No abstract

Infectious keratitis refers to a group of corneal disorders in which corneal tissues suffer inflammation and damage caused by pathogenic infections. Among these disorders, fungal keratitis (FK) and acanthamoeba keratitis (AK) are particularly severe and can cause permanent blindness if not diagnosed early and accurately. In Vivo Confocal Microscopy (IVCM) allows for imaging of different corneal layers and provides an important tool for an early and accurate diagnosis. In this paper, we introduce the IVCM-Keratitis dataset, which comprises of a total of 4001 sample images of AK and FK, as well as non-specific keratitis (NSK) and healthy corneas classes. We use this dataset to develop multiple deep-learning models based on Convolutional Neural Networks (CNNs) to provide automated assistance in enhancing the diagnostic accuracy of confocal microscopy in infectious keratitis. Densenet161 had the best performance among these models, with an accuracy, precision, recall, and F1 score of 93.55%, 92.52%, 94.77%, and 96.93%, respectively. Our study highlights the potential of deep learning models to provide automated diagnostic assistance for infectious keratitis via confocal microscopy images, particularly in the early detection of AK and FK. The proposed model can provide valuable support to both experienced and inexperienced eye-care practitioners in confocal microscopy image analysis, by suggesting the most likely diagnosis. We further demonstrate that these models can highlight the areas of infection in the IVCM images and explain the reasons behind their diagnosis by utilizing saliency maps, a technique used in eXplainable Artificial Intelligence (XAI) to interpret these models.

Purpose: To report eye bank records for pediatric keratoplasty in Iran between 2006 and 2019. Methods: In a retrospective study, all electronic records of the Central Eye Bank of Iran for pediatric keratoplasty between April 2006 and March 2019 were analyzed in terms of indications for keratoplasty, surgical techniques, their corresponding trends, and post-transplantation graft clarity. Results: Our database included 2178 eyes from 2050 pediatric cases. The leading indications for keratoplasty included acquired nontraumatic diseases (75.8%), congenital abnormalities (12.7%), corneal regraft (8.3%), and acquired traumatic diseases (3.2%). Keratoconus was the most common acquired nontraumatic cause (58%) and more common in the age group >12 years than those ≤12 years (P < 0.001). Congenital corneal abnormalities and regrafts were more common in the age group ≤12 years (both Ps < 0.001). The most common surgical technique was penetrating keratoplasty (PKP, 90.9%) followed by deep anterior lamellar keratoplasty (DALK, 7.3%), Descemet stripping automated endothelial keratoplasty (DSAEK, 1.1%), anterior lamellar keratoplasty (0.5%), and keratolimbal allograft transplantation (0.2%). DSAEK was more common in the age group ≤12 years (P = 0.002), which, unlike PKP and DALK, showed a significant ascending trend over the 14-year period (P = 0.018). Posttransplantation graft clarity was 96.8%. Conclusion: Keratoconus was the leading indication for pediatric keratoplasty in Iran. Although PKP was the predominant keratoplasty procedure for the treatment of pediatric corneal disorders, it showed a significant descending trend over the 14 years.

Purpose: Investigating impression cytology (IC) results of various types of clinically suspected ocular surface lesions over a 14-year period in a referral center in Iran. Methods: IC findings obtained from patients with different types of ocular surface disorders between 2005 and 2018 were reviewed. Agreement between clinical suspicions and IC results was evaluated by calculating Cohen’s Kappa coefficient (CKC). Results: Clinical suspicions in 688 surveyed eyes were ocular surface squamous neoplasia (OSSN, 42.0%), limbal stem cell deficiency (LSCD, 36.3%), dry eye-related disorders (DERD, 11.5%), Acanthamoeba keratitis (AK, 7.2%), benign pigmented lesions (BPL, 1.9%), immune-related conjunctivitis (IRC, 0.7%), and malignant pigmented lesions (MPL, 0.4%). General agreement between clinical suspicions and IC results was 0.68 for all groups. This agreement was almost perfect in AK (CKC = 0.966) and BPLs (CKC = 0.843), and was substantial in MPLs (CKC = 0.749), OSSNs (CKC = 0.684), and LSCD (CKC = 0.612). CKC in IRC (0.567) and DERDs (0.443) was moderate. Histopathologic results were available in 22 eyes and were well-correlated with corresponding IC results (CKC = 0.86). Multiple post-treatment follow-up sessions of IC were performed in 51 eyes (11.4%) that had diagnosis of LSCD (31), OSSN (17), and MPL (3) at the first IC session. Conclusion: Our survey not only demonstrated an overall substantial agreement between IC results and primary clinical suspicions, but also showed an almost perfect correlation between IC results and existent histopathologic data. Therefore, IC as a non-invasive diagnostic modality can be of great importance in proper diagnosis of various ocular surface diseases especially when distinguishing malignant from benign lesions is required.