Search

Table S1: Pathomechanics of Joubert Syndrome Table S2: Pathomechanics of Bardet Biedl Syndrome Table S3: Pathomechanics of Neuromyelitis Optica Spectrum Disorder, Genome-wide association study; Phenome-wide association study; Genetic diagnosis; Molecular mechanisms; Cerebral visual impairment; Multiple sclerosis; Joubert syndrome; Mowat Wilson disease; Zellweger spectrum disorder; Neuromyelitis optica spectrum disorder

Pseudophakic bullous keratopathy and Fuchs' endothelial dystrophy are the two most common causes of corneal edema after cataract surgery. We report a 61-year-old alcoholic male with bilateral corneal edema that improved after his alcohol abstinence. He had uneventful bilateral cataract surgery 3 years ago and blurred vision in both eyes developed for weeks. As he had no history of endothelial dystrophy, the treatment for viral endotheliitis was used initially yet in vain. We asked him to stop alcohol and adjusted his psychiatric drugs, but he lied about stopping drinking. The corneal edema progressed, and finally, he underwent penetrating keratoplasty in his left eye 1 year later. During hospitalization for surgery, alcohol withdrawal syndrome was noted because he could not drink alcohol in our hospital. After he quit drinking for months, corneal edema in the right eye disappeared. Our case highlights that alcoholism can result in corneal edema, and stopping drinking is necessary in these patients.

Background Diabetic macular edema (DME) is a sight-threatening condition that need regular examinations and remedies. Optical coherence tomography (OCT) is the most common used examination to evaluate the structure and thickness of the macula but the software in the OCT machine does not tell the clinicians whether DME exists directly. Recently, artificial intelligence (AI) is expected to aid in diagnosis generation and therapy selection. We thus develop a smartphone-based offline AI system that provides diagnostic suggestions and medical strategies through analyzing OCT images from diabetic patients at risk of developing DME. Methods DME patients receiving treatments in 2017 at Taipei Veterans General Hospital were included in this study. We retrospectively collected the OCT images of these patients from January 2008 to July 2018. We established the AI model based on MobileNet architecture to classify the OCT images conditions. The confusion matrix has been applied to present the performance of the trained AI model. Results Based on the convolutional neural network with MobileNet model, our AI system achieved a high DME diagnostic accuracy of 90.02%, which is comparable to other AI systems such as InceptionV3 and VGG16. We further developed a mobile-application based on this AI model available at https://aicl.ddns.net/DME.apk. Conclusion We successful integrated an AI model into the mobile device to provide an offline method to provide the diagnosis for quickly screening the risk of developing diabetic macular edema (DME). With the offline property, our model could help those non-ophthalmological health-care providers in offshore islands or underdeveloped countries.

Background Optical coherence tomography (OCT) is considered as a sensitive and noninvasive tool to evaluate the macular lesions. In patients with diabetes mellitus (DM), the existence of diabetic macular edema (DME) can cause significant vision impairment and further intravitreal injection (IVI) of anti-vascular endothelial growth factor (VEGF) is needed. However, the increasing number of DM patients makes it a big burden for clinicians to manually determine whether DME exists in the OCT images. The artificial intelligence (AI) now enormously applied to many medical territories may help reduce the burden on clinicians. Methods We selected DME patients receiving IVI of anti-VEGF or corticosteroid at Taipei Veterans General Hospital in 2017. All macular cross-sectional scan OCT images were collected retrospectively from the eyes of these patients from January 2008 to July 2018. We further established AI models based on convolutional neural network architecture to determine whether the DM patients have DME by OCT images. Results Based on the convolutional neural networks, InceptionV3 and VGG16, our AI system achieved a high DME diagnostic accuracy of 93.09% and 92.82%, respectively. The sensitivity of the VGG16 and InceptionV3 models was 96.48% and 95.15%., respectively. The specificity was corresponding to 86.67% and 89.63% for VGG16 and InceptionV3, respectively. We further developed an OCT-driven platform based on these AI models. Conclusion We successfully set up AI models to provide an accurate diagnosis of DME by OCT images. These models may assist clinicians in screening DME in DM patients in the future.

The mutations in the genes encoding the subunits of complex I of the mitochondrial electron transport chain are the most common cause of Leber’s hereditary optic neuropathy (LHON), a maternal hereditary disease characterized by retinal ganglion cell (RGC) degeneration. The characteristics of incomplete penetrance indicate that nuclear genetic and environmental factors also determine phenotypic expression of LHON. Therefore, further understanding of the role of mutant mitochondrial nicotinamide adenine dinucleotide dehydrogenase subunit proteins and nuclear genetic factors/environmental effects in the etiology of LHON is needed. In this study, we generated human-induced pluripotent stem cells (hiPSCs) from healthy control, unaffected LHON mutation carrier, and affected LHON patient. hiPSC-derived RGCs were used to study the differences between affected and unaffected carriers of mitochondrial DNA point mutation m.11778G > A in the MT-ND4 gene. We found that both mutated cell lines were characterized by increase in reactive oxygen species production, however, only affected cell line had increased levels of apoptotic cells. We found a significant increase in retrograde mitochondria and a decrease in stationary mitochondria in the affected RGC axons. In addition, the messenger RNA and protein levels of KIF5A in the LHON-affected RGCs were significantly reduced. Antioxidant N-acetyl-L-cysteine could restore the expression of KIF5A and the normal pattern of mitochondrial movement in the affected RGCs. To conclude, we found essential differences in the mutually dependent processes of oxidative stress, mitochondrial transport and apoptosis between two LHON-specific mutation carrier RGC cell lines, asymptomatic carrier and disease-affected, and identified KIF5A as a central modulator of these differences.

Background We investigated the performance of glaucoma assessment using RS-3000 spectral domain optical coherence tomography with a long axial length (AL) normative database versus Cirrus HD-OCT in eyes with high myopia. Methods This is a prospective case-control study. Eyes with AL ≥26 mm were enrolled, including 40 control eyes and 41 eyes with primary open-angle glaucoma. Each participant received OCT imaging with both devices at the same visit. We calculated the area under receiver operating characteristic curve (AUROC) for circumpapillary retinal nerve fiber layer (cpRNFL) and macular ganglion cell complex (GCC) parameters, and compared the false-positive and false-negative rates between the two devices. Results Both devices performed comparably well in glaucoma assessment regarding cpRNFL parameters, with the best parameter being the average cpRNFL for RS-3000 (AUROC: 0.899) and the clock-hour 7 cpRNFL for Cirrus HD (AUROC: 0.912). Regarding macular GCC parameters, although the nasal-inferior outer sector of the RS-3000 (AUROC 0.873) and the inferior-temporal sector of the Cirrus HD (AUROC 0.840) performed well in glaucoma assessment, generally speaking there was a higher false-positive rate using Cirrus HD when compared with that of RS-3000. Conclusion For eyes with long AL, both OCT devices are comparable in the ability of discriminating glaucoma from non-glaucoma in terms of cpRNFL parameters, while the macular GCC parameters of RS-3000 were less likely to over-diagnose glaucoma in highly myopic eyes.

Nanodiamonds (NDs) are considered to be relatively safe carbon nanomaterials used for the transmission of DNA, proteins and drugs. The feasibility of utilizing the NDs to deliver CRISPR-Cas9 system for gene editing has not been clearly studied. Therefore, in this study, we aimed to use NDs as the carriers of CRISPR-Cas9 components designed to introduce the mutation in RS1 gene associated with X-linked retinoschisis (XLRS). ND particles with a diameter of 3 nm were functionalized by carboxylation of the surface and covalently conjugated with fluorescent mCherry protein. Two linear DNA constructs were attached to the conjugated mCherry: one encoded Cas9 endonuclease and GFP reporter, another encoded sgRNA and contained insert of HDR template designed to introduce RS1 c.625C>T mutation. Such nanoparticles were successfully delivered and internalized by human iPSCs and mouse retinas, the efficiency of internalization was significantly improved by mixing with BSA. The delivery of ND particles led to introduction of RS1 c.625C>T mutation in both human iPSCs and mouse retinas. Rs1 gene editing in mouse retinas resulted in several pathological features typical for XLRS, such as aberrant photoreceptor structure. To conclude, our ND-based CRISPR-Cas9 delivery system can be utilized as a tool for creating in vitro and in vivo disease models of XLRS. Statement of significance X-linked retinoschisis (XLRS) is a prevalent hereditary retinal disease, which is caused by mutations in RS1 gene, whose product is important for structural organization of the retina. The recent development of genome editing techniques such as CRISPR-Cas9 significantly improved the prospects for better understanding the pathology and development of treatment for this disease. Firstly, gene editing can allow development of appropriate in vitro and in vivo disease models; secondly, CRISPR-Cas9 can be applied for gene therapy by removing the disease-causative mutation in vivo. The major prerequisite for these approaches is to develop safe and efficient CRISPR-Cas9 delivery system. In this study, we tested specifically modified nanodiamonds for such a delivery system. We were able to introduce Rs1 mutation into the mouse retina and, importantly, observed several XLRS-specific effects.

X-linked juvenile retinoschisis (XLRS) is one of the common early-onset hereditary retinal degenerative diseases in men. The common symptoms of XLRS range from mild to severe central vision loss and radial stripes created by the fovea, the division of the inner layer of the retina in the peripheral retina and the significant decrease in b-wave amplitude (ERG). Retinoschisin, the 224-amino-acid protein product of the retinoschisis 1 (RS1) gene, contains a discoid domain as the primary structural unit, an N-terminal cleavable signal sequence, and an oligomerization-area component. Retinoschisin is a homo-octamer complex with disulfide links that are released by retinal cells. It helps preserve the retina's integrity by binding to the surface of photoreceptors and bipolar cells. As a recessive genetic disease, XLRS was usually treated by prescribing low vision aids in most clinical cases. A gene replacement therapy based on adeno-associated virus vectors was initiated and showed a breakthrough in treating XLRS in 2014. Understanding the revolution of gene therapy for treating XLRS may accelerate its development and make this gene therapy the template for developing therapeutics against other inherited retinal diseases.

Acanthamoeba keratitis (AK) is a vision-threatening disease, usually associated with contact lens (CL) wear. As overnight orthokeratology (OOK) is increasingly used to control myopia, we have found incidence of OOK-associated AK is increasing. This study aimed to investigate the clinical presentation and visual outcomes of OOK-associated AK.Demographic characteristics, clinical features, and treatment outcomes were collected by reviewing the medical charts of CL-associated AK patients (n = 35) diagnosed at Taipei Veterans General Hospital from 2001 to 2016. Cases were OOK-associated AK patients (n = 13), and controls were all other CL-associated AK patients (n = 22). Student t tests and chi-square tests were used to compare cases and controls. Linear regression analyses were used to identify factors associated with the final visual outcome in CL-associated AK.OOK-associated AK accounted for half of all CL-associated AK after 2010. OOK-associated AK patients and other CL-associated patients had similar best-corrected logarithm of the minimum angle of resolution visual acuity (BCLVA) before treatment (1.10 ± 0.75 vs 1.13 ± 0.76, p = 0.893), but OOK-associated AK patients were younger (17.15 ± 3.21 vs 26.36 ± 12.81 years, p = 0.004), had less severe disease (ring infiltration, 0% vs 31.82%, p = 0.023), and had better post-treatment BCLVA (0.06 ± 0.15 vs 0.51 ± 0.95, p = 0.041). Multiple linear regression analysis showed that better BCLVA after treatment in CL-associated AK was associated with initial presentation without ring infiltration (p = 0.002) but not with OOK use itself (p = 0.793). Twenty-six of 35 CL-associated AK patients had final BCLVA equal to or better than 0.10 (Snellen visual acuity of 6/7.5). All 13 OOK-associated AK cases were treated with chlorhexidine 0.02% ± voriconazole 1% ± oral voriconazole, and 12 of these patients had final BCLVA equal to or better than 0.10.Most CL-associated AK patients had satisfactory visual outcomes. Half of AK at our hospital is OOK-associated since 2010. Early diagnosis and correct treatment may be the reason why OOK-associated AK patients had better vision prognosis.

Transepithelial photorefractive keratectomy (Trans-PRK), laser-assisted in situ keratomileusis (LASIK), and small incision lenticule extraction (SMILE) are three mainstay refractive surgeries worldwide. The applicability, efficacy, safety, and predictability of these different techniques are quite similar. Trans-PRK has the strongest biostability, earliest return to normal corneal sensitivity but the longest recovery time, most uncomfortable postoperative experience, and possibility of corneal haze. LASIK possesses the fastest visual rehabilitation but the slowest corneal nerve reinnervation, and flap displacement is possibly lifelong. SMILE incurs no flap-related complications and has intermediate vision recovery time and biomechanics compared with Trans-PRK and LASIK. However, it lacks the cyclotorsion compensation system, eye tracking system, and customized treatment profile for high astigmatism or irregular corneal surface. This review aims to introduce the mechanisms, pros, and cons of these three types of refractive surgery. With full understanding, practitioners could advise patients on the most suitable treatment of choice.

Glaucoma is a progressive and irreversible blindness-causing disease. However, the underlying genetic factors and molecular mechanisms remain poorly understood. Previous genome-wide association studies (GWAS) have made tremendous progress on the SNP-based disease association and characterization. However, most of them were conducted for Europeans. Since differential genetic characteristics among ethnic groups were evident in glaucoma, it is worthwhile to complete its genetic landscape from the larger cohorts of Asian individuals. Here, we present a GWAS based on the Taiwan Biobank. Among 1013 glaucoma patients and 36,562 controls, we identified a total of 138 independent glaucoma-associated SNPs at the significance level of p &lt, 1 × 10−5. After clumping genetically linked SNPs (LD clumping), 134 independent SNPs with p &lt, 10−4 were recruited to construct a Polygenic Risk Score (PRS). The model achieved an area under the receiver operating characteristic curve (AUC) of 0.8387 (95% CI = [0.8269–0.8506]), and those within the top PRS quantile had a 45.48-fold increased risk of glaucoma compared with those within the lowest quantile. The PRS model was validated with an independent cohort that achieved an AUC of 0.7283, thereby showing the effectiveness of our polygenic risk score in predicting individuals in the Han Chinese population with higher glaucoma risks.

Artificial intelligence (AI), Internet of Things (IoT), and telemedicine are deeply involved in our daily life and have also been extensively applied in the medical field, especially in ophthalmology. Clinical ophthalmologists are required to perform a vast array of image exams and analyze images containing complicated information, which allows them to diagnose the disease type and grade, make a decision on remedy, and predict treatment outcomes. AI has a great potential to assist ophthalmologists in their daily routine of image analysis and relieve their work burden. However, in spite of these prospects, the application of AI may also be controversial and associated with several legal, ethical, and sociological concerns. In spite of these issues, AI has indeed become an irresistible trend and is widely used by medical specialists in their daily routines in what we can call now, the era of AI. This review will encompass those issues and focus on recent research on the AI application in ophthalmology and telemedicine.

Background Transient receptor potential vanilloid 1 (TRPV1), recognized as a hyperosmolarity sensor, is a crucial ion channel involved in the pathogenesis of neural and glial signaling. Recently, TRPV1 was determined to play a role in retinal physiology and visual transmission. In this study, we sought to clarify the role of TRPV1 and the downstream pathway in the osmotic stress-related retina ganglion cell (RGC) damage. Methods First, we modified the RGC differentiation protocol to obtain a homogeneous RGC population from human induced pluripotent stem cells (hiPSCs). Subsequently, we induced high osmotic pressure in the hiPSC-derived RGCs by administering NaCl solution and observed the behavior of the TRPV1 channel and its downstream cascade. Results We obtained a purified RGC population from the heterogeneous retina cell population using our modified method. Our findings revealed that TRPV1 was activated after 24 h of NaCl treatment. Upregulation of TRPV1 was noted with autophagy and apoptosis induction. Downstream protein expression analysis indicated increased phosphorylation of CREB and downregulated brain-derived neurotrophic factor (BDNF). However, hyperosmolarity-mediated defective morphological change and apoptosis of RGCs, CREB phosphorylation, and BDNF downregulation were abrogated after concomitant treatment with the PKA inhibitor H89. Conclusion Collectively, our study results indicated that the TRPV1–PKA pathway contributed to cellular response under high levels of osmolarity stress; furthermore, the PKA inhibitor had a protective effect on RGCs exposed to this stress. Therefore, our findings may assist in the treatment of eye diseases involving RGC damage.

Artificial intelligence (AI) based on convolutional neural networks (CNNs) has a great potential to enhance medical workflow and improve health care quality. Of particular interest is practical implementation of such AI-based software as a cloud-based tool aimed for telemedicine, the practice of providing medical care from a distance using electronic interfaces. Methods: In this study, we used a dataset of labeled 35,900 optical coherence tomography (OCT) images obtained from age-related macular degeneration (AMD) patients and used them to train three types of CNNs to perform AMD diagnosis. Results: Here, we present an AI- and cloud-based telemedicine interaction tool for diagnosis and proposed treatment of AMD. Through deep learning process based on the analysis of preprocessed optical coherence tomography (OCT) imaging data, our AI-based system achieved the same image discrimination rate as that of retinal specialists in our hospital. The AI platform's detection accuracy was generally higher than 90% and was significantly superior (p < 0.001) to that of medical students (69.4% and 68.9%) and equal (p = 0.99) to that of retinal specialists (92.73% and 91.90%). Furthermore, it provided appropriate treatment recommendations comparable to those of retinal specialists. Conclusions: We therefore developed a website for realistic cloud computing based on this AI platform, available at https://www.ym.edu.tw/~AI-OCT/. Patients can upload their OCT images to the website to verify whether they have AMD and require treatment. Using an AI-based cloud service represents a real solution for medical imaging diagnostics and telemedicine.

Inherited retinal dystrophies (IRDs) are rare but highly heterogeneous genetic disorders that affect individuals and families worldwide. However, given its wide variability, its analysis of the driver genes for over 50% of the cases remains unexplored. The present study aims to identify novel driver genes, disease-causing variants, and retinitis pigmentosa (RP)-associated pathways. Using family-based whole-exome sequencing (WES) to identify putative RP-causing rare variants, we identified a total of five potentially pathogenic variants located in genes OR56A5, OR52L1, CTSD, PRF1, KBTBD13, and ATP2B4. Of the variants present in all affected individuals, genes OR56A5, OR52L1, CTSD, KBTBD13, and ATP2B4 present as missense mutations, while PRF1 and CTSD present as frameshift variants. Sanger sequencing confirmed the presence of the novel pathogenic variant PRF1 (c.124_128del) that has not been reported previously. More causal-effect or evidence-based studies will be required to elucidate the precise roles of these SNPs in the RP pathogenesis. Taken together, our findings may allow us to explore the risk variants based on the sequencing data and upgrade the existing variant annotation database in Taiwan. It may help detect specific eye diseases such as retinitis pigmentosa in East Asia.

Dry eye disease (DED), also called the keratoconjunctivitis sicca, is one of the most common diseases in the ophthalmology clinics. While DED is not a life-threatening disease, life quality may be substantially affected by the discomfort and the complications of poor vision. As such, a large number of studies have made contributions to the investigation of the DED pathogenesis and novel treatments. DED is a multifactorial disease featured with various phenotypic consequences; therefore, animal models are valuable tools suitable for the related studies. Accordingly, selection of the animal model to recapitulate the clinical presentation of interest is important for appropriately addressing the research objective. To this end, we systemically reviewed different murine and rabbit models of DED, which are categorized into the quantitative (aqueous-deficient) type and the qualitative (evaporative) type, based on the schemes to establish. The clinical manifestations of dry eye on animal models can be induced by mechanical or surgical approaches, iatrogenic immune response, topical eye drops, blockage of neural pathway, or others. Although these models have shown promising results, each has its own limitation and cannot fully reproduce the pathophysiological mechanisms that occur in patients. Nonetheless, the animal models remain the best approximation of human DED and represent the valuable tool for the DED studies.

Angiotensin-converting enzyme 2 (ACE2) was identified as the main host cell receptor for the entry of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and its subsequent infection. In some coronavirus disease 2019 (COVID-19) patients, it has been reported that the nervous tissues and the eyes were also affected. However, evidence supporting that the retina is a target tissue for SARS-CoV-2 infection is still lacking. This present study aimed to investigate whether ACE2 expression plays a role in human retinal neurons during SARS-CoV-2 infection. Human induced pluripotent stem cell (hiPSC)-derived retinal organoids and monolayer cultures derived from dissociated retinal organoids were generated. To validate the potential entry of SARS-CoV-2 infection in the retina, we showed that hiPSC-derived retinal organoids and monolayer cultures endogenously express ACE2 and transmembrane serine protease 2 (TMPRSS2) on the mRNA level. Immunofluorescence staining confirmed the protein expression of ACE2 and TMPRSS2 in retinal organoids and monolayer cultures. Furthermore, using the SARS-CoV-2 pseudovirus spike protein with GFP expression system, we found that retinal organoids and monolayer cultures can potentially be infected by the SARS-CoV-2 pseudovirus. Collectively, our findings highlighted the potential of iPSC-derived retinal organoids as the models for ACE2 receptor-based SARS-CoV-2 infection.

Although current studies suggested that conjunctivitis is not a common presentation of coronavirus disease 2019 (COVID-19), several studies have reported the presence of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) in ocular secretions. Coronavirus had not yet been successfully cultured from tears or conjunctival swabs in humans, neither SARS-CoV-2 nor SARS-CoV. However, live feline coronavirus has been isolated from conjunctival swabs. In addition, infection of COVID-19 through unprotected eye exposure had been suspected in several articles. Reports of ophthalmologists and otolaryngologists died of COVID-19 also raised concern on ocular transmission. As a result, we strongly suggest that personal protective equipment (PPE) should cover the mouth, nose, and eyes of ophthalmologists, especially when conjunctivitis caused by SARS-CoV-2 is clinically indistinguishable from other viral follicular conjunctivitis.

Hsin-Yu Yang,1 Yu-Fan Chang,1,2 Chih-Chien Hsu,1,2 Yu-Chieh Ko,1–3 Catherine Jui-Ling Liu,1,3 Mei-Ju Chen1,3 1Department of Ophthalmology, Taipei Veterans General Hospital, Taipei, Taiwan; 2Institute of Clinical Medicine, National Yang-Ming University, Taipei, Taiwan; 3School of Medicine, National Yang-Ming University, Taipei, Taiwan Purpose: We evaluated macular ganglion cell asymmetry as a tool for diagnosing paracentral scotoma (PCS) in early glaucoma.Patients and methods: This prospective study compared 58 patients with early glaucoma and PCS to 58 age-matched control individuals. All glaucomatous eyes had scotoma within the central 12 degrees of fixation and confined to one hemifield. We measured circumpapillary retinal nerve fiber layer (cpRNFL) thickness, macular ganglion cell-inner plexiform layer (GCIPL) thickness, and optic nerve head (ONH) parameters with Cirrus spectral domain optical coherence tomography. Macular ganglion cell asymmetry was expressed as the absolute differences and the ratios between the inferior and superior hemispheres, inferotemporal (IT) and superotemporal (ST) areas, IT and superonasal (SN) areas, IT and inferonasal (IN) areas, and ST and IN areas. The asymmetry index was the absolute log10 of the ratio. The area under the receiver operating characteristics curve (AUROC) and partial AUROC (pAUROC, specificities ≥90%) were analyzed for each parameter.Results: All GCIPL parameters and most cpRNFL parameters were significantly lower in early glaucoma than in controls. The best discriminating parameters were inferior RNFL thickness, average RNFL thickness, and minimum GCIPL thickness. The macular ganglion cell asymmetry parameters with the largest AUROCs were the log IT/ST index, the absolute difference in IT-ST GCIPL thicknesses, and the log IT/SN index. Performance was comparable between the log IT/ST index and the cpRNFL (inferior RNFL thickness, P=0.129), GCIPL (minimum GCIPL thickness, P=0.358), and ONH parameters (vertical cup-to-disc ratio, P=0.475).Conclusion: The IT/ST asymmetry index may be useful as a new parameter for detecting structural changes in glaucoma patients with PCS.Keywords: macular ganglion cell, paracentral scotoma, glaucoma, optical coherence tomography

Best disease (BD), also termed Best vitelliform macular dystrophy (BVMD), is a juvenile-onset form of macular degeneration and central visual loss. In this report, we generated an induced pluripotent stem cell (iPSC) line, TVGH-iPSC-012-04, from the peripheral blood mononuclear cells of a female patient with BD by using the Sendai virus delivery system. The resulting iPSCs retained the disease-causing DNA mutation, expressed pluripotent markers and could differentiate into three germ layers. We believe that BD patient-specific iPSCs provide a powerful in vitro model for evaluating the pathological phenotypes of the disease.

Artificial intelligence (AI) has been widely applied in the medical field and achieved enormous milestones in helping specialists to make diagnosis and remedy decisions, particularly in the field of eye diseases and ophthalmic screening. With the development of AI-based systems, the enormous hardware and software resources are required for optimal performance. In reality, there are many places on the planet where such resources are highly limited. Hence, the smartphone-based AI systems can be used to provide a remote control route to quickly screen eye diseases such as diabetic-related retinopathy or diabetic macular edema. However, the performance of such mobile-based AI systems is still uncharted territory. In this article, we discuss the issues of computing resource consumption and performance of the mobile device-based AI systems and highlight recent research on the feasibility and future potential of application of the mobile device-based AI systems in telemedicine.

Background To compare the diagnostic ability of Cirrus high-definition spectral-domain optical coherence tomography measurements of the macular ganglion cell-inner plexiform layer (GCIPL) vs the circumferential retinal nerve fiber layer (cpRNFL) to detect early glaucoma with hemifield visual field (VF) defects. Methods This prospective study included 96 patients with primary open-angle glaucoma (48 with superior hemifield defects and 48 with inferior hemifield defects) and 48 normal control subjects. All glaucomatous eyes had a mean deviation of the VF defect ≥-6.0 dB confined to one hemifield. cpRNFL and GCIPL thicknesses were recorded. Area under the receiver operating characteristic curve (AUROC) was calculated for each parameter and compared. Results All GCIPL parameters and most cpRNFL parameters (except at the nasal quadrant, and 2-, 3-, and 4-o'clock sectors) were significantly lower in glaucomatous eyes vs those in normal controls. In the superior hemifield defect group, the best discriminating parameters were 7-o'clock-sector cpRNFL thickness (AUROC value, 0.963), inferior cpRNFL thickness (0.926), and inferotemporal GCIPL thickness (0.923). Performance was comparable between the best measures of GCIPL analysis (inferotemporal GCIPL thickness) and those of cpRNFL (7-o'clock-sector thickness, p = 0.28). In the inferior hemifield defect group, the best discriminating parameters were 11- and 10-o'clock-sector cpRNFL thickness (0.940 and 0.904, respectively), and average cpRNFL thickness (0.909). Performance was comparable between the best measures from each method (superotemporal GCIPL thickness vs. 11-o'clock-sector cpRNFL thickness [0.857 vs 0.940, p = 0.07]). Conclusion Diagnostic abilities of GCIPL parameters and cpRNFL parameters for early glaucoma were comparable for eyes with either superior or inferior hemifield VF defects.

Human amniotic membrane (AM) offers unique advantages as a matrix to support the transplantation of limbal stem cells (LSCs) due to its inherent pro-regenerative and anti-inflammatory properties. However, the widespread use of AM in clinical treatments of ocular surface disorders is limited by its weak mechanical strength and fast degradation, and high cost associated with preserving freshly isolated AM. Here we constructed a composite membrane consisting of an electrospun bioabsorbable poly(e-caprolactone) (PCL) nanofiber mesh to significantly improve the ultimate tensile strength, toughness, and suture retention strength by 4-10-fold in comparison with decellularized AM sheet. The composite membrane showed extended stability and conferred longer-lasting coverage on wounded cornea surface compared with dAM. The composite membrane maintained the pro-regenerative and immunomodulatory properties of dAM, promoted LSC survival, retention, and organization, improved re-epithelialization of the defect area, and reduced inflammation and neovascularization. This study demonstrates the translational potential of our composite membrane for stem cell-based treatment of ocular surface damage. STATEMENT OF SIGNIFICANCE: Human decellularized amniotic membrane (dAM) has been widely shown as a biodegradable and bioactive matrix for regenerative tissue repair. However, the weak mechanical property has limited its widespread use in the clinic. Here we constructed a composite membrane using a layer of electrospun poly(e-caprolactone) (PCL) nanofiber mesh to reinforce the dAM sheet through covalent interfacial bonding, while retaining the unique bioactivity of dAM. In a rabbit model of limbal stem cell (LSC) deficiency induced by alkaline burn, we demonstrated the superior property of this PCL-dAM composite membrane for repairing damaged cornea through promoting LSC transplantation, improving re-epithelialization, and reducing inflammation and neovascularization. This new composite membrane offers great translational potential in supporting stem cell-based treatment of ocular surface damage.