There was a significant short-term learning effect for 20 (P = 0.004), 40 (P < 0.001), and 80 (P = 0.014) dps DVA test of horizontal motion, 40 dps DVA test of vertical (P = 0.003), and diagonal motion (P = 0.036). The long-term learning effect was detected in the 40 dps diagonal motion DVA test (P = 0.015). The short- and long-term learning effects were positively associated with initial DVA in most combinations of motion type and velocity tests (P < 0.05, respectively). The short- (P = 0.031) and long-term (P = 0.024) learning effect of 80 dps horizontal motion DVA test was greater in male than female participants. Conclusions: There is a significant short-term learning effect in the DVA test of various motion types, but the long-term learning effect was rarely observed, and it is greater in participants with worse initial DVA.

The present research aims to evaluate the feasibility of a deep-learning model in identifying bulbar conjunctival injection grading. Methods: We collected 1401 color anterior segment photographs demonstrating the cornea and bulbar conjunctival. The ground truth was bulbar conjunctival injection scores labeled by human ophthalmologists. Two convolutional neural network-based models were constructed and trained. Accuracy, precision, recall, F1-score, Kappa, and the area under the curve (AUC) were calculated to evaluate the efficiency of the deep learning models. The micro-average and macro-average AUC values for model grading bulbar conjunctival injection were 0.98 and 0.98, respectively. The deep learning model achieved a high accuracy of 87.12%, a precision of 87.13%, a recall of 87.12%, an F1-score of 87.07%, and Cohen’s Kappa of 0.8153. The deep learning model demonstrated excellent performance in evaluating the severity of bulbar conjunctival injection, and it has the potential to help evaluate ocular surface diseases and determine disease progression and recovery.

Purpose: This study aimed to evaluate the diagnostic accuracy of a novel shape index, the Pentacam Random Forest Index (PRFI), in detecting keratoconus (KC), specifically subclinical keratoconus, in Chinese refractive surgery candidates. Methods: This prospective cohort study included 856 participants who were divided into four groups based on their tomographic outcomes: the KC group (n = 137), the very asymmetric ectasia (VAE) group (n = 73), the normal cornea group (n = 363) and the tomographically suspected KC (TSK) group (n = 283). The diagnostic performance of PRFI and other widely used indices, including the shape index BAD-D and the combined index TBI, was assessed using receiver operating characteristic (ROC) curve analysis and compared using DeLong’s test. The area under the curve (AUC), best cutoff values, and Youden index for each parameter are reported. Additionally, the false-positive rates of BAD-D and PRFI were calculated and compared in “normal corneas”. Results: All shape and biomechanical parameters collected in this study were found to be significantly different among the four groups (KC, VAE, TSK, and normal groups; p = 0.000). The AUC of PRFI was the highest in detecting any form of KC (including clinical KC eyes and VAE-NT eyes) in Chinese refractive surgery candidates, outperforming the widely used shape index BAD-D (0.919 vs. 0.890, p < 0.001). There was no significant difference in performance between the PRFI and the combined TBI index (0.919 vs. 0.916, p > 0.05). For detecting subclinical KC eyes (i.e., VAE-NT), the AUC of PRFI was 0.774, which was statistically comparable to TBI (0.774 vs. 0.776, p > 0.05), while outperforming BAD-D (0.774 vs. 0.684, p < 0.001). The best cutoff values of PRFI for detecting any KC and VAE-NT eyes were determined to be 0.37 and 0.27, respectively. Additionally, PRFI demonstrated a lower false-positive rate than BAD-D (13.8% vs. 43.8%, p < 0.001). Notably, the relatively high false-positive rate of BAD-D observed in this study might be attributed to the smaller horizontal corneal diameter in tomographically suspected eyes. Conclusions: The PRFI proved to be a superior shape index compared to BAD-D in detecting any form of keratoconus, including subclinical cases, in Chinese refractive surgery candidates. This finding may be attributed to the relatively small corneas commonly observed in Asians.