Deletion of IFT20 exclusively in the RPE ablates primary cilia and leads to retinal degeneration.

Vision impairment places a serious burden on the aging society, affecting the lives of millions of people. Many retinal diseases are of genetic origin, of which over 50% are due to mutations in cilia-associated genes. Most research on retinal degeneration has focused on the ciliated photoreceptor cells of the retina. However, the contribution of primary cilia in other ocular cell types has largely been ignored. The retinal pigment epithelium (RPE) is a monolayer epithelium at the back of the eye intricately associated with photoreceptors and essential for visual function. It is already known that primary cilia in the RPE are critical for its development and maturation; however, it remains unclear whether this affects RPE function and retinal tissue homeostasis. We generated a conditional knockout mouse model, in which IFT20 is exclusively deleted in the RPE, ablating primary cilia. This leads to defective RPE function, followed by photoreceptor degeneration and, ultimately, vision impairment. Transcriptomic analysis offers insights into mechanisms underlying pathogenic changes, which include transcripts related to epithelial homeostasis, the visual cycle, and phagocytosis. Due to the loss of cilia exclusively in the RPE, this mouse model enables us to tease out the functional role of RPE cilia and their contribution to retinal degeneration, providing a powerful tool for basic and translational research in syndromic and non-syndromic retinal degeneration. Non-ciliary mechanisms of IFT20 in the RPE may also contribute to pathogenesis and cannot be excluded, especially considering the increasing evidence of non-ciliary functions of ciliary proteins.

Single-cell-resolution map of human retinal pigment epithelium helps discover subpopulations with differential disease sensitivity.

Regional phenotypic and functional differences in the retinal pigment epithelium (RPE) monolayer have been suggested to account for regional susceptibility in ocular diseases such as age-related macular degeneration (AMD), late-onset retinal degeneration (L-ORD), and choroideremia (CHM). However, a comprehensive description of human topographical RPE diversity is not yet available, thus limiting the understanding of regional RPE diversity and degenerative disease sensitivity in the eye. To develop a complete morphometric RPE map of the human eye, artificial intelligence­based software was trained to recognize, segment, and analyze RPE borders. Five statistically different, concentric RPE subpopulations (P1 to P5) were identified using cell area as a parameter, including a subpopulation (P4) with cell area comparable to that of macular cells in the far periphery of the eye. This work provides a complete reference map of human RPE subpopulations and their location in the eye. In addition, the analysis of cadaver non-AMD and AMD eyes and ultra-widefield fundus images of patients revealed differential vulnerability of the five RPE subpopulations to different retinal diseases.

Unbiased image segmentation assessment toolkit for quantitative differentiation of state-of-the-art algorithms and pipelines.

BACKGROUND: Image segmentation pipelines are commonly used in microscopy to identify cellular compartments like nucleus and cytoplasm, but there are few standards for comparing segmentation accuracy across pipelines. The process of selecting a segmentation assessment pipeline can seem daunting to researchers due to the number and variety of metrics available for evaluating segmentation quality. RESULTS: Here we present automated pipelines to obtain a comprehensive set of 69 metrics to evaluate segmented data and propose a selection methodology for models based on quantitative analysis, dimension reduction or unsupervised classification techniques and informed selection criteria. CONCLUSION: We show that the metrics used here can often be reduced to a small number of metrics that give a more complete understanding of segmentation accuracy, with different groups of metrics providing sensitivity to different types of segmentation error. These tools are delivered as easy to use python libraries, command line tools, Common Workflow Language Tools, and as Web Image Processing Pipeline interactive plugins to ensure a wide range of users can access and use them. We also present how our evaluation methods can be used to observe the changes in segmentations across modern machine learning/deep learning workflows and use cases.

Preexisting Autoimmunity Is Associated With Increased Severity of Coronavirus Disease 2019: A Retrospective Cohort Study Using Data From the National COVID Cohort Collaborative (N3C).

BACKGROUND: Identifying individuals with a higher risk of developing severe coronavirus disease 2019 (COVID-19) outcomes will inform targeted and more intensive clinical monitoring and management. To date, there is mixed evidence regarding the impact of preexisting autoimmune disease (AID) diagnosis and/or immunosuppressant (IS) exposure on developing severe COVID-19 outcomes. METHODS: A retrospective cohort of adults diagnosed with COVID-19 was created in the National COVID Cohort Collaborative enclave. Two outcomes, life-threatening disease and hospitalization, were evaluated by using logistic regression models with and without adjustment for demographics and comorbidities. RESULTS: Of the 2 453 799 adults diagnosed with COVID-19, 191 520 (7.81%) had a preexisting AID diagnosis and 278 095 (11.33%) had a preexisting IS exposure. Logistic regression models adjusted for demographics and comorbidities demonstrated that individuals with a preexisting AID (odds ratio [OR], 1.13; 95% confidence interval [CI]: 1.09-1.17; P < .001), IS exposure (OR, 1.27; 95% CI: 1.24-1.30; P < .001), or both (OR, 1.35; 95% CI: 1.29-1.40; P < .001) were more likely to have a life-threatening disease. These results were consistent when hospitalization was evaluated. A sensitivity analysis evaluating specific IS revealed that tumor necrosis factor inhibitors were protective against life-threatening disease (OR, 0.80; 95% CI: .66-.96; P = .017) and hospitalization (OR, 0.80; 95% CI: .73-.89; P < .001). CONCLUSIONS: Patients with preexisting AID, IS exposure, or both are more likely to have a life-threatening disease or hospitalization. These patients may thus require tailored monitoring and preventative measures to minimize negative consequences of COVID-19.

Data quality considerations for evaluating COVID-19 treatments using real world data: learnings from the National COVID Cohort Collaborative (N3C).

BACKGROUND: Multi-institution electronic health records (EHR) are a rich source of real world data (RWD) for generating real world evidence (RWE) regarding the utilization, benefits and harms of medical interventions. They provide access to clinical data from large pooled patient populations in addition to laboratory measurements unavailable in insurance claims-based data. However, secondary use of these data for research requires specialized knowledge and careful evaluation of data quality and completeness. We discuss data quality assessments undertaken during the conduct of prep-to-research, focusing on the investigation of treatment safety and effectiveness. METHODS: Using the National COVID Cohort Collaborative (N3C) enclave, we defined a patient population using criteria typical in non-interventional inpatient drug effectiveness studies. We present the challenges encountered when constructing this dataset, beginning with an examination of data quality across data partners. We then discuss the methods and best practices used to operationalize several important study elements: exposure to treatment, baseline health comorbidities, and key outcomes of interest. RESULTS: We share our experiences and lessons learned when working with heterogeneous EHR data from over 65 healthcare institutions and 4 common data models. We discuss six key areas of data variability and quality. (1) The specific EHR data elements captured from a site can vary depending on source data model and practice. (2) Data missingness remains a significant issue. (3) Drug exposures can be recorded at different levels and may not contain route of administration or dosage information. (4) Reconstruction of continuous drug exposure intervals may not always be possible. (5) EHR discontinuity is a major concern for capturing history of prior treatment and comorbidities. Lastly, (6) access to EHR data alone limits the potential outcomes which can be used in studies. CONCLUSIONS: The creation of large scale centralized multi-site EHR databases such as N3C enables a wide range of research aimed at better understanding treatments and health impacts of many conditions including COVID-19. As with all observational research, it is important that research teams engage with appropriate domain experts to understand the data in order to define research questions that are both clinically important and feasible to address using these real world data.

Modeling, validation and verification of three-dimensional cell-scaffold contacts from terabyte-sized images.

BACKGROUND: Cell-scaffold contact measurements are derived from pairs of co-registered volumetric fluorescent confocal laser scanning microscopy (CLSM) images (z-stacks) of stained cells and three types of scaffolds (i.e., spun coat, large microfiber, and medium microfiber). Our analysis of the acquired terabyte-sized collection is motivated by the need to understand the nature of the shape dimensionality (1D vs 2D vs 3D) of cell-scaffold interactions relevant to tissue engineers that grow cells on biomaterial scaffolds. RESULTS: We designed five statistical and three geometrical contact models, and then down-selected them to one from each category using a validation approach based on physically orthogonal measurements to CLSM. The two selected models were applied to 414 z-stacks with three scaffold types and all contact results were visually verified. A planar geometrical model for the spun coat scaffold type was validated from atomic force microscopy images by computing surface roughness of 52.35 nm ±31.76 nm which was 2 to 8 times smaller than the CLSM resolution. A cylindrical model for fiber scaffolds was validated from multi-view 2D scanning electron microscopy (SEM) images. The fiber scaffold segmentation error was assessed by comparing fiber diameters from SEM and CLSM to be between 0.46% to 3.8% of the SEM reference values. For contact verification, we constructed a web-based visual verification system with 414 pairs of images with cells and their segmentation results, and with 4968 movies with animated cell, scaffold, and contact overlays. Based on visual verification by three experts, we report the accuracy of cell segmentation to be 96.4% with 94.3% precision, and the accuracy of cell-scaffold contact for a statistical model to be 62.6% with 76.7% precision and for a geometrical model to be 93.5% with 87.6% precision. CONCLUSIONS: The novelty of our approach lies in (1) representing cell-scaffold contact sites with statistical intensity and geometrical shape models, (2) designing a methodology for validating 3D geometrical contact models and (3) devising a mechanism for visual verification of hundreds of 3D measurements. The raw and processed data are publicly available from https://isg.nist.gov/deepzoomweb/data/ together with the web -based verification system.

Biomaterial Strategies for Immunomodulation.

Strategies to enhance, suppress, or qualitatively shape the immune response are of importance for diverse biomedical applications, such as the development of new vaccines, treatments for autoimmune diseases and allergies, strategies for regenerative medicine, and immunotherapies for cancer. However, the intricate cellular and molecular signals regulating the immune system are major hurdles to predictably manipulating the immune response and developing safe and effective therapies. To meet this challenge, biomaterials are being developed that control how, where, and when immune cells are stimulated in vivo, and that can finely control their differentiation in vitro. We review recent advances in the field of biomaterials for immunomodulation, focusing particularly on designing biomaterials to provide controlled immunostimulation, targeting drugs and vaccines to lymphoid organs, and serving as scaffolds to organize immune cells and emulate lymphoid tissues. These ongoing efforts highlight the many ways in which biomaterials can be brought to bear to engineer the immune system.

Assessing the effect of selective serotonin reuptake inhibitors in the prevention of post-acute sequelae of COVID-19.

Background: Post-acute sequelae of COVID-19 (PASC) produce significant morbidity, prompting evaluation of interventions that might lower risk. Selective serotonin reuptake inhibitors (SSRIs) potentially could modulate risk of PASC via their central, hypothesized immunomodulatory, and/or antiplatelet properties although clinical trial data are lacking. Materials and Methods: This retrospective study was conducted leveraging real-world clinical data within the National COVID Cohort Collaborative (N3C) to evaluate whether SSRIs with agonist activity at the sigma-1 receptor (S1R) lower the risk of PASC, since agonism at this receptor may serve as a mechanism by which SSRIs attenuate an inflammatory response. Additionally, determine whether the potential benefit could be traced to S1R agonism. Presumed PASC was defined based on a computable PASC phenotype trained on the U09.9 ICD-10 diagnosis code. Results: Of the 17,908 patients identified, 1521 were exposed at baseline to a S1R agonist SSRI, 1803 to a non-S1R agonist SSRI, and 14,584 to neither. Using inverse probability weighting and Poisson regression, relative risk (RR) of PASC was assessed.A 29% reduction in the RR of PASC (0.704 [95% CI, 0.58-0.85]; P = 4 ×10-4) was seen among patients who received an S1R agonist SSRI compared to SSRI unexposed patients and a 21% reduction in the RR of PASC was seen among those receiving an SSRI without S1R agonist activity (0.79 [95% CI, 0.67 - 0.93]; P = 0.005).Thus, SSRIs with and without reported agonist activity at the S1R were associated with a significant decrease in the risk of PASC.

Paxlovid (nirmatrelvir/ritonavir) effectiveness against hospitalization and death in N3C: A target trial emulation study.

Importance: COVID-19 has placed a monumental burden on the health care system globally. Although no longer a public health emergency, there is still a pressing need for effective treatments to prevent hospitalization and death. Paxlovid (nirmatrelvir/ritonavir) is a promising and potentially effective antiviral that has received emergency use authorization by the U.S. FDA. Objective: Determine real world effectiveness of Paxlovid nationwide and investigate disparities between treated and untreated eligible patients. Design/Setting/Participants: Population-based cohort study emulating a target trial, using inverse probability weighted models to balance treated and untreated groups on baseline confounders. Participants were patients with a SARS-CoV-2 positive test or diagnosis (index) date between December 2021 and February 2023 selected from the National COVID Cohort Collaborative (N3C) database who were eligible for Paxlovid treatment. Namely, adults with at least one risk factor for severe COVID-19 illness, no contraindicated medical conditions, not using one or more strictly contraindicated medications, and not hospitalized within three days of index. From this cohort we identified patients who were treated with Paxlovid within 5 days of positive test or diagnosis (n = 98,060) and patients who either did not receive Paxlovid or were treated outside the 5-day window (n = 913,079 never treated; n = 1,771 treated after 5 days). Exposures: Treatment with Paxlovid within 5 days of positive COVID-19 test or diagnosis. Main Outcomes and Measures: Hospitalization and death in the 28 days following COVID-19 index date. Results: A total of 1,012,910 COVID-19 positive patients at risk for severe COVID-19 were included, 9.7% of whom were treated with Paxlovid. Uptake varied widely by geographic region and timing, with top adoption areas near 50% and bottom near 0%. Adoption increased rapidly after EUA, reaching steady state by 6/2022. Participants who were treated with Paxlovid had a 26% (RR, 0.742; 95% CI, 0.689-0.812) reduction in hospitalization risk and 73% (RR, 0.269, 95% CI, 0.179-0.370) reduction in mortality risk in the 28 days following COVID-19 index date. Conclusions/Relevance: Paxlovid is effective in preventing hospitalization and death in at-risk COVID-19 patients. These results were robust to a large number of sensitivity considerations. Disclosure: The authors report no disclosures. Key points: Question: Is treatment with Paxlovid (nirmatrelvir/ritonavir) associated with a reduction in 28-day hospitalization and mortality in patients at risk for severe COVID-19? Findings: In this multi-institute retrospective cohort study of 1,012,910 patients, Paxlovid treatment within 5 days after COVID-19 diagnosis reduced 28-day hospitalization and mortality by 26% and 73% respectively, compared to no treatment with Paxlovid within 5 days. Paxlovid uptake was low overall (9.7%) and highly variable. Meaning: In Paxlovid-eligible patients, treatment was associated with decreased risk of hospitalization and death. Results align with prior randomized trials and observational studies, thus supporting the real-world effectiveness of Paxlovid.

Assessing the Effect of Selective Serotonin Reuptake Inhibitors in the Prevention of Post-Acute Sequelae of COVID-19.

Importance: Post-acute sequelae of COVID-19 (PASC) produce significant morbidity, prompting evaluation of interventions that might lower risk. Selective serotonin reuptake inhibitors (SSRIs) potentially could modulate risk of PASC via their central, hypothesized immunomodulatory, and/or antiplatelet properties and therefore may be postulated to be of benefit in patients with PASC, although clinical trial data are lacking. Objectives: The main objective was to evaluate whether SSRIs with agonist activity at the sigma-1 receptor lower the risk of PASC, since agonism at this receptor may serve as a mechanism by which SSRIs attenuate an inflammatory response. A secondary objective was to determine whether potential benefit could be traced to sigma-1 agonism by evaluating the risk of PASC among recipients of SSRIs that are not S1R agonists. Design: Retrospective study leveraging real-world clinical data within the National COVID Cohort Collaborative (N3C), a large centralized multi-institutional de-identified EHR database. Presumed PASC was defined based on a computable PASC phenotype trained on the U09.9 ICD-10 diagnosis code to more comprehensively identify patients likely to have the condition, since the ICD code has come into wide-spread use only recently. Setting: Population-based study at US medical centers. Participants: Adults (≥ 18 years of age) with a confirmed COVID-19 diagnosis date between October 1, 2021 and April 7, 2022 and at least one follow up visit 45 days post-diagnosis. Of the 17 933 patients identified, 2021 were exposed at baseline to a S1R agonist SSRI, 1328 to a non-S1R agonist SSRI, and 14 584 to neither. Exposures: Exposure at baseline (at or prior to COVID-19 diagnosis) to an SSRI with documented or presumed agonist activity at the S1R (fluvoxamine, fluoxetine, escitalopram, or citalopram), an SSRI without agonist activity at S1R (sertraline, an antagonist, or paroxetine, which does not appreciably bind to the S1R), or none of these agents. Main Outcome and Measurement: Development of PASC based on a previously validated XGBoost-trained algorithm. Using inverse probability weighting and Poisson regression, relative risk (RR) of PASC was assessed. Results: A 26% reduction in the RR of PASC (0.74 [95% CI, 0.63-0.88]; P = 5 × 10-4) was seen among patients who received an S1R agonist SSRI compared to SSRI unexposed patients and a 25% reduction in the RR of PASC was seen among those receiving an SSRI without S1R agonist activity (0.75 [95% CI, 0.62 - 0.90]; P = 0.003) compared to SSRI unexposed patients. Conclusions and Relevance: SSRIs with and without reported agonist activity at the S1R were associated with a significant decrease in the risk of PASC. Future prospective studies are warranted.

Pre-existing autoimmunity is associated with increased severity of COVID-19: A retrospective cohort study using data from the National COVID Cohort Collaborative (N3C).

Importance: Identifying individuals with a higher risk of developing severe COVID-19 outcomes will inform targeted or more intensive clinical monitoring and management. Objective: To examine, using data from the National COVID Cohort Collaborative (N3C), whether patients with pre-existing autoimmune disease (AID) diagnosis and/or immunosuppressant (IS) exposure are at a higher risk of developing severe COVID-19 outcomes. Design setting and participants: A retrospective cohort of 2,453,799 individuals diagnosed with COVID-19 between January 1 st , 2020, and June 30 th , 2022, was created from the N3C data enclave, which comprises data of 15,231,849 patients from 75 USA data partners. Patients were stratified as those with/without a pre-existing diagnosis of AID and/or those with/without exposure to IS prior to COVID-19. Main outcomes and measures: Two outcomes of COVID-19 severity, derived from the World Health Organization severity score, were defined, namely life-threatening disease and hospitalization. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated using logistic regression models with and without adjustment for demographics (age, BMI, gender, race, ethnicity, smoking status), and comorbidities (cardiovascular disease, dementia, pulmonary disease, liver disease, type 2 diabetes mellitus, kidney disease, cancer, and HIV infection). Results: In total, 2,453,799 (16.11% of the N3C cohort) adults (age> 18 years) were diagnosed with COVID-19, of which 191,520 (7.81%) had a prior AID diagnosis, and 278,095 (11.33%) had a prior IS exposure. Logistic regression models adjusted for demographic factors and comorbidities demonstrated that individuals with a prior AID (OR = 1.13, 95% CI 1.09 - 1.17; p =2.43E-13), prior exposure to IS (OR= 1.27, 95% CI 1.24 - 1.30; p =3.66E-74), or both (OR= 1.35, 95% CI 1.29 - 1.40; p =7.50E-49) were more likely to have a life-threatening COVID-19 disease. These results were confirmed after adjusting for exposure to antivirals and vaccination in a cohort subset with COVID-19 diagnosis dates after December 2021 (AID OR = 1.18, 95% CI 1.02 - 1.36; p =2.46E-02; IS OR= 1.60, 95% CI 1.41 - 1.80; p =5.11E-14; AID+IS OR= 1.93, 95% CI 1.62 - 2.30; p =1.68E-13). These results were consistent when evaluating hospitalization as the outcome and also when stratifying by race and sex. Finally, a sensitivity analysis evaluating specific IS revealed that TNF inhibitors were protective against life-threatening disease (OR = 0.80, 95% CI 0.66-0.96; p =1.66E-2) and hospitalization (OR = 0.80, 95% CI 0.73 - 0.89; p =1.06E-05). Conclusions and Relevance: Patients with pre-existing AID, exposure to IS, or both are more likely to have a life-threatening disease or hospitalization. These patients may thus require tailored monitoring and preventative measures to minimize negative consequences of COVID-19.

Measuring dimensionality of cell-scaffold contacts of primary human bone marrow stromal cells cultured on electrospun fiber scaffolds.

The properties and structure of the cellular microenvironment can influence cell behavior. Sites of cell adhesion to the extracellular matrix (ECM) initiate intracellular signaling that directs cell functions such as proliferation, differentiation, and apoptosis. Electrospun fibers mimic the fibrous nature of native ECM proteins and cell culture in fibers affects cell shape and dimensionality, which can drive specific functions, such as the osteogenic differentiation of primary human bone marrow stromal cells (hBMSCs), by. In order to probe how scaffolds affect cell shape and behavior, cell-fiber contacts were imaged to assess their shape and dimensionality through a novel approach. Fluorescent polymeric fiber scaffolds were made so that they could be imaged by confocal fluorescence microscopy. Fluorescent polymer films were made as a planar control. hBSMCs were cultured on the fluorescent substrates and the cells and substrates were imaged. Two different image analysis approaches, one having geometrical assumptions and the other having statistical assumptions, were used to analyze the 3D structure of cell-scaffold contacts. The cells cultured in scaffolds contacted the fibers in multiple planes over the surface of the cell, while the cells cultured on films had contacts confined to the bottom surface of the cell. Shape metric analysis indicated that cell-fiber contacts had greater dimensionality and greater 3D character than the cell-film contacts. These results suggest that cell adhesion site-initiated signaling could emanate from multiple planes over the cell surface during culture in fibers, as opposed to emanating only from the cell's basal surface during culture on planar surfaces.

In vitro disease modeling of oculocutaneous albinism type 1 and 2 using human induced pluripotent stem cell-derived retinal pigment epithelium.

Oculocutaneous albinism (OCA) encompasses a set of autosomal recessive genetic conditions that affect pigmentation in the eye, skin, and hair. OCA patients display reduced best-corrected visual acuity, reduced to absent ocular pigmentation, abnormalities in fovea development, and/or abnormal decussation of optic nerve fibers. It has been hypothesized that improving eye pigmentation could prevent or rescue some of the vision defects. The goal of the present study was to develop an in vitro model for studying pigmentation defects in human retinal pigment epithelium (RPE). We developed a "disease in a dish" model for OCA1A and OCA2 types using induced pluripotent stem cells to generate RPE. The RPE is a monolayer of cells that are pigmented, polarized, and polygonal in shape, located between the neural retina and choroid, with an important role in vision. Here we show that RPE tissue derived in vitro from OCA patients recapitulates the pigmentation defects seen in albinism, while retaining the apical-basal polarity and normal polygonal morphology of the constituent RPE cells.

Deep-learning based multi-modal retinal image registration for the longitudinal analysis of patients with age-related macular degeneration.

This work reports a deep-learning based registration algorithm that aligns multi-modal retinal images collected from longitudinal clinical studies to achieve accuracy and robustness required for analysis of structural changes in large-scale clinical data. Deep-learning networks that mirror the architecture of conventional feature-point-based registration were evaluated with different networks that solved for registration affine parameters, image patch displacements, and patch displacements within the region of overlap. The ground truth images for deep learning-based approaches were derived from successful conventional feature-based registration. Cross-sectional and longitudinal affine registrations were performed across color fundus photography (CFP), fundus autofluorescence (FAF), and infrared reflectance (IR) image modalities. For mono-modality longitudinal registration, the conventional feature-based registration method achieved mean errors in the range of 39-53 µm (depending on the modality) whereas the deep learning method with region overlap prediction exhibited mean errors in the range 54-59 µm. For cross-sectional multi-modality registration, the conventional method exhibited gross failures with large errors in more than 50% of the cases while the proposed deep-learning method achieved robust performance with no gross failures and mean errors in the range 66-69 µm. Thus, the deep learning-based method achieved superior overall performance across all modalities. The accuracy and robustness reported in this work provide important advances that will facilitate clinical research and enable a detailed study of the progression of retinal diseases such as age-related macular degeneration.

Deep Learning-Based Automatic Detection of Ellipsoid Zone Loss in Spectral-Domain OCT for Hydroxychloroquine Retinal Toxicity Screening.

Purpose: Retinal toxicity resulting from hydroxychloroquine use manifests photoreceptor loss and disruption of the ellipsoid zone (EZ) reflectivity band detectable on spectral-domain (SD) OCT imaging. This study investigated whether an automatic deep learning-based algorithm can detect and quantitate EZ loss on SD OCT images with an accuracy comparable with that of human annotations. Design: Retrospective analysis of data acquired in a prospective, single-center, case-control study. Participants: Eighty-five patients (168 eyes) who were long-term hydroxychloroquine users (average exposure time, 14 ± 7.2 years). Methods: A mask region-based convolutional neural network (M-RCNN) was implemented and trained on individual OCT B-scans. Scan-by-scan detections were aggregated to produce an en face map of EZ loss per 3-dimensional SD OCT volume image. To improve the accuracy and robustness of the EZ loss map, a dual network architecture was proposed that learns to detect EZ loss in parallel using horizontal (horizontal mask region-based convolutional neural network [M-RCNNH]) and vertical (vertical mask region-based convolutional neural network [M-RCNNV]) B-scans independently. To quantify accuracy, 10-fold cross-validation was performed. Main Outcome Measures: Precision, recall, intersection over union (IOU), F1-score metrics, and measured total EZ loss area were compared against human grader annotations and with the determination of toxicity based on the recommended screening guidelines. Results: The combined projection network demonstrated the best overall performance: precision, 0.90 ± 0.09; recall, 0.88 ± 0.08; and F1 score, 0.89 ± 0.07. The combined model performed superiorly to the M-RCNNH only model (precision, 0.79 ± 0.17; recall, 0.96 ± 0.04; IOU, 0.78 ± 0.15; and F1 score, 0.86 ± 0.12) and M-RCNNV only model (precision, 0.71 ± 0.21; recall, 0.94 ± 0.06; IOU, 0.69 ± 0.21; and F1 score, 0.79 ± 0.16). The accuracy was comparable with the variability of human experts: precision, 0.85 ± 0.09; recall, 0.98 ± 0.01; IOU, 0.82 ± 0.12; and F1 score, 0.91 ± 0.06. Automatically generated en face EZ loss maps provide quantitative SD OCT metrics for accurate toxicity determination combined with other functional testing. Conclusions: The algorithm can provide a fast, objective, automatic method for measuring areas with EZ loss and can serve as a quantitative assistance tool to screen patients for the presence and extent of toxicity.

Loss of Ciliary Gene Bbs8 Results in Physiological Defects in the Retinal Pigment Epithelium.

Primary cilia are sensory organelles vital for developmental and physiological processes. Their dysfunction causes a range of phenotypes including retinopathies. Although primary cilia have been described in the retinal pigment epithelium (RPE), little is known about their contribution to biological processes within this tissue. Ciliary proteins are increasingly being identified in non-ciliary locations and might carry out additional functions, disruption of which possibly contributes to pathology. The RPE is essential for maintaining photoreceptor cells and visual function. We demonstrate that upon loss of Bbs8, predominantly thought to be a ciliary gene, the RPE shows changes in gene and protein expression initially involved in signaling pathways and developmental processes, and at a later time point RPE homeostasis and function. Differentially regulated molecules affecting the cytoskeleton and cellular adhesion, led to defective cellular polarization and morphology associated with a possible epithelial-to-mesenchymal transition (EMT)-like phenotype. Our data highlights the benefit of combinatorial "omics" approaches with in vivo data for investigating the function of ciliopathy proteins. It also emphasizes the importance of ciliary proteins in the RPE and their contribution to visual disorders, which must be considered when designing treatment strategies for retinal degeneration.

Comparison of Artificial Intelligence based approaches to cell function prediction.

Predicting Retinal Pigment Epithelium (RPE) cell functions in stem cell implants using non-invasive bright field microscopy imaging is a critical task for clinical deployment of stem cell therapies. Such cell function predictions can be carried out using Artificial Intelligence (AI) based models. In this paper we used Traditional Machine Learning (TML) and Deep Learning (DL) based AI models for cell function prediction tasks. TML models depend on feature engineering and DL models perform feature engineering automatically but have higher modeling complexity. This work aims at exploring the tradeoffs between three approaches using TML and DL based models for RPE cell function prediction from microscopy images and at understanding the accuracy relationship between pixel-, cell feature-, and implant label-level accuracies of models. Among the three compared approaches to cell function prediction, the direct approach to cell function prediction from images is slightly more accurate in comparison to indirect approaches using intermediate segmentation and/or feature engineering steps. We also evaluated accuracy variations with respect to model selections (five TML models and two DL models) and model configurations (with and without transfer learning). Finally, we quantified the relationships between segmentation accuracy and the number of samples used for training a model, segmentation accuracy and cell feature error, and cell feature error and accuracy of implant labels. We concluded that for the RPE cell data set, there is a monotonic relationship between the number of training samples and image segmentation accuracy, and between segmentation accuracy and cell feature error, but there is no such a relationship between segmentation accuracy and accuracy of RPE implant labels.

CSF1R blockade induces macrophage ablation and results in mouse choroidal vascular atrophy and RPE disorganization.

The choroid, which provides vascular supply to the outer retina, demonstrates progressive degeneration in aging and age-related macular degeneration (AMD). However mechanisms that maintain or compromise choroidal homeostasis are obscure. We discovered that the ablation of choroidal macrophages via CSF1R blockade was associated with choroidal vascular atrophy and retinal pigment epithelial (RPE) changes including structural disruption, downregulation of visual cycle genes, and altered angiogenic factor expression. Suspending CSF1R blockade following ablation enabled spontaneous macrophage regeneration, which fully restored original macrophage distributions and morphologies. Macrophage regeneration was accompanied by arrested vascular degeneration and ameliorated pathological RPE alterations. These findings suggest that choroidal macrophages play a previously unappreciated trophic role in maintaining choroidal vasculature and RPE cells, implicating insufficiency in choroidal macrophage function as a factor in aging- and AMD-associated pathology. Modulating macrophage function may constitute a strategy for the therapeutic preservation of the choroid and RPE in age-related retinal disorders.

Deep learning predicts function of live retinal pigment epithelium from quantitative microscopy.

Increases in the number of cell therapies in the preclinical and clinical phases have prompted the need for reliable and noninvasive assays to validate transplant function in clinical biomanufacturing. We developed a robust characterization methodology composed of quantitative bright-field absorbance microscopy (QBAM) and deep neural networks (DNNs) to noninvasively predict tissue function and cellular donor identity. The methodology was validated using clinical-grade induced pluripotent stem cell-derived retinal pigment epithelial cells (iPSC-RPE). QBAM images of iPSC-RPE were used to train DNNs that predicted iPSC-RPE monolayer transepithelial resistance, predicted polarized vascular endothelial growth factor (VEGF) secretion, and matched iPSC-RPE monolayers to the stem cell donors. DNN predictions were supplemented with traditional machine-learning algorithms that identified shape and texture features of single cells that were used to predict tissue function and iPSC donor identity. These results demonstrate noninvasive cell therapy characterization can be achieved with QBAM and machine learning.

High-yield, automated intracellular electrophysiology in retinal pigment epithelia.

BACKGROUND: Recent advancements with induced pluripotent stem cell-derived (iPSC) retinal pigment epithelium (RPE) have made disease modeling and cell therapy for macular degeneration feasible. However, current techniques for intracellular electrophysiology - used to validate epithelial function - are painstaking and require manual skill; limiting experimental throughput. NEW METHOD: A five-stage algorithm, leveraging advances in automated patch clamping, systematically derived and optimized, improves yield and reduces skill when compared to conventional, manual techniques. RESULTS: The automated algorithm improves yield per attempt from 17% (manually, n = 23) to 22% (automated, n = 120) (chi-squared, p = 0.004). Specifically for RPE, depressing the local cell membrane by 6 µm and electroporating (buzzing) just prior to this depth (5 µm) maximized yield. COMPARISON WITH EXISTING METHOD: Conventionally, intracellular epithelial electrophysiology is performed by manually lowering a pipette with a micromanipulator, blindly, towards a monolayer of cells and spontaneously stopping when the magnitude of the instantaneous measured membrane potential decreased below a predetermined threshold. The new method automatically measures the pipette tip resistance during the descent, detects the cell surface, indents the cell membrane, and briefly buzzes to electroporate the membrane while descending, overall achieving a higher yield than conventional methods. CONCLUSIONS: This paper presents an algorithm for high-yield, automated intracellular electrophysiology in epithelia; optimized for human RPE. Automation reduces required user skill and training while, simultaneously, improving yield. This algorithm could enable large-scale exploration of drug toxicity and physiological function verification for numerous kinds of epithelia.