A multi-ancestry GWAS of Fuchs corneal dystrophy highlights the contributions of laminins, collagen, and endothelial cell regulation.

Fuchs endothelial corneal dystrophy (FECD) is a leading indication for corneal transplantation, but its molecular etiology remains poorly understood. We performed genome-wide association studies (GWAS) of FECD in the Million Veteran Program followed by multi-ancestry meta-analysis with the previous largest FECD GWAS, for a total of 3970 cases and 333,794 controls. We confirm the previous four loci, and identify eight novel loci: SSBP3, THSD7A, LAMB1, PIDD1, RORA, HS3ST3B1, LAMA5, and COL18A1. We further confirm the TCF4 locus in GWAS for admixed African and Hispanic/Latino ancestries and show an enrichment of European-ancestry haplotypes at TCF4 in FECD cases. Among the novel associations are low frequency missense variants in laminin genes LAMA5 and LAMB1 which, together with previously reported LAMC1, form laminin-511 (LM511). AlphaFold 2 protein modeling, validated through homology, suggests that mutations at LAMA5 and LAMB1 may destabilize LM511 by altering inter-domain interactions or extracellular matrix binding. Finally, phenome-wide association scans and colocalization analyses suggest that the TCF4 CTG18.1 trinucleotide repeat expansion leads to dysregulation of ion transport in the corneal endothelium and has pleiotropic effects on renal function.

Deep learning segmentation of endothelial cell images using an active learning paradigm with guided label corrections.

Purpose: To create Guided Correction Software for informed manual editing of automatically generated corneal endothelial cell (EC) segmentations and apply it to an active learning paradigm to analyze a diverse set of post-keratoplasty EC images. Approach: An original U-Net model trained on 130 manually labeled post-Descemet stripping automated endothelial keratoplasty (EK) images was applied to 841 post-Descemet membrane EK images generating "uncorrected" cell border segmentations. Segmentations were then manually edited using the Guided Correction Software to create corrected labels. This dataset was split into 741 training and 100 testing EC images. U-Net and DeepLabV3+ were trained on the EC images and the corresponding uncorrected and corrected labels. Model performance was evaluated in a cell-by-cell analysis. Evaluation metrics included the number of over-segmentations, under-segmentations, correctly identified new cells, and endothelial cell density (ECD). Results: Utilizing corrected segmentations for training U-Net and DeepLabV3+ improved their performance. The average number of over- and under-segmentations per image was reduced from 23 to 11 with the corrected training set. Predicted ECD values generated by networks trained on the corrected labels were not significantly different than the ground truth counterparts (p=0.02, paired t-test). These models also correctly segmented a larger percentage of newly identified cells. The proposed Guided Correction Software and semi-automated approach reduced the time to accurately segment EC images from 15 to 30 to 5 min, an ∼80% decrease compared to manual editing. Conclusions: Guided Correction Software can efficiently label new training data for improved deep learning performance and generalization between EC datasets.

Effect of Simultaneous Endothelial Keratoplasty and Cataract Surgery: A Secondary Analysis of the Descemet Endothelial Thickness Comparison Trial.

PURPOSE: Combining cataract surgery with endothelial keratoplasty (triple EK) is a common practice and may be safer because it commits the patient to only one surgery. This study aimed to determine whether outcomes of pseudophakic endothelial keratoplasty and triple EK have similar outcomes. METHODS: This was a non-prespecified secondary analysis of a multicenter, double-masked, randomized, controlled clinical trial. Enrollment centers included the Casey Eye Institute at Oregon Health and Sciences University and the Byers Eye Institute at Stanford University. Patients with damaged or diseased endothelium and were considered good candidates for either Descemet membrane endothelial keratoplasty or ultrathin Descemet stripping automated endothelial keratoplasty were randomized to one of the two surgeries 1-2 days prior to surgery. If the patient had a cataract, they underwent simultaneous cataract surgery. The primary outcome was visual acuity at 6 months. Visual acuity at 3, 12, and 24 months, 3, 6, 12, and 24 month endothelial cell density and intraoperative and postoperative complications were also recorded. RESULTS: Those who had pseudophakic EK generally were older, more likely to be male, and more likely to have worse baseline vision with higher corneal densitometry values. After controlling for baseline visual acuity and type of keratoplasty, those that underwent Triple-EK had 0.09 better LogMAR lines better visual acuity at 6 (95% CI -0.17 to -0.10; P = 0.02) and 12 months (95% CI -0.21 to -0.07; P = 0.03). Although those receiving Triple- EK had nearly one-half LogMAR line better visual acuity at 24 months, it was no longer statistically significant (95% CI -0.20 to -0.05; P = 0.36). There was no statistically significant difference in ECD between groups at any timepoint. Those undergoing triple EK were more likely to have at least one adverse event (P = 0.02). CONCLUSIONS: Earlier intervention with Triple-EK may improve visual acuity outcomes after endothelial keratoplasty compared with staged procedures, but may also increase the risks of adverse events including the need for re-bubble.

Descemet Membrane Endothelial Keratoplasty: 10-Year Cell Loss and Failure Rate Compared With Descemet Stripping Endothelial Keratoplasty and Penetrating Keratoplasty.

PURPOSE: The aim of this study was to assess long-term endothelial cell loss (ECL) and graft failure with Descemet membrane endothelial keratoplasty (DMEK) and Descemet stripping endothelial keratoplasty (DSEK) versus penetrating keratoplasty (PK) performed for the same indications (primarily Fuchs dystrophy and pseudophakic corneal edema) in the Cornea Donor Study. METHODS: This retrospective study included consecutive primary DMEK (529 recipients, 739 eyes) and DSEK cases (585 recipients, 748 eyes) with 1 or more endothelial cell density (ECD) measurements at 6 months to 16 years. Main outcomes were ECD, longitudinal ECL, and graft failure. RESULTS: Between 6 months and 8 years the ECD declined linearly by approximately 118 cells/mm2/yr after DMEK and 112 cells/mm2/yr after DSEK. Beyond 8 years postoperatively the rate of decline slowed substantially. Selective dropout from graft failure did not significantly affect the ECD trend. At 10 years, median ECL (interquartile range) was 63% (45, 73) with DMEK, 68% (48, 78) with DSEK, and 76% (70, 82) with PK (P = 0.01 DMEK vs. DSEK, P <0.001 DMEK vs. PK, and P < 0.001 DSEK vs. PK). The proportion of surviving grafts with 10-year ECD <500 cells/mm2 was 1.4% with DMEK, 7.3% with DSEK, and 23.9% with PK. The cumulative risk of graft failure between 6 months and 10 years was 5% with DMEK, 11% with DSEK, and 19% with PK (P < 0.001). CONCLUSIONS: Compared with PK and DSEK, DMEK had significantly lower ECL and significantly lower risk of secondary graft failure through 10 years.

Trends in Ophthalmological Patents, 2005-2020.

Purpose: Technological development drives the optimization of therapeutics in ophthalmology, but quantifiable and systematic review of such innovation is lacking. To fill this gap, we characterize trends in ophthalmology-related patents in the United States from 2005 to 2020. Methods: Publicly available patent data from the US Patent and Trademark Office was analyzed with the R programming language. Ophthalmology-related patents were identified with a keyword search of their titles and claims text. Temporal trends were assessed with the Mann-Kendall trend test (α = 0.05, two-sided). Results: Of 4.5 million collected patents, some 21,000 (0.5%) were ophthalmology related. The number of annually granted ophthalmology patents increased over time (Mann-Kendall test: z = 4.91; P < 0.001), from 619 patents released in 2005 to 2,019 patents in 2020. Patent counts also increased over time for all ophthalmic subspecialties except oculoplastics, with steepest rises in retina (z = 4.91; P < 0.001) and cornea (z = 4.64; P < 0.001). The most cited patents were in biocompatible intraocular implants and implantable controlled-release drug delivery systems, which underscores particular advancement in therapeutic efficacy and safety in devices used in the treatment and management of common yet debilitating eye conditions. Conclusion: This exploratory analysis reveals hotspots for ophthalmology-related innovation in the United States that may predict current and future growth trends in device development and pharmacologic advancement in ophthalmology, paving the way for more diverse and effective treatment options for preserving vision.

Association Between Fuchs Endothelial Corneal Dystrophy, Diabetes Mellitus, and Multimorbidity.

PURPOSE: The aim of this study was to assess risk for demographic variables and other health conditions that are associated with Fuchs endothelial corneal dystrophy (FECD). METHODS: We developed a FECD case-control algorithm based on structured electronic health record data and confirmed accuracy by individual review of charts at 3 Veterans Affairs (VA) Medical Centers. This algorithm was applied to the Department of VA Million Veteran Program cohort from whom sex, genetic ancestry, comorbidities, diagnostic phecodes, and laboratory values were extracted. Single-variable and multiple variable logistic regression models were used to determine the association of these risk factors with FECD diagnosis. RESULTS: Being a FECD case was associated with female sex, European genetic ancestry, and a greater number of comorbidities. Of 1417 diagnostic phecodes evaluated, 213 had a significant association with FECD, falling in both ocular and nonocular conditions, including diabetes mellitus (DM). Five of 69 laboratory values were associated with FECD, with the direction of change for 4 being consistent with DM. Insulin dependency and type 1 DM raised risk to a greater degree than type 2 DM, like other microvascular diabetic complications. CONCLUSIONS: Female sex, European ancestry, and multimorbidity increased FECD risk. Endocrine/metabolic clinic encounter codes and altered patterns of laboratory values support DM increasing FECD risk. Our results evoke a threshold model in which the FECD phenotype is intensified by DM and potentially other health conditions that alter corneal physiology. Further studies to better understand the relationship between FECD and DM are indicated and may help identify opportunities for slowing FECD progression.

Long-Term Endothelial Safety Profile With iStent Inject in Patients With Open-Angle Glaucoma.

PURPOSE: To report 5-year postoperative safety data of iStent inject, including overall stability, endothelial cell density (ECD), and endothelial cell loss (ECL) in patients with mild-to-moderate primary open-angle glaucoma (POAG). DESIGN: 5-year follow-up safety study of the prospective, randomized, single-masked, concurrently controlled, multicenter iStent inject pivotal trial. METHODS: In this 5-year follow-up safety study of the 2-year iStent inject pivotal randomized controlled trial, patients receiving iStent inject placement and phacoemulsification or phacoemulsification alone were studied for the incidence of clinically relevant complications associated with iStent inject placement and stability. Corneal endothelial endpoints were mean change in ECD from screening and proportion of patients with >30% ECL from screening, from analysis of central specular endothelial images by a central image analysis reading center at several time points through 60 months postoperatively. RESULTS: Of the 505 original randomized patients, 227 elected to participate (iStent inject and phacoemulsification group, n = 178; phacoemulsification-alone control group, n = 49). No specific device-related adverse events or complications were reported through month 60. No significant differences were observed in mean ECD, mean percentage change in ECD, or proportion of eyes with >30% ECL between the iStent inject and control groups at any time point; mean percentage decrease in ECD at 60 months was 14.3% ± 13.4% in the iStent inject group and 14.8% ± 10.3% in the control group (P = .8112). The annualized rate of ECD change from 3 to 60 months was neither clinically nor statistically significant between groups. CONCLUSIONS: Implantation of iStent inject during phacoemulsification in patients with mild-to-moderate POAG did not produce any device-related complications or ECD safety concerns compared to phacoemulsification alone through 60 months.

Machine Learning Analysis of Postkeratoplasty Endothelial Cell Images for the Prediction of Future Graft Rejection.

Purpose: This study developed machine learning (ML) classifiers of postoperative corneal endothelial cell images to identify postkeratoplasty patients at risk for allograft rejection within 1 to 24 months of treatment. Methods: Central corneal endothelium specular microscopic images were obtained from 44 patients after Descemet membrane endothelial keratoplasty (DMEK), half of whom had experienced graft rejection. After deep learning segmentation of images from all patients' last and second-to-last imaging, time points prior to rejection were analyzed (175 and 168, respectively), and 432 quantitative features were extracted assessing cellular spatial arrangements and cell intensity values. Random forest (RF) and logistic regression (LR) models were trained on novel-to-this-application features from single time points, delta-radiomics, and traditional morphometrics (endothelial cell density, coefficient of variation, hexagonality) via 10 iterations of threefold cross-validation. Final assessments were evaluated on a held-out test set. Results: ML classifiers trained on novel-to-this-application features outperformed those trained on traditional morphometrics for predicting future graft rejection. RF and LR models predicted post-DMEK patients' allograft rejection in the held-out test set with >0.80 accuracy. RF models trained on novel features from second-to-last time points and delta-radiomics predicted post-DMEK patients' rejection with >0.70 accuracy. Cell-graph spatial arrangement, intensity, and shape features were most indicative of graft rejection. Conclusions: ML classifiers successfully predicted future graft rejections 1 to 24 months prior to clinically apparent rejection. This technology could aid clinicians to identify patients at risk for graft rejection and guide treatment plans accordingly. Translational Relevance: Our software applies ML techniques to clinical images and enhances patient care by detecting preclinical keratoplasty rejection.

Comparison of Graft Outcomes Reusing Original Intermediate-Term Cold Storage Solution for Entire Corneal Donor Storage Period With Exchanged Fresh Storage Solution After Donor Preparation in the Cornea Preservation Time Study.

PURPOSE: The purpose of this study was to evaluate outcomes of Descemet stripping automated endothelial keratoplasty comparing exchange with fresh intermediate-term cold storage solution after lenticule preparation versus reuse of the original solution in the Cornea Preservation Time Study. METHODS: In the Cornea Preservation Time Study, 508 donor corneas had lenticules prepared by the eye bank with fresh solution exchange (Fresh group) and 283 with reuse of the original solution (Original group). Storage time for all donors was ≤11 days. Graft success rates, central endothelial cell loss at 3 years, and frequency of positive donor rim cultures were compared between the 2 groups. RESULTS: The 3-year graft success rate (95% confidence interval) was 93.4% (90.7%-95.3%) in the Fresh group and 95.2% (91.8%-97.2%) in the Original group (adjusted hazard ratio for graft failure = 0.64, 95% confidence interval, 0.33-1.24, P = 0.19). The mean percentage endothelial cell loss was significantly greater in the Fresh group versus Original group (45% ± 22% vs. 38% ± 20%, respectively, P = 0.004). Cultures were positive in 4 (1.5%) of 267 donor rims (3 fungal and 1 bacterial) in the Fresh group and in 4 (2.5%) of 158 in the Original group ( P = 0.57). There were 2 postoperative infections in the Original group and none in the Fresh group. CONCLUSIONS: The use of the original intermediate-term cold storage solution did not reduce the 3-year graft success rate compared with exchanging with fresh solution after lenticule preparation for Descemet stripping automated endothelial keratoplasty, while the frequency of positive donor rim cultures did not significantly differ between groups.

Comparison of Donor Corneal Endothelial Cell Density Determined by Eye Banks and by a Central Image Analysis Reading Center Using the Same Image Analysis Method.

PURPOSE: The purpose of this study was to evaluate agreement between eye banks (EBs) and an image analysis reading center on endothelial cell density (ECD) determinations using the same image analysis method. METHODS: The Cornea Image Analysis Reading Center (CIARC) determined ECD with a single experienced analyst on EB-obtained central endothelial images from donors intended for keratoplasty from 2 eye banks, Eversight and Lions VisionGift, using the Konan center analysis method. The EBs performed ECD determination on their respective sets of images using the same analysis method with experienced eye bank technicians. RESULTS: The mean age of the 200 donors was 54 years (range 30-75 years). Seventy (35%) of the 200 patients were women, and 57 (29%) were diabetic. The mean ECD was 10 cells/mm2 greater by the EBs than by CIARC (P = 0.39), with 95% limits of agreement of [-304 to 323 cells/mm2]. The mean difference was not substantially changed when the difference between EBs and CIARC ECD was adjusted for sex, donor age, donor diabetes, CV, HEX, number of cells analyzed, and EBs as a random effect (estimated mean difference of 20 cells/mm2 after adjustment in a linear mixed model; P = 0.73). The EB-determined preoperative ECD was within 10% of the CIARC-determined ECD for 178 (89%) image sets, with 15 (8%) higher by >10% and 7 (3%) lower by >10%. CONCLUSIONS: Well-trained eye bank technicians achieve comparable results for ECD determination with an experienced image analyst from an image analysis reading center when the same image analysis method is used.

Random Survival Forests Analysis of Intraoperative Complications as Predictors of Descemet Stripping Automated Endothelial Keratoplasty Graft Failure in the Cornea Preservation Time Study.

Importance: A new analytic method can evaluate factors of interest associated with graft failure after Descemet stripping automated endothelial keratoplasty (DSAEK) or more generally in any ophthalmic surgical setting with a time-to-event outcome. Objective: To reanalyze types of intraoperative complications associated with DSAEK graft failure in the Cornea Preservation Time Study using random survival forests. Design, Setting, and Participants: This cohort study, initially conceived in April 2019, used a prediction model to conduct a post hoc secondary analysis of data collected in a multicenter, double-masked, randomized clinical trial. Forty US clinical sites with 70 surgeons participated, with donor corneas provided by 23 US eye banks. The study included 1090 participants, representing 1330 eyes, undergoing DSAEK for Fuchs dystrophy (1255 eyes [94.4%]) or pseudophakic or aphakic corneal edema (75 eyes [5.6%]). Enrollment occurred between April 16, 2012, and February 20, 2014, and follow-up ended June 5, 2017. Statistical analysis was performed from July 10, 2019, to May 29, 2020. Intervention: Descemet stripping automated endothelial keratoplasty with random assignment of a donor cornea with preservation time of 7 days or less or 8 to 14 days. Main Outcomes and Measures: Ranked variable importance for intraoperative complications among 50 donor, recipient, and eye bank variables and restricted mean survival time through 47 months (1434 days) after DSAEK were examined. Random survival forests, a nonparametric method (with less restrictive model assumptions) that is far more flexible in its ability to model nonlinear effects and interactions, was used to analyze the data. Results: This study included 1090 participants (663 women [60.8%]; median age, 70 years [range, 42-90 years]), representing 1330 eyes. Random survival forests ranked a DSAEK intraoperative complication as the third most predictive factor of graft failure, after surgeon and eye bank, in the final model with 5 predictors. In the first 47 months after DSAEK, the estimated mean difference in restricted mean survival time for grafts that experienced a DSAEK intraoperative complication vs those that did not was -227 days (99% CI, -352 to -70 days) based on the final RSF model. Conclusions and Relevance: These findings, while post hoc, support the hypothesis that random survival forests allow for an improved analytic approach for identifying factors predictive of graft failure and for obtaining adjusted graft survival estimates. Random survival forests offer the opportunity to guide the development of future population-based cohort ophthalmic surgical studies, establishing definitive factors for procedural success.

The Impact of Virtual Interviews on the Geographic Distribution of Ophthalmology Match Results in the 2020-2021 Cycle.

Background Instead of the traditional in-person interviews, the 2020 to 2021 ophthalmology application cycle was conducted with virtual interviews due to coronavirus disease 2019 (COVID-2019). Little is known about differences between the results of this application cycle with previous years. Objectives The aim of this study was to determine the effect of virtual interviews on the geographic distribution of matched ophthalmology residency applicants. Methods Information was collected on the medical school location and matched residency program location for 2020 to 2021 applicants as well as applicants during the 2016 to 2017, 2017 to 2018, and 2018 to 2019 cycles from publicly available Web sites. Pearson chi-squared tests were conducted to determine whether there was a significant difference in the proportion of applicants matching in the same region, state, and institution as their medical schools in the 2020 to 2021 interview cycle when compared with past cycles. Results Three-hundred seventy-five applicants from 2020 to 2021 and 1,190 applicants from 2016 to 2019 application cycles were analyzed. There was no difference in the type of medical school attended (allopathic vs. osteopathic vs. international medical graduate) ( p = 0.069), the likelihood of attending a residency program in the same region as the home medical school (54% for 2020-2021 vs. 57% for 2016-2019 applicants, p = 0.3), and the likelihood of attending a residency program in the same state as the home medical school (31 vs. 28%, p = 0.2). There was a higher likelihood of applicants during the 2020 to 2021 cycle matching at a residency program affiliated with their home medical school than previous cycles (23 vs. 18%, p = 0.03). Conclusions Virtual interviews did not increase the likelihood of medical students staying in the same region or state as their medical school, while there was a higher likelihood of applicants matching at residency programs at institutions affiliated with their medical schools. A hybrid approach to maintain geographic diversity of applicants' final residency programs involving virtual interviews with the addition of in-person away rotations is suggested.

Acute corneal edema decades after penetrating keratoplasty for keratoconus in eyes wearing scleral contact lenses.

PURPOSE: To report three cases of acute corneal edema occurring decades after penetrating keratoplasty (PK) for keratoconus in eyes wearing scleral contact lenses (ScCLs) with previously clear corneal grafts. METHODS: Retrospective chart review of three ScCL wearers presenting for sudden onset pain and blurred vision. Data extracted included clinical presentation, year and reason for PK, ocular medications and comorbidities, contact lens wearing history, results of any ancillary testing available including corneal topography, anterior segment optical coherence tomography (OCT), and specular microscopy surrounding the event, treatment and outcomes of intervention. The number of PK eyes fit with ScCLs in the author's practice was determined to estimate the prevalence of this event. RESULTS: The three patients each had a longstanding PK for keratoconus performed between 33-35 years prior to presentation and recurrent ectasia. Each patient presented with an acute, painful eye and reduced vision either 3 days, 4 months or 9 years after refitting into ScCLs. Each eye had well demarcated focal microcystic epithelial and stromal edema within the graft and crossing the wound margin onto the host cornea. Although a definitive break or detachment of Descemet's membrane was not visualized, the presentations suggest these were episodes of acute hydrops. CONCLUSIONS: Longstanding PKs with recurrent ectasia and acute focal edema suggestive of corneal hydrops is demonstrated in this case series of ScCL wearers. Although similar events have occurred as part of the natural history of post-PK corneas for keratoconus, the proximity of ScCL refitting to two of the events suggests some association.

Quantitative and qualitative evaluation of deep learning automatic segmentations of corneal endothelial cell images of reduced image quality obtained following cornea transplant.

We are developing automated analysis of corneal-endothelial-cell-layer, specular microscopic images so as to determine quantitative biomarkers indicative of corneal health following corneal transplantation. Especially on these images of varying quality, commercial automated image analysis systems can give inaccurate results, and manual methods are very labor intensive. We have developed a method to automatically segment endothelial cells with a process that included image flattening, U-Net deep learning, and postprocessing to create individual cell segmentations. We used 130 corneal endothelial cell images following one type of corneal transplantation (Descemet stripping automated endothelial keratoplasty) with expert-reader annotated cell borders. We obtained very good pixelwise segmentation performance (e.g., Dice coefficient = 0.87 ± 0.17 , Jaccard index = 0.80 ± 0.18 , across 10 folds). The automated method segmented cells left unmarked by analysts and sometimes segmented cells differently than analysts (e.g., one cell was split or two cells were merged). A clinically informative visual analysis of the held-out test set showed that 92% of cells within manually labeled regions were acceptably segmented and that, as compared to manual segmentation, automation added 21% more correctly segmented cells. We speculate that automation could reduce 15 to 30 min of manual segmentation to 3 to 5 min of manual review and editing.

Comparative corneal endothelial cell toxicity of differing intracameral moxifloxacin doses after phacoemulsification.

PURPOSE: To determine whether intracameral moxifloxacin 500 µg is noninferior to 250 µg for central endothelial cell loss (ECL) after phacoemulsification. SETTING: Aravind Eye Care System. DESIGN: Prospective masked randomized study. METHODS: Eyes with bilateral nuclear cataracts, central endothelial cell density (ECD) of more than 2000 cells/mm, and ECD not differing between eyes by more than 200 cells/mm underwent phacoemulsification at least 14 days apart. Intraoperatively, the first eye was randomized to receive either a 500 or 250 µg dose of moxifloxacin intracamerally and received the other dose for the second-eye surgery. Postoperative course was monitored at 1 day, 1 week, 1 month, and 3 months. Preoperative and 30-day and 90-day postoperative central ECD was determined by a reading center for a masked analysis of ECL at 3 months postoperatively. RESULTS: Fifty eyes of 25 patients (aged 48 to 69 years) underwent uneventful surgery and had normal postoperative courses. The point estimate (PE) and 95% CI for the mean difference in % ECL between the 500 µg and 250 µg doses at 3 months postoperatively was 0.8% (-5.8%, 7.4%). Upon identifying and removing 2 outliers, noninferiority was proven with a mean difference of the PE, -2.2% (CI, -6.5%, 2.1%). CONCLUSIONS: Clinical and corneal endothelial cell were comparable in this study population for the 250 µg and 500 µg doses of intracameral moxifloxacin. Both doses were well tolerated clinically, supporting the use of the higher dose for improved antimicrobial coverage for the prevention of postoperative endophthalmitis.

Machine learning for segmenting cells in corneal endothelium images.

Images of the endothelial cell layer of the cornea can be used to evaluate corneal health. Quantitative biomarkers extracted from these images such as cell density, coefficient of variation of cell area, and cell hexagonality are commonly used to evaluate the status of the endothelium. Currently, fully-automated endothelial image analysis systems in use often give inaccurate results, while semi-automated methods, requiring trained image analysis readers to identify cells manually, are both challenging and time-consuming. We are investigating two deep learning methods to automatically segment cells in such images. We compare the performance of two deep neural networks, namely U-Net and SegNet. To train and test the classifiers, a dataset of 130 images was collected, with expert reader annotated cell borders in each image. We applied standard training and testing techniques to evaluate pixel-wise segmentation performance, and report corresponding metrics such as the Dice and Jaccard coefficients. Visual evaluation of results showed that most pixel-wise errors in the U-Net were rather non-consequential. Results from the U-Net approach are being applied to create endothelial cell segmentations and quantify important morphological measurements for evaluating cornea health.