An AS-OCT image dataset for deep learning-enabled segmentation and 3D reconstruction for keratitis.

Infectious keratitis is among the major causes of global blindness. Anterior segment optical coherence tomography (AS-OCT) images allow the characterizing of cross-sectional structures in the cornea with keratitis thus revealing the severity of inflammation, and can also provide 360-degree information on anterior chambers. The development of image analysis methods for such cases, particularly deep learning methods, requires a large number of annotated images, but to date, there is no such open-access AS-OCT image repository. For this reason, this work provides a dataset containing a total of 1168 AS-OCT images of patients with keratitis, including 768 full-frame images (6 patients). Each image has associated segmentation labels for lesions and cornea, and also labels of iris for full-frame images. This study provides a great opportunity to advance the field of image analysis on AS-OCT images in both two-dimensional (2D) and three-dimensional (3D) and would aid in the development of artificial intelligence-based keratitis management.

Diagnostic Potential of Anterior Segment Optical Coherence Tomography Scans for Pseudomonas Keratitis.

Purpose: The purpose of this study was to investigate the diagnostic value of anterior segment optical coherence tomography (AS-OCT) scans for Pseudomonas keratitis. Methods: Patients with treatment-naïve keratitis underwent AS-OCT imaging. The following parameters were evaluated: corneal thickness (CT), infiltrate thickness (IT), infiltrate diameter (ID), tissue loss/gain, entropy, and distance of the lesion from the corneal pupillary center. Three different OCT devices were used for the analysis. The relationship between the detected pathogen and the OCT patterns was analyzed. Results: Nineteen eyes of 19 patients were included in the analysis: seven cases in the Pseudomonas group and 12 cases in the Gram-positive group. The mean (SD) values for the Pseudomonas and Gram-positive groups, respectively, were as follows: CT, 834 µm (165 µm) and 760 µm (120 µm); IT, 290 µm (152 µm) and 287 µm (84 µm); ID, 2067 µm (1470 µm) and 1307 µm (745 µm); distance to center, 3.0 mm (1.2 mm) and 3.0 mm (1.6 mm); epithelial defect, 1193 µm (586 µm) and 484 µm (615 µm); tissue gain, +31% (19%) and +10% (12%); and entropy level, 4.0 (0.8) and 3.9 (1.1). Conclusions: This study introduces novel insights by identifying specific OCT parameters that distinguish Pseudomonas keratitis, including a 30% tissue gain. These findings align with earlier research that underscores the potential of OCT in differentiating various pathogens causing keratitis. Translational Relevance: The findings of this study could be used to develop new diagnostic strategies for Pseudomonas keratitis. The OCT findings could be used to develop new biomarkers for the infection.

Treatment of immune-mediated keratitis (IMMK) in dogs with immunosuppressants observed with spectral domain optical coherence tomography (SD-OCT).

Two dogs presented with bilateral pattern-forming corneal opacity. Treatment with topical immunosuppressants was initiated after a complete ophthalmic examination. The response to treatment was assessed by analyzing serial images using slit-lamp biomicroscopy and spectral domain optical coherence tomography (SD-OCT). Both dogs responded to topical immunosuppressants; however, the lesions recurred once the treatment was abated or withdrawn. The most effective immunosuppressant in both dogs was 0.03% tacrolimus ointment. Early and continuous treatment with topical immunosuppressants may be necessary to improve corneal clarity and prevent scarring. SD-OCT could provide useful structural information regarding presumed immune-mediated keratitis and aid in monitoring treatment response.

A Structure-Aware Convolutional Neural Network for Automatic Diagnosis of Fungal Keratitis with In Vivo Confocal Microscopy Images.

Fungal keratitis (FK) is a common and severe corneal disease, which is widely spread in tropical and subtropical areas. Early diagnosis and treatment are vital for patients, with confocal microscopy cornea imaging being one of the most effective methods for the diagnosis of FK. However, most cases are currently diagnosed by the subjective judgment of ophthalmologists, which is time-consuming and heavily depends on the experience of the ophthalmologists. In this paper, we introduce a novel structure-aware automatic diagnosis algorithm based on deep convolutional neural networks for the accurate diagnosis of FK. Specifically, a two-stream convolutional network is deployed, combining GoogLeNet and VGGNet, which are two commonly used networks in computer vision architectures. The main stream is used for feature extraction of the input image, while the auxiliary stream is used for feature discrimination and enhancement of the hyphae structure. Then, the features are combined by concatenating the channel dimension to obtain the final output, i.e., normal or abnormal. The results showed that the proposed method achieved accuracy, sensitivity, and specificity of 97.73%, 97.02%, and 98.54%, respectively. These results suggest that the proposed neural network could be a promising computer-aided FK diagnosis solution.

A case of Candida albicans-induced fungal keratitis in a Chihuahua dog - with a focus on optical coherence tomographic features.

A 4-year-old Chihuahua dog was referred for bilateral corneal ulcers. Slightly raised white fluorescein-positive plaque-like corneal lesions in both eyes appeared as intense hyperreflective areas with posterior shadowing on optical coherence tomography (OCT). Based on corneal cytology and culture, Candida albicans-induced fungal keratitis was diagnosed. Despite treatment, on OCT, endothelial plaques, increased stromal infiltration thickness, vertical shapes of the ulcer edge, and necrotic stromal space were judged to be aggravation of the disease, and surgery was performed. Conjunctival grafting surgery with topical 1% voriconazole effectively resolved fungal keratitis. OCT can provide detailed and objective information related to the disease prognosis.

Class-Aware Attention Network for infectious keratitis diagnosis using corneal photographs.

Infectious keratitis is one of the common ophthalmic diseases and also one of the main blinding eye diseases in China, hence rapid and accurate diagnosis and treatment for infectious keratitis are urgent to prevent the progression of the disease and limit the degree of corneal injury. Unfortunately, the traditional manual diagnosis accuracy is usually unsatisfactory due to the indistinguishable visual features. In this paper, we propose a novel end-to-end fully convolutional network, named Class-Aware Attention Network (CAA-Net), for automatically diagnosing infectious keratitis (normal, viral keratitis, fungal keratitis, and bacterial keratitis) using corneal photographs. In CAA-Net, a class-aware classification module is first trained to learn class-related discriminative features using separate branches for each class. Then, the learned class-aware discriminative features are fed into the main branch and fused with other feature maps using two attention strategies to assist the final multi-class classification performance. For the experiments, we have built a new corneal photograph dataset with 1886 images from 519 patients and conducted comprehensive experiments to verify the effectiveness of our proposed method. The code is available at https://github.com/SWF-hao/CAA-Net_Pytorch.

Subbasal corneal nerve damage in patients with bacterial keratitis: in vivo confocal microscopy study.

PURPOSE: To examine subbasal corneal nerve changes in patients with bacterial infectious keratitis using in vivo confocal microscopy. METHODS: Thirteen patients (13 eyes) with unilateral bacterial keratitis and 12 healthy controls were prospectively enrolled in the study. In vivo confocal microscopy was performed in all the patients at 2 time points, in the acute phase of infectious keratitis and at 28 ± 0.6 months after resolution of the infection. RESULTS: The subbasal nerve length was 5.15 ± 1.03 mm/mm2 during the acute phase of bacterial keratitis (compared with that of the controls: 19.02 ± 1.78 mm/mm2, p<0.05). Despite the significant corneal nerve regeneration over the interval of 28 months after the resolution of the infection, the nerve density was still significantly reduced as compared with that of the controls (9.73 ± 0.93 mm/mm2, p<0.05). Moreover, in vivo confocal microscopy images showed diffuse high-reflecting areas referring to the scar tissue areas with thin and tortuous nerve branches regenerating toward these areas. CONCLUSIONS: A partial corneal nerve regeneration of subbasal nerve plexus during the first 28 months after the acute phase of infectious keratitis was observed. Moreover, the regenerated nerves of the patients remained morphologically altered as compared with those of the healthy controls. These results may be relevant to the clinical follow-up and surgical planning for these patients.

Comparisons of deep learning algorithms for diagnosing bacterial keratitis via external eye photographs.

Bacterial keratitis (BK), a painful and fulminant bacterial infection of the cornea, is the most common type of vision-threatening infectious keratitis (IK). A rapid clinical diagnosis by an ophthalmologist may often help prevent BK patients from progression to corneal melting or even perforation, but many rural areas cannot afford an ophthalmologist. Thanks to the rapid development of deep learning (DL) algorithms, artificial intelligence via image could provide an immediate screening and recommendation for patients with red and painful eyes. Therefore, this study aims to elucidate the potentials of different DL algorithms for diagnosing BK via external eye photos. External eye photos of clinically suspected IK were consecutively collected from five referral centers. The candidate DL frameworks, including ResNet50, ResNeXt50, DenseNet121, SE-ResNet50, EfficientNets B0, B1, B2, and B3, were trained to recognize BK from the photo toward the target with the greatest area under the receiver operating characteristic curve (AUROC). Via five-cross validation, EfficientNet B3 showed the most excellent average AUROC, in which the average percentage of sensitivity, specificity, positive predictive value, and negative predictive value was 74, 64, 77, and 61. There was no statistical difference in diagnostic accuracy and AUROC between any two of these DL frameworks. The diagnostic accuracy of these models (ranged from 69 to 72%) is comparable to that of the ophthalmologist (66% to 74%). Therefore, all these models are promising tools for diagnosing BK in first-line medical care units without ophthalmologists.

Medical image management and analysis system based on web for fungal keratitis images.

The medical image management and analysis system proposed in this paper is a medical software developed by the Browser/Server (B/S) architecture after investigating the workflow of the relevant departments of the hospital, which realizes the entire process of patients from consultation to printing of reports. The computer-aided diagnosis function is added based on image management. Due to the difficulty in collecting medical image data, in the computer-aided diagnosis module, this paper only uses the common fungal keratitis collected from the hospital in the laboratory. Focused microscope images are used for experiments. First, the images were trained with three convolutional neural networks, AlexNet, ZFNet, and VGG16. These models which classify fungal keratitis were obtained and integrated was performed to obtain better classification results. Finally, the model was integrated with the system designed in this paper, which realized the automatic diagnosis of Confocal Microscopy (CM) images of fungal keratitis online and provided it to medical staff for reference. The system can improve the work efficiency of the image-related departments while reducing the workload of doctors in the department to manually read the films.

A deep transfer learning framework for the automated assessment of corneal inflammation on in vivo confocal microscopy images.

PURPOSE: Infiltration of activated dendritic cells and inflammatory cells in cornea represents an important marker for defining corneal inflammation. Deep transfer learning has presented a promising potential and is gaining more importance in computer assisted diagnosis. This study aimed to develop deep transfer learning models for automatic detection of activated dendritic cells and inflammatory cells using in vivo confocal microscopy images. METHODS: A total of 3453 images was used to train the models. External validation was performed on an independent test set of 558 images. A ground-truth label was assigned to each image by a panel of cornea specialists. We constructed a deep transfer learning network that consisted of a pre-trained network and an adaptation layer. In this work, five pre-trained networks were considered, namely VGG-16, ResNet-101, Inception V3, Xception, and Inception-ResNet V2. The performance of each transfer network was evaluated by calculating the area under the curve (AUC) of receiver operating characteristic, accuracy, sensitivity, specificity, and G mean. RESULTS: The best performance was achieved by Inception-ResNet V2 transfer model. In the validation set, the best transfer system achieved an AUC of 0.9646 (P<0.001) in identifying activated dendritic cells (accuracy, 0.9319; sensitivity, 0.8171; specificity, 0.9517; and G mean, 0.8872), and 0.9901 (P<0.001) in identifying inflammatory cells (accuracy, 0.9767; sensitivity, 0.9174; specificity, 0.9931; and G mean, 0.9545). CONCLUSIONS: The deep transfer learning models provide a completely automated analysis of corneal inflammatory cellular components with high accuracy. The implementation of such models would greatly benefit the management of corneal diseases and reduce workloads for ophthalmologists.

Open-Source Automatic Segmentation of Ocular Structures and Biomarkers of Microbial Keratitis on Slit-Lamp Photography Images Using Deep Learning.

We propose a fully-automatic deep learning-based algorithm for segmentation of ocular structures and microbial keratitis (MK) biomarkers on slit-lamp photography (SLP) images. The dataset consisted of SLP images from 133 eyes with manual annotations by a physician, P1. A modified region-based convolutional neural network, SLIT-Net, was developed and trained using P1's annotations to identify and segment four pathological regions of interest (ROIs) on diffuse white light images (stromal infiltrate (SI), hypopyon, white blood cell (WBC) border, corneal edema border), one pathological ROI on diffuse blue light images (epithelial defect (ED)), and two non-pathological ROIs on all images (corneal limbus, light reflexes). To assess inter-reader variability, 75 eyes were manually annotated for pathological ROIs by a second physician, P2. Performance was evaluated using the Dice similarity coefficient (DSC) and Hausdorff distance (HD). Using seven-fold cross-validation, the DSC of the algorithm (as compared to P1) for all ROIs was good (range: 0.62-0.95) on all 133 eyes. For the subset of 75 eyes with manual annotations by P2, the DSC for pathological ROIs ranged from 0.69-0.85 (SLIT-Net) vs. 0.37-0.92 (P2). DSCs for SLIT-Net were not significantly different than P2 for segmenting hypopyons (p > 0.05) and higher than P2 for WBCs (p < 0.001) and edema (p < 0.001). DSCs were higher for P2 for segmenting SIs (p < 0.001) and EDs (p < 0.001). HDs were lower for P2 for segmenting SIs (p = 0.005) and EDs (p < 0.001) and not significantly different for hypopyons (p > 0.05), WBCs (p > 0.05), and edema (p > 0.05). This prototype fully-automatic algorithm to segment MK biomarkers on SLP images performed to expectations on an exploratory dataset and holds promise for quantification of corneal physiology and pathology.

Anterior Segment Optical Coherence Tomography in the Early Management of Microbial Keratitis: A Cross-Sectional Study.

INTRODUCTION: Infectious keratitis is an important cause of visual loss. The purpose of this study was to investigate anterior segment optical coherence tomography patterns in infectious keratitis and evaluate the role of this tool in the early management of this disorder. MATERIAL AND METHODS: In this cross-sectional study, we included patients with proven infectious keratitis, either by culture or therapeutic trial. Subjects underwent baseline anterior segment optical coherence tomography (Spectralis® anterior segment module, Heidelberg Engineering, Germany) performed by the same operator. We used anterior segment optical coherence tomography vertical and horizontal raster default scans with 6.0 mm scan lines. RESULTS: Twenty-five patients (14 men and 11 women) were included. The most common risk factors identified were ocular trauma (11 cases) and contact lens wear (7 cases). Fifteen patients presented bacterial infection; three, fungal infection; two parasitic infection; and five cases presented a negative microbiological culture. Anterior segment optical coherence tomography depicted nine distinct morphological patterns. DISCUSSION: Anterior segment optical coherence tomography allows the depth of corneal involvement to be assessed. When the only patterns identified were hyperreflective stromal lesion and stromal edema, the visual outcome was better. Cystic spaces were present in severe bacterial keratitis. CONCLUSION: Anterior segment optical coherence tomography can complement biomicroscopy, allowing for a better characterization of corneal involvement at presentation that can help in staging and providing useful prognostic information.

Introdução: As queratites infeciosas são uma causa importante de perda da acuidade visual. O objetivo deste estudo foi investigar padrões das queratites infeciosas na tomografia de coerência ótica do segmento anterior e avaliar o seu papel na orientação precoce desta patologia. Material e Métodos: Neste estudo transversal, incluímos doentes com queratite infeciosa comprovada, por cultura microbiológica ou prova terapêutica. Os doentes foram submetidos a tomografia de coerência ótica do segmento anterior na baseline (Spectralis® anterior segment module, Heidelberg Engineering, Germany) executados pelo mesmo operador. Adquiriram-se cortes verticais e horizontais de 6,0 mm com tomografia de coerência ótica spectral domain. Resultados: Foram incluídos 25 doentes (14 homens e 11 mulheres). Os fatores de risco mais comummente identificados foram o trauma ocular (11 casos) e o uso de lentes de contato (sete casos). Quinze doentes apresentaram infeção bacteriana, três infeção fúngica, dois infeção parasitária e cinco casos apresentaram culturas microbiológicas negativas. Distinguiram-se nove padrões morfológicos na tomografia de coerência ótica do segmento anterior. Discussão: A tomografia de coerência ótica do segmento anterior permite avaliar a profundidade do envolvimento corneano na queratite infeciosa. Quando os únicos padrões identificados eram a lesão estromal hiper-refletiva e edema estromal o outcome visual foi melhor. Os espaços quísticos observaram-se nas queratites bacterianas graves. Conclusão: A tomografia de coerência ótica do segmento anterior permite complementar a biomicroscopia, melhorar o estadiamento e fornecer informação prognóstica útil nas queratites infeciosas.

Antenatal Findings of Keratitis-Ichthyosis-Deafness Syndrome.

BACKGROUND: Keratitis-ichthyosis-deafness (KID) syndrome is a congenital ectodermal disorder characterized by keratitis, ichthyosis, and deafness. This syndrome affects multiple systems and can be fatal. CASE: A 34-year-old G2, P1 woman was admitted to the Ege University School of Medicine in Izmir, Turkey because of a rapid increase in abdominal circumference at 32 weeks gestation. Fetal anatomic screening revealed complete chorioamniotic separation, hypoplasia of the cerebellar vermis, and dysmorphic facial findings such as frontal bulging. After the delivery, the baby's whole body had granular thickened skin. Bilateral dry eye, corneal edema, and bilateral retinopathy of prematurity were diagnosed. CONCLUSION: This case report highlights the importance of prenatal diagnosis through ultrasonography and magnetic resonance imaging. This is the first case report that has antenatal ultrasonographic features in the literature.

Cellular morphological changes detected by laser scanning in vivo confocal microscopy associated with clinical outcome in fungal keratitis.

HRT3 in vivo confocal microscopy (IVCM) images may indicate clinical outcome, but few studies have analysed this in fungal keratitis (FK). Adults with FK (diameter ≥3 mm) presenting to Aravind Eye Hospital, India from 2012-3 were enrolled prospectively. IVCM was performed at baseline, days 7, 14 and 21 post-enrolment (+/- 3 days where possible). Specific morphologies were identified in IVCM images by a grader masked to microbiology and clinical outcome (defined as good: healed/improving, or poor: enlarged ulcer, perforation or transplant/glue). Associations with final visit outcome assessed using logistic regression. 143 FK participants were enrolled; 87 had good outcome, 56 had poor outcome. Poor outcomes were associated with stellate interconnected cellular processes with no visible nuclei (OR 2.28, 95% CI: 1.03-5.06, p = 0.043) in baseline IVCM images, and fungal filaments (OR 6.48, 95% CI:2.50-16.78, p < 0.001) or honeycomb distribution of inflammatory cells (OR 5.24, 95% CI: 1.44-19.06, p = 0.012) in final visit images. Fungal filaments (OR 3.61, 95% CI:1.64-7.95, p = 0.001), stromal dendritiform cells (OR 2.88, 95% CI:1.17-7.11, p = 0.022), or stellate cellular processes with no visible nuclei (OR 2.09, 95% CI:1.14-3.82, p = 0.017) were associated with poor outcome if not in baseline but present in final visit images. IVCM can reveal morphological changes associated with clinical outcome.

Interstitial keratitis diagnosis and treatment.

Interstitial keratitis is a non-ulcerative, non-suppurative, more or less vascularized inflammation of the corneal stroma. The corneal lesions result from the host response to bacterial, viral (40% of cases) or parasitic antigens, or from an autoimmune response (1% of cases) without active corneal infection. The natural history of the disease is divided into two phases: acute and cicatricial. This type of keratitis is less common than ulcerative bacterial keratitis, but it is a non-negligible cause of visual loss. It is associated with systemic or infectious disease and requires early diagnosis and appropriate treatment to optimize visual prognosis and avoid other complications.

Longitudinal Morphometric Analysis of Sub-Basal Nerve Plexus in Contralateral Eyes of Patients with Unilateral Neurotrophic Keratitis.

Objectives: To investigate longitudinally corneal sub-basal nerve plexus (SNP) by means of in vivo confocal microscopy (IVCM) in the contralateral eye (CE) of patients with unilateral neurotrophic keratitis (NK) secondary to central nervous system (CNS) diseases who underwent different treatments. Methods: Ten patients with NK and 10 matched controls were included. In 7 NK patients, conservative treatment maintained unchanged the clinical picture over the 1-year follow-up (Group 1), while NK progressed in 3 patients who underwent direct corneal neurotization (Group 2). IVCM scans of SNP of NK patients were acquired in CE at baseline (V0) ad after 1-year follow-up (V1). All images were analyzed with the automated software "ACCMetrics" and compared with controls. The following IVCM corneal nerve parameters were calculated at V0 and V1 with ACCMetrics: fiber density (CNFD), branch density (CNBD), fiber length (CNFL), total branch density (CTBD), fiber area (CNFA), fiber width (CNFW), and fractal dimension (CNFrD). Results: At V0, significantly lower mean values of CNFD and CNBD, and higher values of CNFW were detected in CE of NK patients compared to controls (respectively, 16.9 ± 8.7 vs 25.0 ± 8.3 n/mm2, P= .029; 19.3 ± 13.8 vs 33.8 ± 18.9 n/mm2, P= .023; 0.022 ± 0.002 vs 0.020 ± 0.001 mm/mm2, P< .001). From V0 to V1, all IVCM metrics of CE remained unchanged in Group 1, while they improved in Group 2. Conclusions: Contralateral eye of patients with unilateral NK secondary to CNS disease showed lower CNFD and CNBD and higher CNFW compared to controls. Unlike conservative treatment, direct corneal neurotization was able to improve SNP metrics also in CE.

Sterile keratitis following standard corneal collagen crosslinking: A case series and literature review.

Standard corneal collagen crosslinking (S-CXL) is a safe, approved procedure, but it may result in severe pain, early vision loss and possible complications, such as infectious or sterile keratitis, in some cases. We describe four cases of sterile infiltrates after uneventful S-CXL for keratoconus, from diagnosis to medical management with six months of follow-up, reporting their pathophysiological features, and comparing our findings with published reports. We discuss various possibilities for diagnosing sterile infiltration more rapidly. In terms of the pathophysiology of sterile infiltrate formation, we separated our patients into two types, one with sterile infiltrate from an antigen reaction and the other with sterile infiltrate due to excessive scarring. Early local steroid treatment resulted in a good visual outcome in our cases.

In vivo confocal microscopy of the inflamed anterior segment: A review of clinical and research applications.

In vivo confocal microscopy (IVCM) allows non-invasive imaging of the living human cornea, specifically enabling the detection of immune cells in the healthy and diseased ocular anterior segment. Studies using IVCM have provided insight into the effects of contact lens wear on corneal Langerhans cell density and morphology, and the effects of eye drops on conjunctiva-associated lymphoid tissue. IVCM has also been shown to be a useful adjunctive diagnostic tool in distinguishing infective and non-infective uveitis and in diagnosing atypical infective keratitis. In the research setting, this technology has enhanced our understanding of the role of inflammatory cells in corneal neuropathy and angiogenesis. In vivo-ex vivo correlation using animal models has helped overcome some of the difficulties in identifying cell type on IVCM images. As highlighted in this review, currently there are multiple established, and emerging, clinical and research applications for IVCM in the inflamed anterior segment.