DenseUNets with feedback non-local attention for the segmentation of specular microscopy images of the corneal endothelium with guttae.

Corneal guttae, which are the abnormal growth of extracellular matrix in the corneal endothelium, are observed in specular images as black droplets that occlude the endothelial cells. To estimate the corneal parameters (endothelial cell density [ECD], coefficient of variation [CV], and hexagonality [HEX]), we propose a new deep learning method that includes a novel attention mechanism (named fNLA), which helps to infer the cell edges in the occluded areas. The approach first derives the cell edges, then infers the well-detected cells, and finally employs a postprocessing method to fix mistakes. This results in a binary segmentation from which the corneal parameters are estimated. We analyzed 1203 images (500 contained guttae) obtained with a Topcon SP-1P microscope. To generate the ground truth, we performed manual segmentation in all images. Several networks were evaluated (UNet, ResUNeXt, DenseUNets, UNet++, etc.) and we found that DenseUNets with fNLA provided the lowest error: a mean absolute error of 23.16 [cells/mm[Formula: see text]] in ECD, 1.28 [%] in CV, and 3.13 [%] in HEX. Compared with Topcon's built-in software, our error was 3-6 times smaller. Overall, our approach handled notably well the cells affected by guttae, detecting cell edges partially occluded by small guttae and discarding large areas covered by extensive guttae.

A multi-ethnic genome-wide association study implicates collagen matrix integrity and cell differentiation pathways in keratoconus.

Keratoconus is characterised by reduced rigidity of the cornea with distortion and focal thinning that causes blurred vision, however, the pathogenetic mechanisms are unknown. It can lead to severe visual morbidity in children and young adults and is a common indication for corneal transplantation worldwide. Here we report the first large scale genome-wide association study of keratoconus including 4,669 cases and 116,547 controls. We have identified significant association with 36 genomic loci that, for the first time, implicate both dysregulation of corneal collagen matrix integrity and cell differentiation pathways as primary disease-causing mechanisms. The results also suggest pleiotropy, with some disease mechanisms shared with other corneal diseases, such as Fuchs endothelial corneal dystrophy. The common variants associated with keratoconus explain 12.5% of the genetic variance, which shows potential for the future development of a diagnostic test to detect susceptibility to disease.

The rising incidence of Acanthamoeba keratitis: A 7-year nationwide survey and clinical assessment of risk factors and functional outcomes.

PURPOSE: To evaluate the incidence of Acanthamoeba keratitis in the Netherlands between 2009 and 2015 and to analyse predicting factors for treatment outcome. METHODS: Patient characteristics, diagnostic methods, diagnostic delay, therapy prior to and after diagnosis, and visual outcome were obtained from medical files of all patients diagnosed with Acanthamoeba keratitis in the Netherlands between 2009 and 2015. A logistic regression analysis on treatment failure, defined as a best corrected visual acuity of less than 20/40 Snellen decimals (i.e. >0.3 logMAR or an approximate loss of three lines of visual acuity) and/or the need for keratoplasty, was performed to determine predicting factors. RESULTS: Two hundred and twenty-four eyes of 224 patients were included. Ninety-five percent of the patients were contact lens wearers, of whom 74% wore soft contact lenses. The number of cases increased from 16 in 2009 to 49 in 2015. This resulted in an estimated incidence of 1 in 21,000 for soft contact lens wearers in 2015. Eighty-seven eyes (39%) met the criteria for treatment failure. In a multivariable regression analysis, higher age at presentation, a higher severity stage and corticosteroid use before diagnosis were positively correlated with treatment failure. Early referral to a cornea specialist was associated with better clinical outcomes. CONCLUSIONS: Although Acanthamoeba keratitis is still a relatively uncommon disease, the incidence in soft contact lens wearers has increased to reach 1 in 21,000 in 2015. Treatment failure occurred in 39% of cases, with age, higher severity stage, corticosteroid use before diagnosis and indirect referral to a cornea specialist as important risks factors.

Endothelial cell density after posterior lamellar keratoplasty: five- to seven-year follow-up.

PURPOSE: To report endothelial cell density (ECD) after posterior lamellar keratoplasty (PLK). DESIGN: Prospective, interventional case series. METHODS: In a single institution, PLK was performed in 15 eyes of 15 patients with pseudophakic bullous keratopathy (PPBK), Fuchs endothelial dystrophy (FED), or both. In 11 eyes, the donor tissue was inserted unfolded through a 9.0-mm sclerocorneal incision. In four eyes, the donor was folded and inserted through a 5.0-mm incision. The ECD measurements were performed in each patient at regular intervals until three years after surgery, and then in all patients simultaneously at 54 to 84 months after surgery. RESULTS: Three patients were not available for follow-up. One graft had decompensated. In the remaining 11 eyes, ECDs ranged from 368 to 1576 cells/mm(2). Monophasic and biphasic regression models of ECD against time were fitted for each technique separately. CONCLUSIONS: After PLK, ECD declined at a high rate. Both early- and longer-term loss rates may be related to surgical technique.

Validity of endothelial cell analysis methods and recommendations for calibration in Topcon SP-2000P specular microscopy.

PURPOSE: To report on the calibration of the Topcon SP-2000P specular microscope and the Endothelial Cell Analysis Module of the IMAGEnet 2000 software, and to establish the validity of the different endothelial cell density (ECD) assessment methods available in these instruments. METHODS: Using an external microgrid, we calibrated the magnification of the SP-2000P and the IMAGEnet software. In both eyes of 36 volunteers, we validated 4 ECD assessment methods by comparing these methods to the gold standard manual ECD, manual counting of cells on a video print. These methods were: the estimated ECD, estimation of ECD with a reference grid on the camera screen; the SP-2000P ECD, pointing out whole contiguous cells on the camera screen; the uncorrected IMAGEnet ECD, using automatically drawn cell borders, and the corrected IMAGEnet ECD, with manual correction of incorrectly drawn cell borders in the automated analysis. Validity of each method was evaluated by calculating both the mean difference with the manual ECD and the limits of agreement as described by Bland and Altman. RESULTS: Preset factory values of magnification were incorrect, resulting in errors in ECD of up to 9%. All assessments except 1 of the estimated ECDs differed significantly from manual ECDs, with most differences being similar (< or =6.5%), except for uncorrected IMAGEnet ECD (30.2%). Corrected IMAGEnet ECD showed the narrowest limits of agreement (-4.9 to +19.3%). CONCLUSIONS: We advise checking the calibration of magnification in any specular microscope or endothelial analysis software as it may be erroneous. Corrected IMAGEnet ECD is the most valid of the investigated methods in the Topcon SP-2000P/IMAGEnet 2000 combination.

Biocompatibility of trypan blue with human corneal cells.

OBJECTIVE: To quantify the toxicity of trypan blue on human corneal cells according to exposure time and concentration. METHODS: Three in vitro experiments were performed. (1) We exposed cultured human corneal fibroblasts to trypan blue (0.0001% to 0.1%) in Eagle modified minimum essential medium (EMEM) or phosphate-buffered saline (PBS) for 15 minutes to 24 hours. Cytotoxicity was evaluated by Mosmann's colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MMT) assay. (2) We exposed human corneas in EMEM for 24 hours to trypan blue (0.001% to 0.1%). Fellow donor corneas served as controls. Endothelial survival was evaluated morphologically and by cell density assessment. (3) We morphologically compared the endothelial viability of human donor corneas after exposure to 0.1% trypan blue for 5 to 30 minutes with control corneas. RESULTS: In experiment 1, trypan blue in EMEM was not significantly toxic at concentrations of 0.005% or lower. Higher concentrations were toxic only after exposure to trypan blue for at least 6 hours. In PBS, significant toxicity was found after exposure to 0.1% trypan blue for 30 minutes or longer. Lower concentrations were toxic after longer exposures. In experiment 2, exposure to 0.01% and 0.1% trypan blue for 24 hours resulted in significant loss in cell density. At lower concentrations, the endothelium was affected only morphologically. In experiment 3, endothelial morphology changed in control corneas and after exposure to 0.1% trypan blue for as little as 5 minutes. After 30-minute exposure, morphologic deterioration was more pronounced. CONCLUSIONS: Trypan blue was toxic in vitro to corneal endothelium and corneal fibroblasts at higher concentrations and notably longer exposure times. Toxicity was less in EMEM than in PBS. Clinical Relevance At commonly used concentrations, both during cataract surgery and in the cornea bank, trypan blue is safe for corneal cells. At higher concentrations or longer exposures, however, caution is warranted.

Endothelial cell density after deep anterior lamellar keratoplasty (Melles technique).

PURPOSE: To measure the recipient endothelial cell loss after the Melles technique for deep anterior lamellar keratoplasty. METHODS: In 21 eyes of 21 patients, a deep anterior lamellar keratoplasty procedure was performed. Before surgery and at 6, 12, and 24 months after surgery, specular microscopy was performed to evaluate the endothelial cell density. For each postoperative time interval, the mean endothelial cell loss relative to the preoperative value was calculated. RESULTS: Mean postoperative endothelial cell loss averaged 283 cells/mm(2) (+/- 293) at 6 months, 335 cells/mm(2) (+/- 309) at 12 months, and 421 cells/mm(2) (+/- 316) at 24 months. Estimate relative endothelial cell density losses obtained by mixed model analysis of variance were 11.1%, 2.0%, and 1.2%, respectively, each time compared with its previous measurement point. Second order comparisons showed that the loss within the first 6 months was significantly higher than after 6 months. CONCLUSION: In deep anterior lamellar keratoplasty, the recipient corneal endothelium showed a small initial drop in endothelial cell density followed by a physiologic rate of cell loss. Cell survival after lamellar keratoplasty may be expected to be better when compared with that following penetrating keratoplasty.