Deep learning-based automatic meibomian gland segmentation and morphology assessment in infrared meibography.

Meibomian glands (MG) are large sebaceous glands located below the tarsal conjunctiva and the abnormalities of these glands cause Meibomian gland dysfunction (MGD) which is responsible for evaporative dry eye disease (DED). Accurate MG segmentation is a key prerequisite for automated imaging based MGD related DED diagnosis. However, Automatic MG segmentation in infrared meibography is a challenging task due to image artifacts. A deep learning-based MG segmentation has been proposed which directly learns MG features from the training image dataset without any image pre-processing. The model is trained and evaluated using 728 anonymized clinical meibography images. Additionally, automatic MG morphometric parameters, gland number, length, width, and tortuosity assessment were proposed. The average precision, recall, and F1 score were achieved 83%, 81%, and 84% respectively on the testing dataset with AUC value of 0.96 based on ROC curve and dice coefficient of 84%. Single image segmentation and morphometric parameter evaluation took on average 1.33 s. To the best of our knowledge, this is the first time that a validated deep learning-based approach is applied in MG segmentation and evaluation for both upper and lower eyelids.

Fibrillar Layer as a Marker for Areas of Pronounced Corneal Endothelial Cell Loss in Advanced Fuchs Endothelial Corneal Dystrophy.

PURPOSE: We sought to assess the correlation of corneal endothelial cell (CEC) density to alterations of collagen composition of Descemet membrane (DM) in advanced Fuchs endothelial corneal dystrophy (FECD) and to image such changes by slit-lamp biomicroscopy in vivo. DESIGN: Prospective, observational consecutive case series. METHODS: Fifty eyes (50 subjects) with advanced FECD were enrolled. After slit-lamp biomicroscopy and corneal Scheimpflug imaging, the Descemet endothelium complex (DEC) was retrieved during DM endothelial keratoplasty (DMEK) surgery. The expression of collagens I, III, and IV (COL I, COL III, and COL IV) and corresponding CEC density were analyzed by immunofluorescence flat mount-staining. Presence, diameter and surface area of collagen expression, and CEC density served as the main outcome measures. RESULTS: Immunofluorescence staining revealed central coherent collagen positive areas (mean surface area = 10 mm2 ± 6 mm2) corresponding to a fibrillar layer burying the guttae of DM in 84% (42/50) of DECs. CEC density overlying the fibrillar layer compared with the periphery was significantly reduced (-54.8%, P < .0001) with a steep decline of CEC density at its borders. Subgroup analysis revealed that the fibrillar layer may be imaged by slit-lamp biomicroscopy in vivo with significant positive correlation of mean maximum diameter detected by slit-lamp biomicroscopy (dSL max = 4.1 mm ± 0.9 mm) and by immunofluorescence staining (dIF max = 4.7 mm ± 1.1 mm; r = 0.76; P = .001). CONCLUSION: A fibrillar layer with a clear geographic pattern marks areas of pronounced loss of CEC density in advanced FECD eyes and may be imaged by slit-lamp biomicroscopy in vivo.

New Technologies in Clinical Trials in Corneal Diseases and Limbal Stem Cell Deficiency: Review from the European Vision Institute Special Interest Focus Group Meeting.

To discuss and evaluate new technologies for a better diagnosis of corneal diseases and limbal stem cell deficiency, the outcomes of a consensus process within the European Vision Institute (and of a workshop at the University of Cologne) are outlined. Various technologies are presented and analyzed for their potential clinical use also in defining new end points in clinical trials. The disease areas which are discussed comprise dry eye and ocular surface inflammation, imaging, and corneal neovascularization and corneal grafting/stem cell and cell transplantation. The unmet needs in the abovementioned disease areas are discussed, and realistically achievable new technologies for better diagnosis and use in clinical trials are outlined. To sum up, it can be said that there are several new technologies that can improve current diagnostics in the field of ophthalmology in the near future and will have impact on clinical trial end point design.

Clinical Dry Eye and Meibomian Gland Features Among Dry Eye Patients in Rural and Urban Ghana.

AIM: To compare differences in clinical dry eye features and meibomian gland health status between dry eye patients from rural and urban populations in Ghana. METHODS: We examined 211 (rural=109, urban=102) participants with subjective dry eye symptoms. Tear film break -up time (TBUT), Schirmer's test and ocular surface staining (OSS) were assessed. Symptoms were evaluated using the SPEED II questionnaire. Meibomian glands (MG) in the right eye upper (UL) and lower lids (LL) were imaged using a custom meibographer. MG area was determined by intensity threshold segmentation using Image J software. MG loss (MGL) was also graded based on Pult's grading scheme. Mann-Whitney, Spearman correlation, chi-square and odds analyses were performed; p<0.05 was considered significant. RESULTS: Rural participants showed greater SPEED scores, reduced TBUT, and lower Schirmer scores, p <0.05. The proportion of rural participants with MGL were significantly more (82.3%) than urban participants (63.3%), p <0.05. They also showed greater MGL than urban participants, p <0.05. Chi-square test revealed significantly different meiboscale distributions (UL: χ2 =13.58, LL: χ2 =15.29) between the groups, p <0.05. Overall significant relationships were observed between MGL and age [rs= 0.61], OSS [rs= 0.35], TBUT [rs= -0.52], and Schirmer scores [rs= -0.40], p <0.05. CONCLUSION: The data suggest that the participants from the rural population have worse dry eye and meibomian gland health status than those from the urban population. The significant relationships between the various clinical variables suggest important links between MGD and DED. Subtle differences in the everyday working and living environment could likely account for the differences in the severity of DED and MGD between the two groups. And considering the increased pattern of urbanization, industrialization and modernization and the related environmental effects in Africa, future longitudinal studies on specific environmental risk factors or mediators of DED and MGD are necessary to ascertain the MGD and DED situation in Ghana and Africa at large.

Analysis of peripapillary vessel density and Bruch's membrane opening-based neuroretinal rim parameters in glaucoma using OCT and OCT-angiography.

PURPOSE: To compare peripapillary vessel density (VD) measured by spectral domain optical coherence tomography angiography (SD-OCT-A) with morphometric parameters assessing the neuroretinal rim based on Bruch's membrane opening (BMO) by spectral domain optical coherence tomography (SD-OCT) in glaucoma. METHODS: In this non-interventional cohort study, 50 eyes of 25 consecutively enrolled patients with diagnosis of glaucoma underwent SD-OCT and SD-OCT-A imaging of the optic nerve head (ONH). BMO minimum rim width (BMO-MRW) and area (BMO-MRA) as well as peripapillary retinal nerve fiber layer (RNFL) thickness were compared to peripapillary VD in the RNFL layer around the ONH. RESULTS: Mean BMO-MRW was 221.46 ± 81.5 µm, mean BMO-MRA was 1.05 ± 0.04 mm2, mean RNFL thickness was 72.46 ± 23.16 µm, and mean VD was 43.8 ± 11.4%. VD was significantly lower when morphometric parameters had lower thickness values (p < 0.01). Correlation coefficients and their 95%- confidence intervals (95%-CI) with VD were r = 0.53 (95%-CI: 0.21-0.77) for BMO-MRW, r = 0.55 (95%-CI: 0.21-0.77) for BMO-MRA, and r = 0.57 (95%-CI: 0.13-0.73) for RNFL thickness. Intra-individual VD in both eyes correlated with r = 0.72 (p < 0.001), mean VD was comparable (p = 0.6). Eyes with high global RNFL thickness (>90 µm) showed less VD variance (σ2 = 48.1) compared to eyes with highly reduced RNFL thickness (<65 µm; σ2 = 82.0). Best corrected visual acuity, perimetric mean defect, and PSD correlated significantly with VD (95%-CI: -0.66 to -0.10, 0.16 to 0.6, and -0.65 to -0.02, respectively). CONCLUSIONS: Peripapillary vessel density measured by SD-OCT angiography correlates significantly with Bruch's membrane opening-based parameters measured by SD-OCT in glaucoma patients.

Immune reactions after modern lamellar (DALK, DSAEK, DMEK) versus conventional penetrating corneal transplantation.

In the past decade, novel lamellar keratoplasty techniques such as Deep Anterior Lamellar Keratoplasty (DALK) for anterior keratoplasty and Descemet stripping automated endothelial keratoplasty (DSAEK)/Descemet membrane endothelial keratoplasty (DMEK) for posterior keratoplasty have been developed. DALK eliminates the possibility of endothelial allograft rejection, which is the main reason for graft failure after penetrating keratoplasty (PK). Compared to PK, the risk of endothelial graft rejection is significantly reduced after DSAEK/DMEK. Thus, with modern lamellar techniques, the clinical problem of endothelial graft rejection seems to be nearly solved in the low-risk situation. However, even with lamellar grafts there are epithelial, subepithelial and stromal immune reactions in DALK and endothelial immune reactions in DSAEK/DMEK, and not all keratoplasties can be performed in a lamellar fashion. Therefore, endothelial graft rejection in PK is still highly relevant, especially in the "high-risk" setting, where the cornea's (lymph)angiogenic and immune privilege is lost due to severe inflammation and pathological neovascularization. For these eyes, currently available treatment options are still unsatisfactory. In this review, we will describe currently used keratoplasty techniques, namely PK, DALK, DSAEK, and DMEK. We will summarize their indications, provide surgical descriptions, and comment on their complications and outcomes. Furthermore, we will give an overview on corneal transplant immunology. A specific focus will be placed on endothelial graft rejection and we will report on its incidence, clinical presentation, and current/future treatment and prevention options. Finally, we will speculate how the field of keratoplasty and prevention of corneal allograft rejection will develop in the future.

Assessment of meibomian glands using a custom-made meibographer in dry eye patients in Ghana.

BACKGROUND: Meibomian Gland Dysfunction (MGD) is a leading cause of evaporative Dry Eye Disease (DED). This makes non-invasive meibography an important procedure in the clinical evaluation of DED patients. Our purpose was to conduct a lead-off investigation focused on the practicality of performing meibography in a developing country, with limited access to complex ophthalmic imaging systems, using a custom meibographer, as a step to future comparative studies on meibomian glands and DED in Africa. METHODS: Meibomian glands(MG) in 76 upper eyelids (UL) and 49 lower eyelids (LL) in 1 eye each of 125 patients randomly selected from a patient population presenting with subjective DED symptoms at a clinic were photographed using a custom meibographer. Single frames were captured, and the MG area determined by intensity threshold segmentation and area calculation using Image J software. MG loss (MGL) was quantified by outlining its area and expressing it as a percentage of the total MG per Pult's grading scheme. Dry eye measures included Tear Film Break Up - Time (TUBT), Schirmer's test and Ocular Surface Staining (OSS). Symptoms were evaluated using the SPEED II questionnaire. Correlations between MGL and age, ocular signs and symptoms were analyzed by Pearson's. Differences between comparable groups were analyzed by Mann - Whitney test; p < 0.05 was considered significant. RESULTS: Overall mean MGL was 32.10% ± 25.0% (26.25% ± 22.40% for UL and 40.33% ± 26.70% for LL). MGL correlated significantly with age [r = 0.91, p = 0.001], SPEED scores [r = 0.90, p = 0.001], OSS [r = 0.75, p = 0.001] and TBUT [r = - 0.81, p = 0.001]. MGL scores were significantly higher in the UL than LL [U = 1293.5 p = 0.004]. CONCLUSION: This study for the first time presents data on the status of Meibomian glands in Africa. It furthermore suggests that it is feasible to examine Meibomian glands using a custom meibographer in developing countries with limited access to complex imaging systems. It also demonstrates the benefit and cost-effectiveness of a simple device by the observed significant relations between meibomian gland loss and DED in these patients.

UV light crosslinking regresses mature corneal blood and lymphatic vessels and promotes subsequent high-risk corneal transplant survival.

Immunologic graft rejection is the main complication after corneal transplant into pathologically prevascularized so-called high-risk eyes. The aim of this study was to evaluate whether ultraviolet (UV) light crosslinking can regress pathologic corneal blood and lymphatic vessels and thereby improve subsequent graft survival. Using the murine model of suture-induced corneal neovascularization, we found that corneal crosslinking with UVA light and riboflavin regressed both preexisting blood and lymphatic vessels significantly via induction of apoptosis in vascular endothelial cells. In addition, macrophages and CD45+ cell counts were significantly reduced. Consistently, corneal crosslinking reduced keratocyte density and corneal thickness without affecting corneal nonvascular endothelial cells, iris, and lens depending on the crosslinking duration. Furthermore, using the murine model of corneal transplant, long-term graft survival was significantly promoted (P < .05) and CD4+ CD25+ FoxP3+ T regulatory cells were upregulated (P < .01) in high-risk eyes preoperatively treated with crosslinking. Our results suggest UV light crosslinking as a novel method to regress both pathologic corneal blood and lymphatic vessels and to reduce CD45+ inflammatory cells. Furthermore, this study demonstrates for the first time that preoperative corneal crosslinking in prevascularized high-risk eyes can significantly improve subsequent graft survival and may become a promising novel therapy in the clinic.

Intraoperative changes in corneal structure during excimer laser phototherapeutic keratectomy (PTK) assessed by intraoperative optical coherence tomography.

PURPOSE: Excimer laser phototherapeutic keratectomy (PTK) is a safe treatment for superficial corneal opacities, e.g., in corneal dystrophies or degenerations. Nevertheless, no standardized treatment protocols are available and intraoperative monitoring was not possible, so far. Here we evaluate the potential benefits of the intraoperative assessment by microscope-integrated intraoperative optical coherence tomography (MI-OCT) of corneal optical properties during PTK. METHODS: Retrospective study of eight patients (one male, seven females; age range, 43-80 years, mean = 66.1 years) using an 840-nm microscope-integrated spectral-domain OCT (iOCT; OptoMedical Technologies, Luebeck, Germany adapted to HS Hi-R Neo 900A, Haag Streit Surgical, Wedel, Germany). Images were acquired before and after corneal abrasion and after PTK. For PTK, a SCHWIND Amaris 750S excimer laser (SCHWIND eye-tech-solutions GmbH und KO. KG) was used. Parameters assessed were the central corneal thickness (CCT), changes in central depth-dependent corneal tissue intensity (TI), and corneal surface roughness (SR) in cross-sectional images of the cornea. RESULTS: Intraoperative monitoring using microscope-integrated OCT was possible in all patients at all time points. TI of the anterior corneal stroma decreased significantly (p = 0.037) after PTK (T1 = 15.1 ± 3.6, T2 = 15.0 ± 3.84, T3 = 13.7 ± 3.38), but not after corneal abrasion alone, indicating increased transparency caused by excimer laser PTK. CCT was significantly lower after corneal abrasion (p = 0.017), but not after PTK (T1 = 630.4 ± 70 µm, T2 = 544.1 ± 59.4 µm, T3 = 558.3 ± 52.5 µm. SR significantly decreased (p = 0.043) after PTK (T1 = 614.4 ± 37.5 pixels, T2 = 634.4 ± 35.6 pixels, T3 = 611.0 ± 40.3 pixels). CONCLUSIONS: Intraoperative OCT allows real-time imaging during PTK and the assessment of corneal optical transparency and its surface roughness. It has to be clarified in larger studies if these parameters correlate with later postoperative visual outcomes.

Novel Method to Detect Corneal Lymphatic Vessels In Vivo by Intrastromal Injection of Fluorescein.

PURPOSE: Corneal lymphatic vessels are clinically invisible because of their thin walls and clear lymph fluid. There is no easy and established method for in vivo imaging of corneal lymphatic vessels so far. In this study, we present a novel approach to visualize corneal lymphatic vessels in vivo by injecting intrastromal fluorescein sodium. METHODS: Six- to eight-week-old female BALB/c mice were used in the mouse model of suture-induced corneal neovascularization. Two weeks after the suture placement, fluorescein sodium was injected intrastromally. The fluorescein, taken up by the presumed lymphatic vessels, was then tracked using a clinically used Spectralis HRA + OCT device. Immunohistochemistry staining with specific lymphatic marker LYVE-1 and pan-endothelial marker CD31 was used to confirm the indirect lymphangiography findings. RESULTS: By injecting fluorescein intrastromally, both corneal blood and lymphatic vessels were detected. While the lymphatic vessels were visible as bright vessel-like structures using HRA, the blood vessels appeared as dark networks. Fluorescein-labeled lymphatic vessels were colocalized with LYVE-1 in immunohistochemically stained sections of the same specimen. CONCLUSIONS: Corneal lymphatic vessels can be easily imaged in vivo in the murine model using intrastromal fluorescein injection.

Label-Free In Vivo Imaging of Corneal Lymphatic Vessels Using Microscopic Optical Coherence Tomography.

Purpose: Corneal neovascularization, in particular lymphangiogenesis, is a limiting factor in corneal transplant survival. Novel treatment approaches focus on (selective) inhibition and regression of lymphatic vessels. Imaging clinically invisible corneal lymphatic vessels is a prerequisite for these strategies. Using a murine model, this study investigates whether corneal lymphatic vessels can be imaged using microscopic optical coherence tomography (mOCT). Methods: Corneal neovascularization was induced by intrastromal placement of 11.0 nylon sutures in one eye of BALB/c mice. After 2 weeks, cross-sectional images and volumes of the corneas with a 0.5 mm lateral and axial field of view were acquired using a custom-built mOCT system enabling a resolution of 1 µm at a B-scan rate of 165/s. Three of the six animals received an additional intrastromal injection of India ink 24 hours before the measurement to stain the corneal lymphatic system in vivo. Immunohistochemistry using CD31 and LYVE-1 was used to validate the mOCT findings. Results: Using mOCT, lymphatic vessels were visible as dark vessel-like structures with the lumen lacking a hyperreflective wall and mostly lacking cells. However, individual, slowly moving particles, which most likely are immune cells, occasionally could be observed inside the lumen. In lymphatic vessels of ink-stained corneas, hyperreflection and shadowing underneath was observed. Ink-filled lymphatic vessels were colocalized in consecutive corneal flat mounts of the same specimen. Conclusions: Corneal lymphatic vessels can be imaged using mOCT. This novel approach opens new options for noninvasive clinical imaging of corneal lymphatic vessels for diagnostic and therapeutic indications.

Noncontact holographic detection for photoacoustic tomography.

A holographic method for high-speed, noncontact photoacoustic tomography is introduced and evaluated. Relative changes of the object's topography, induced by the impact of thermoelastic pressure waves, were determined at nanometer sensitivity without physical contact. The object's surface was illuminated with nanosecond laser pulses and imaged with a high-speed CMOS camera. From two interferograms measured before and after excitation of the acoustic wave, surface displacement was calculated and then used as the basis for a tomographic reconstruction of the initial pressure caused by optical absorption. The holographic detection scheme enables variable sampling rates of the photoacoustic signal of up to 50 MHz. The total acquisition times for complete volumes with 230 MVoxel is far below 1 s. Measurements of silicone and porcine skin tissue phantoms with embedded artificial absorbers, which served as a model for human subcutaneous vascular networks, were possible. Three-dimensional reconstructions of the absorbing structures show details with a diameter of 310 µm up to a depth of 2.5 mm. Theoretical limitations and the experimental sensitivity, as well as the potential for in vivo imaging depending on the detection repetition rate, are analyzed and discussed.

Transient Ingrowth of Lymphatic Vessels into the Physiologically Avascular Cornea Regulates Corneal Edema and Transparency.

Lymphangiogenesis is essential for fluid homeostasis in vascularized tissues. In the normally avascular cornea, however, pathological lymphangiogenesis mediates diseases like corneal transplant rejection, dry eye disease, and allergy. So far, a physiological role for lymphangiogenesis in a primarily avascular site such as the cornea has not been described. Using a mouse model of perforating corneal injury that causes acute and severe fluid accumulation in the cornea, we show that lymphatics transiently and selectively invade the cornea and regulate the resolution of corneal edema. Pharmacological blockade of lymphangiogenesis via VEGFR-3 inhibition results in increased corneal thickness due to delayed drainage of corneal edema and a trend towards prolonged corneal opacification. Notably, lymphatics are also detectable in the cornea of a patient with acute edema due to spontaneous Descemet´s (basement) membrane rupture in keratoconus, mimicking this animal model and highlighting the clinical relevance of lymphangiogenesis in corneal fluid homeostasis. Together, our findings provide evidence that lymphangiogenesis plays an unexpectedly beneficial role in the regulation of corneal edema and transparency. This might open new treatment options in blinding diseases associated with corneal edema and transparency loss. Furthermore, we demonstrate for the first time that physiological lymphangiogenesis also occurs in primarily avascular sites.

Full-field speckle interferometry for non-contact photoacoustic tomography.

A full-field speckle interferometry method for non-contact and prospectively high speed Photoacoustic Tomography is introduced and evaluated as proof of concept. Thermoelastic pressure induced changes of the objects topography are acquired in a repetitive mode without any physical contact to the object. In order to obtain high acquisition speed, the object surface is illuminated by laser pulses and imaged onto a high speed camera chip. In a repetitive triple pulse mode, surface displacements can be acquired with nanometre sensitivity and an adjustable sampling rate of e.g. 20 MHz with a total acquisition time far below one second using kHz repetition rate lasers. Due to recurring interferometric referencing, the method is insensitive to thermal drift of the object due to previous pulses or other motion. The size of the investigated area and the spatial and temporal resolution of the detection are scalable. In this study, the approach is validated by measuring a silicone phantom and a porcine skin phantom with embedded silicone absorbers. The reconstruction of the absorbers is presented in 2D and 3D. The sensitivity of the measurement with respect to the photoacoustic detection is discussed. Potentially, Photoacoustic Imaging can be brought a step closer towards non-anaesthetized in vivo imaging and new medical applications not allowing acoustic contact, such as neurosurgical monitoring or burnt skin investigation.

Refractive index determination of transparent samples by noniterative phase retrieval.

We present a simple method to determine the refractive indices of transparent specimens. The refractive index of an object under investigation is received by evaluating the optical path difference introduced by the object, while taking into account geometric parameters. The optical path difference that corresponds to the phase distribution is obtained by a noninterferometric, noniterative phase retrieval method based on Green's functions. It will be shown that this technique is a highly accurate and quantitative method for refractive index determination.