A Deep Feature Fusion of Improved Suspected Keratoconus Detection with Deep Learning.

Detection of early clinical keratoconus (KCN) is a challenging task, even for expert clinicians. In this study, we propose a deep learning (DL) model to address this challenge. We first used Xception and InceptionResNetV2 DL architectures to extract features from three different corneal maps collected from 1371 eyes examined in an eye clinic in Egypt. We then fused features using Xception and InceptionResNetV2 to detect subclinical forms of KCN more accurately and robustly. We obtained an area under the receiver operating characteristic curves (AUC) of 0.99 and an accuracy range of 97-100% to distinguish normal eyes from eyes with subclinical and established KCN. We further validated the model based on an independent dataset with 213 eyes examined in Iraq and obtained AUCs of 0.91-0.92 and an accuracy range of 88-92%. The proposed model is a step toward improving the detection of clinical and subclinical forms of KCN.

A Hybrid Deep Learning Construct for Detecting Keratoconus From Corneal Maps.

Purpose: To develop and assess the accuracy of a hybrid deep learning construct for detecting keratoconus (KCN) based on corneal topographic maps. 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.

Association between visual field damage and corneal structural parameters.

The main goal of this study is to identify the association between corneal shape, elevation, and thickness parameters and visual field damage using machine learning. A total of 676 eyes from 568 patients from the Jichi Medical University in Japan were included in this study. Corneal topography, pachymetry, and elevation images were obtained using anterior segment optical coherence tomography (OCT) and visual field tests were collected using standard automated perimetry with 24-2 Swedish Interactive Threshold Algorithm. The association between corneal structural parameters and visual field damage was investigated using machine learning and evaluated through tenfold cross-validation of the area under the receiver operating characteristic curves (AUC). The average mean deviation was - 8.0 dB and the average central corneal thickness (CCT) was 513.1 µm. Using ensemble machine learning bagged trees classifiers, we detected visual field abnormality from corneal parameters with an AUC of 0.83. Using a tree-based machine learning classifier, we detected four visual field severity levels from corneal parameters with an AUC of 0.74. Although CCT and corneal hysteresis have long been accepted as predictors of glaucoma development and future visual field loss, corneal shape and elevation parameters may also predict glaucoma-induced visual functional loss.

The effects of riboflavin and ultraviolet light on keratocytes cultured in vitro.

PURPOSE: To culture quiescent human keratocytes and evaluate the effects of ultraviolet light and riboflavin on human corneal keratocytes in vitro. METHODS: Keratocytes were obtained from remaining corneoscleral ring donor corneas previously used in corneal transplant surgeries and cultured in DMEM/F12 with 2% FBS until confluence. Characterization of cultured cells was performed by immunofluorescence analysis for anti-cytokeratin-3, anti-Thy-1, anti-α-smooth muscle actin, and anti-lumican. Immunofluorescence was performed before and after treatment of cultured cells with either ultraviolet light or riboflavin. Corneal stromal cells were covered with collagen (200 µL or 500 µL) and 0.1% riboflavin, and then exposed to ultraviolet light at 370 nm for 30 minutes. After 24 hours, cytotoxicity was determined using MTT colorimetric assays, whereas cell viability was assessed using Hoechst 33342 and propidium iodide. RESULTS: Cell cultures achieved confluence in approximately 20 days. Expression of the lumican was high, whereas no expression of CK3, Thy-1, and α-SMA was observed. After crosslinking, MTT colorimetric assays demonstrated a low toxicity rate, whereas Hoechst 33342/propidium iodide staining demonstrated a low rate of apoptosis and necrosis, respectively, in all collagen-treatment groups. CONCLUSION: Keratocytes can be successfully cultured in vitro and characterized by immunofluorescence using lumican. MTT colorimetric assays, and Hoechst 33342, and propidium iodide staining demonstrated a higher rate of cell death in cells cultured without collagen, indicating collagen protects keratocytes from the cytotoxic effects of ultraviolet light.

The effects of riboflavin and ultraviolet light on keratocytes cultured in vitro / Avaliação do efeito da riboflavina e luz ultravioleta sobre os ceratócitos in vitro

ABSTRACT Purpose: To culture quiescent human keratocytes and evaluate the effects of ultraviolet light and riboflavin on human corneal keratocytes in vitro. Methods: Keratocytes were obtained from remaining corneoscleral ring donor corneas previously used in corneal transplant surgeries and cultured in DMEM/F12 with 2% FBS until confluence. Characterization of cultured cells was performed by immunofluorescence analysis for anti-cytokeratin-3, anti-Thy-1, anti-α-smooth muscle actin, and anti-lumican. Immunofluorescence was performed before and after treatment of cultured cells with either ultraviolet light or riboflavin. Corneal stromal cells were covered with collagen (200 µL or 500 µL) and 0.1% riboflavin, and then exposed to ultraviolet light at 370 nm for 30 minutes. After 24 hours, cytotoxicity was determined using MTT colorimetric assays, whereas cell viability was assessed using Hoechst 33342 and propidium iodide. Results: Cell cultures achieved confluence in approximately 20 days. Expression of the lumican was high, whereas no expression of CK3, Thy-1, and α-SMA was observed. After crosslinking, MTT colorimetric assays demonstrated a low toxicity rate, whereas Hoechst 33342/propidium iodide staining demonstrated a low rate of apoptosis and necrosis, respectively, in all collagen-treatment groups. Conclusion: Keratocytes can be successfully cultured in vitro and characterized by immunofluorescence using lumican. MTT colorimetric assays, and Hoechst 33342, and propidium iodide staining demonstrated a higher rate of cell death in cells cultured without collagen, indicating collagen protects keratocytes from the cytotoxic effects of ultraviolet light.

RESUMO Objetivo: Avaliar o efeito da aplicação da luz ultravioleta e riboflavina sobre ceratócitos da córnea humana in vitro. Métodos: Os ceratócitos foram obtidos a partir das rimas corneoesclerais remanescentes da trepanação de córneas previamente utilizadas em cirurgias de transplante de córnea e cultivadas em meio DMEM/F12 com 2% de FBS até atingir confluência. As culturas de células foram caracterizadas por imunofluorescência com os anticorpos K3 (marcador de células epiteliais), Thy-1 (marcador de fibroblasto) SMA (marcador de miofibroblasto) e Lumican (marcador de ceratócitos). Imunofluorescência também foi feita após o tratamento. As células do estroma da córnea foram cobertas com colágeno (200 µL e 500 µL) e 0,1% de riboflavina e exposta a luz UVA a 370 nm por 30 minutos. Após 24 horas, citotoxicidade foi determinada por ensaio de MTT e a viabilidade celular foi feita por Hoechst 33342/Iodeto de propideo. Resultados: As culturas de células atingiram confluência em aproximadamente 20 dias. Imunofluorescência apontou alta expressão para o marcador de ceratócitos (Lumican) e expressão negativa par os marcadores de células epiteliais (K3), fibroblasto (Thy-1) e miofibroblasto (α-SMA). Após o cross linking a análise de MTT mostrou baixa taxa de toxicidade e com a coloração de Hoechst 33342/Iodeto de propideo baixa taxa de apoptose e necrose respectivamente em todos os grupos que continham colágeno. Conclusão: As culturas de ceratócitos foram obtidas e caracterizadas por imunofluorescência através do marcador Lumican com sucesso. O ensaio de MTT e a coloração por Hoechst 33342 e iodeto de propídio, apresentaram maior índice de morte celular nos grupos que não continham colágeno, provando que protege as células contra os efeitos da luz UVA.

Conjunctival epithelial cells cultivated ex vivo from patients with total limbal stem cell deficiency.

PURPOSE: Reconstruction of the ocular surface is challenging. As an alternative to mucosal and limbal epithelial, we study the feasibility of cultivated human conjunctival epithelial (HCjE) cells of patients with total limbal stem cell deficiency (LSCD). METHODS: We studied superior forniceal conjunctival biopsies harvested from 9 living donors with total LSCD of several etiologies who underwent surgery for ocular surface reconstruction. The conjunctival explants were cultivated on serum and growth factor supplemented DMEM/F12 under submerged conditions on denuded human amniotic membrane and tissue culture dishes. The area of cell growth was assessed. Cell morphology was analyzed by light microscopy, impression cytology, and transmission electron microscopy. Cultures were evaluated for epithelial cytokeratins (CK3, CK19), proliferation marker (Ki-67), and putative stem cells markers (ABCG2 and p63). Confocal immunofluorescence was also performed to assess CK3, CK19, Ki-67, ABCG2, and p63. RESULTS: The HCjE cells cultivated ex vivo were successfully expanded on denuded amniotic membrane but with a slower growth than in the tissue culture dish. Transmission electron microscopy showed stratified epithelium with microvilli, desmosomes, and hemidesmosomes. Impression cytology showed PAS+ cells that resembled goblet cells. Immunocytochemical analysis showed positivity for CK3, CK19, Ki-67, ABCG2, and p63. Confocal immunofluorescence was positive for CK3, CK19, Ki-67, ABCG2, and p63. CONCLUSIONS: Our results showed that it is possible to cultivate HCjE cells ex vivo of patients with ocular surface diseases. This method is important for ocular surface reconstruction in patients with bilateral total LSCD.

Comparative study of conjunctival limbal transplantation not associated with the use of amniotic membrane transplantation for treatment of total limbal deficiency secondary to chemical injury.

PURPOSE: The aim of this study was to compare the outcome of conjunctival limbal transplantation with and without the use of amniotic membrane (AM) transplantation for the treatment of limbal deficiency after chemical burns to the eyes. METHODS: This is a retrospective, comparative, interventional case series that included 34 eyes (34 patients) with total limbal deficiency after chemical burns, who were submitted to conjunctival limbal transplantation. Two groups were formed: group 1 (15 eyes) limbal transplantation associated with AM transplantation and group 2 (19 eyes) only limbal transplantation. Success and failure rates, epithelialization time, and visual acuity were compared between the 2 groups. RESULTS: The mean follow-up period was 19.71 ± 5.6 months in group 1 and 18.26 ± 7.78 months in group 2. The proportion of conjunctival limbal autograft and living-related conjunctival limbal allograft was similar in both groups (P = 0.914). Time until complete epithelialization of the corneal surface was significantly longer in group 1 (P = 0.007). Graft survival was similar in both groups (P = 0.581). Failure rate was also similar in both groups (P = 0.232). The most common cause of failure was transplanted stem cell depletion in group 1 (20% of cases) and rejection in group 2 (10.5%). The visual acuity improved postoperatively in 12 eyes (80%) in group 1 and 16 (84.2%) in group 2 (P = 0.430). CONCLUSIONS: Results with the use of AM associated with conjunctival limbal transplantation in ocular surface reconstruction were similar to those obtained with limbal transplantation alone. Both techniques were found to be satisfactory for ocular surface reconstruction in patients with total limbal deficiency after chemical burns.

Minor salivary glands and labial mucous membrane graft in the treatment of severe symblepharon and dry eye in patients with Stevens-Johnson syndrome.

OBJECTIVE: To evaluate minor salivary glands and labial mucous membrane graft in patients with severe symblepharon and dry eye secondary to Stevens-Johnson syndrome (SJS). METHODS: A prospective, non-comparative, interventional case series of 19 patients with severe symblepharon and dry eye secondary to SJS who underwent labial mucous membrane and minor salivary glands transplantation. A complete ophthalmic examination including the Schirmer I test was performed prior to and following surgery. All patients had a preoperative Schirmer I test value of zero. RESULTS: Nineteen patients with severe symblepharon and dry eye secondary to SJS were included in the study. There was a statistically significant improvement in the best spectacle-corrected visual acuity in eight patients (t test; p=0.0070). Values obtained in the Schirmer I test improved significantly in 14 eyes (73.7%) 6 months following surgery (χ(2) test; p=0.0094). A statistically significant increase in tear production (Schirmer I test) was found in eyes that received more than 10 glands per graft compared with eyes that received fewer glands (χ(2) test; p=0.0096). Corneal transparency improved significantly in 11 (72.2%) eyes and corneal neovascularisation improved significantly in five eyes (29.4%) (McNemar test; p=0.001 and p=0.0005). The symptoms questionnaire revealed improvement in foreign body sensation in 53.6% of the patients, in photophobia in 50.2% and in pain in 54.8% (Kruskal-Wallis test; p=0.0167). CONCLUSION: Labial mucous membrane and minor salivary glands transplantation were found to constitute a good option for the treatment of severe symblepharon and dry eye secondary to SJS. This may be considered as a step prior to limbal stem cell and corneal transplantation in these patients.

Phototherapeutic keratectomy for the treatment of corneal opacities after epidemic keratoconjunctivitis.

PURPOSE: To assess visual results and compare methods of measuring central corneal thickness (CCT) and corneal opacity thickness (COT) in patients with corneal opacities induced by epidemic keratoconjunctivitis (EKC) and treated with phototherapeutic keratectomy (PTK) using low-dose mitomycin C (MMC). DESIGN: Prospective consecutive case series. METHODS: Patients with chronic adenoviral corneal opacity underwent transepithelial PTK with MMC 0.002% for 1 minute. The presence of photophobia, the best spectacle-corrected visual acuity (BSCVA), and the contrast sensitivity were evaluated. CCT measurements were obtained with ultrasound pachymeter (US), ultrasound biomicroscopy (UBM), Scheimpflug tomography (Pentacam Oculus), and optical coherence tomography (OCT Visante). COT measurements were obtained with UBM, Pentacam, and OCT. RESULTS: Thirty-one eyes of 23 patients, comprising 15 women (65.2%) and 8 men (34.8%), mean age 41.8 years, were enrolled in the study. Duration of visual disturbance was 19.1 ± 14 months. The number of patients with photophobia was reduced from 100% to 29% after surgery. BSCVA improved 2 or more lines in 78% of the patients at 12 months. A hyperopic shift of 1.52 ± 0.91 diopters was achieved. Contrast sensitivity improved in both photopic and mesopic conditions. For each of the instruments, the CCT postoperative mean was significantly smaller than the preoperative measurement (P < .0001) and COT values were significantly reduced in comparison to the preoperative values (P < .001). CONCLUSION: Improvements in photophobia, BSCVA, and contrast sensitivity were observed in patients treated using excimer laser PTK with low-dose MMC for subepithelial infiltrates.

Pentacam Scheimpflug evaluation of corneal volume after LASIK.

PURPOSE: To assess the change in corneal volume and laser ablation volume over time after LASIK using a rotating Scheimpflug camera. METHODS: Twenty-six patients (49 eyes) underwent LASIK. Pentacam Scheimpflug measurements were performed pre- and postoperatively at 1, 3, 7, 15, and 30 days. Central corneal thickness, total corneal volume (10-mm diameter), and partial corneal volumes (3, 5, and 7 mm) based on the apex of the cornea, were measured. Main outcome measures were differences between pre- and postoperative volume measurements of total and partial corneal regions, volume changes over time postoperatively, and comparison between laser ablation volume and corneal volume. RESULTS: A strong linear relationship was noted between spherical equivalent refraction and laser ablation volume (r = -0.931, P < .0001). Furthermore, 30 days after LASIK, a strong linear relationship was noted between planned thickness and achieved thickness (r = 0.9457, P < .001). Corneal volumes for 3-, 5-, and 7-mm diameter regions significantly decreased over time postoperatively (one-way analysis of variance, P < .05). CONCLUSIONS: Alterations in corneal volume correlated with laser ablation volume when the diameter of the region considered was 3, 5, or 7 mm from the corneal apex.