Changes in arterial myocyte excitability induced by subarachnoid hemorrhage in a rat model.

Aneurismal subarachnoid hemorrhage (aSAH) is a neurovascular disease produced by the rupture of the cerebral arteries and the extravasation of blood to the subarachnoid space and is accompanied by severe comorbidities. Secondarily associated vasospasm is one of the main side effects after hydrocephalus and possible rebleeding. Here, we analyze the alterations in function in the arteries of a rat model of SAH. For this, autologous blood was injected into the cisterna magna. We performed electrophysiological, microfluorimetric, and molecular biology experiments at different times after SAH to determine the functional and molecular changes induced by the hemorrhage. Our results confirmed that in SAH animals, arterial myocytes were depolarized on days 5 and 7, had higher [Ca2+]i on baseline, peaks and plateaus, and were more excitable at low levels of depolarization on day 7, than in the control and sham animals. Microarray analysis showed that, on day 7, the sets of genes related to voltage-dependent Ca2+ channels and K+ dynamics in SAH animals decreased, while the voltage-independent Ca2+ dynamics genes were over-represented. In conclusion, after SAH, several mechanisms involved in arterial reactivity were altered in our animal model, suggesting that there is no unique cause of vasospasm and alterations in several signaling pathways are involved in its development.

Changes in Adhesion and the Expression of Adhesion Molecules in PBMCs after Aneurysmal Subarachnoid Hemorrhage: Relation to Cerebral Vasospasm.

Aneurysmal subarachnoid hemorrhage (aSAH) is a neurovascular disease produced by extravasation of blood to the subarachnoid space after rupture of the cerebral vessels. After bleeding, the immune response is activated. The role of peripheral blood mononuclear cells (PBMCs) in this response is a current subject of research. We have analysed the changes in PBMCs of patients with aSAH and their interaction with the endothelium, focusing on their adhesion and the expression of adhesion molecules. Using an in vitro adhesion assay, we observed that the adhesion of PBMCs of patients with aSAH is increased. Flow cytometry analysis shows that monocytes increased significantly in patients, especially in those who developed vasospasm (VSP). In aSAH patients, the expression of CD162, CD49d, CD62L and CD11a in T lymphocytes and of CD62L in monocytes increased. However, the expression of CD162, CD43, and CD11a decreased in monocytes. Furthermore, monocytes from patients who developed arteriographic VSP had lower expression of CD62L. In conclusion, our results confirm that after aSAH, monocyte count and adhesion of PBMCs increase, especially in patients with VSP, and that the expression of several adhesion molecules is altered. These observations can help predict VSP and to improve the treatment of this pathology.

The Effect of Ti/Ta Ratio and Processing Routes on the Hardness and Elastic Modulus of Porous TiNbZrTa Alloys.

TiNbZrTa alloys are promising for multidisciplinary applications, such as refractory and biomedical purposes, due to their high thermal stability and non-toxicity. Hardness and elastic modulus are among the key features for their adequate industrial applications. The influence of porosity and Ti/Ta ratio were investigated on TiNbZrTa alloys produced by three different processing routes, i.e., (i) blend element and posterior press and sintering (BE + P&S); (ii) mechanical alloying with press and sintering (MA + P&S); and (iii) arc melting and casting. Porosity decreased in the following order: casting < MA + P&S < BE + P&S. The total porosity of alloys increased with increasing Ta contents, i.e., by lowering the Ti/Ta ratio. However, the Ti/Ta ratio did not considerably affect the bonding energy or the elastic modulus. Hardness was increased significantly in dense alloys compared to porous ones. However, porosity and Ti/Ta ratio did not show a clear trend in hardness among the porous alloys.

The direct effect of fibroblast growth factor 23 on vascular smooth muscle cell phenotype and function.

BACKGROUND: In chronic kidney disease (CKD) patients, increased levels of fibroblast growth factor 23 (FGF23) are associated with cardiovascular mortality. The relationship between FGF23 and heart hypertrophy has been documented, however, it is not known whether FGF23 has an effect on vasculature. Vascular smooth muscle cells VSMCs may exhibit different phenotypes; our hypothesis is that FGF23 favours a switch from a contractile to synthetic phenotype that may cause vascular dysfunction. Our objective was to determine whether FGF23 may directly control a change in VSMC phenotype. METHODS: This study includes in vitro, in vivo and ex vivo experiments and evaluation of patients with CKD stages 2-3 studying a relationship between FGF23 and vascular dysfunction. RESULTS: In vitro studies show that high levels of FGF23, by acting on its specific receptor FGFR1 and Erk1/2, causes a change in the phenotype of VSMCs from contractile to synthetic. This change is mediated by a downregulation of miR-221/222, which augments the expression of MAP3K2 and PAK1. miR-221/222 transfections recovered the contractile phenotype of VSMCs. Infusion of recombinant FGF23 to rats increased vascular wall thickness, with VSMCs showing a synthetic phenotype with a reduction of miR-221 expression. Ex-vivo studies on aortic rings demonstrate also that high FGF23 increases arterial stiffening. In CKD 2-3 patients, elevation of FGF23 was associated with increased pulse wave velocity and reduced plasma levels of miR-221/222. CONCLUSION: In VSMCs, high levels of FGF23, through the downregulation of miR-221/222, causes a change to a synthetic phenotype. This change in VSMCs increases arterial stiffening and impairs vascular function, which might ultimately worsen cardiovascular disease.

Pandemic response gaps: Infection prevention and control lessons learned during coronavirus disease 2019 (COVID-19) outbreaks in skilled nursing facilities in Detroit, Michigan.

BACKGROUND: Hospitalizations among skilled nursing facility (SNF) residents in Detroit increased in mid-March 2020 due to the coronavirus disease 2019 (COVID-19) pandemic. Outbreak response teams were deployed from local healthcare systems, the Centers for Disease Control and Prevention (CDC), and the Detroit Health Department (DHD) to understand the infection prevention and control (IPC) gaps in SNFs that may have accelerated the outbreak. METHODS: We conducted 2 point-prevalence surveys (PPS-1 and PPS-2) at 13 Detroit SNFs from April 8 to May 8, 2020. The DHD and partners conducted facility-wide severe acute respiratory coronavirus virus 2 (SARS-CoV-2) testing of all residents and staff and collected information regarding resident cohorting, staff cohorting, and personnel protective equipment (PPE) utilized during that time. RESULTS: Resident cohorting had been implemented in 7 of 13 (58.3%) SNFs prior to point-prevalence survey 1 (PPS-1), and other facilities initiated cohorting after obtaining PPS-1 results. Cohorting protocols of healthcare practitioners and environmental service staff were not established in 4 (31%) of 13 facilities, and in 3 facilities (23.1%) the ancillary staff were not assigned to cohorts. Also, 2 SNFs (15%) had an observation unit prior to PPS-1, 2 (15%) had an observation unit after PPS-1, 4 (31%) could not establish an observation unit due to inadequate space, and 5 (38.4%) created an observation unit after PPS-2. CONCLUSION: On-site consultations identified gaps in IPC knowledge and cohorting that may have contributed to ongoing transmission of SARS-CoV-2 among SNF residents despite aggressive testing measures. Infection preventionists (IPs) are critical in guiding ongoing IPC practices in SNFs to reduce spread of COVID-19 through response and prevention.

A Deep Learning Framework for the Detection and Quantification of Reticular Pseudodrusen and Drusen on Optical Coherence Tomography.

Purpose: The purpose of this study was to develop and validate a deep learning (DL) framework for the detection and quantification of reticular pseudodrusen (RPD) and drusen on optical coherence tomography (OCT) scans. Methods: A DL framework was developed consisting of a classification model and an out-of-distribution (OOD) detection model for the identification of ungradable scans; a classification model to identify scans with drusen or RPD; and an image segmentation model to independently segment lesions as RPD or drusen. Data were obtained from 1284 participants in the UK Biobank (UKBB) with a self-reported diagnosis of age-related macular degeneration (AMD) and 250 UKBB controls. Drusen and RPD were manually delineated by five retina specialists. The main outcome measures were sensitivity, specificity, area under the receiver operating characteristic (ROC) curve (AUC), kappa, accuracy, intraclass correlation coefficient (ICC), and free-response receiver operating characteristic (FROC) curves. Results: The classification models performed strongly at their respective tasks (0.95, 0.93, and 0.99 AUC, respectively, for the ungradable scans classifier, the OOD model, and the drusen and RPD classification models). The mean ICC for the drusen and RPD area versus graders was 0.74 and 0.61, respectively, compared with 0.69 and 0.68 for intergrader agreement. FROC curves showed that the model's sensitivity was close to human performance. Conclusions: The models achieved high classification and segmentation performance, similar to human performance. Translational Relevance: Application of this robust framework will further our understanding of RPD as a separate entity from drusen in both research and clinical settings.

Outcomes of peroral endoscopic myotomy in patients with spastic esophageal motility disorders that do not fulfill Chicago Classification criteria

Introduction: per-oral endoscopic myotomy (POEM) hasbecome a mainstream treatment for achalasia and is apromising therapy in spastic disorders.Methods: this is a retrospective study of prospectively collected data (case series). We present the first results of theuse of POEM in patients with atypical spastic esophagealmotor disorders that do not satisfy current Chicago Classification criteria. Seven consecutive patients with troublesome and persistent symptoms (12-180 months) relatedto atypical spastic esophageal motor dysfunction weresystematically assessed before and after POEM, the extentof which was tailored by manometric findings. In five ofthe patients, other endoscopic or surgical procedures hadfailed.Results: high-resolution manometry (HRM) showed a spasticesophageal body contractile segment in varying positionsand lengths along the esophageal body which did not meet Chicago Classification criteria. After POEM, dysphagia and/or chest pain had either resolved or was greatly reduced.HRM 3-6 months after myotomy showed that the regions ofspastic contraction targeted by myotomy had been ablated.There were no major complications. The clinical responseswere fully maintained up to the most recent assessmentsafter POEM (range 7-44 months).Conclussion: in our seven patients, POEM was a highly effective treatment for patients with troublesome symptoms related to atypical spastic esophageal motility disorders. (AU)

Evaluación y caracterización de la sirtuina 2.3 de Giardia duodenalis (CdSir2.3)

La giardiasis es la enfermedad gastrointestinal de mayor incidencia mundial, causada por el protozoario Giardia duodenalis, para la cual no se cuenta con una vacuna o tratamiento eficiente. En aras de buscar nuevos blancos farmacológicos contra este parásito, se han estudiado las enzimas del metabolismo energético, como las sirtuinas, deacetilasas dependientes del dinucleótido de adenina y nicotinamida (NAD). Previamente se identificó a GdSir2.1 y GdSir2.2 como deacetilasas dependientes de NAD, con localizaciones subcelulares diferentes. En este trabajo se estudió otro candidato a sirtuina (GdSir2.3) mediante herramientas bioinformáticas para la identificación de características típicas de la familia sirtuina en la secuencia del candidato, y experimentales como la obtención de la proteína recombinante 6xHis-GdSir2.3 que demostró actividad deacetilasa dependiente de NAD y que sirvió como antígeno en la producción de los IgY - α -6xHis-GdSir2.3 para la localización subcelular de la proteína endógena en G. duodenalis. Lo anterior concuerda con otros estudios donde se señala a GdSir2.3 como un importante regulador de la enquistación, debido a su aumento de expresión durante esta etapa del ciclo de vida, constituyéndola como un blanco farmacológico promisorio para el control de esta parasitemia.

Giardiasis is the gastrointestinal disease with the highest incidence worldwide, caused by the protozoan Giardia duodenalis, for which there is no vaccine or efficient treatment. In order to find new pharmacological targets against this parasite, energy metabolism enzymes such as sirtuins, deacetylases dependent on the nicotinamide adenine dinucleotide (NAD), have been studied. GdSir2.1 and GdSir2.2 were previously identified as NAD-dependent deacetylases, with different subcellular locations. In this work, another candidate for sirtuin (GdSir2.3) was studied using bioinformatic tools for the identification of typical characteristics of the sirtuin family in the sequence of the candidate; and experimental ones such as obtaining the recombinant protein 6xHis-GdSir2.3 that demonstrated NAD-dependent deacetylase activity; and that it served as an antigen in the production of IgY - α - 6xHis-GdSir2.3 for the subcellular localization of the endogenous protein in G. duodenalis. The foregoing is consistent with other studies where GdSir2.3 is indicated as an important regulator of encyst due to its increased expression during this stage of the life cycle, constituting it as a promising drug target for the control of this parasitaemia.

A giardíase é a doença gastrointestinal de maior incidência no mundo, causada pelo protozoário Giardia duodenalis, para a qual não existe vacina ou tratamento eficaz. Com o objetivo de encontrar novos alvos farmacológicos contra esse parasita, têm sido estudadas enzimas do metabolismo energético, como as sirtuínas, desacetilases dependentes do dinucleotídeo adenina nicotinamida (NAD). GdSir2.1 e GdSir2.2 foram previamente identificados como desacetilases dependentes de NAD, com diferentes localizações subcelulares. Neste trabalho, outro candidato a sirtuin (GdSir2.3) foi estudado usando ferramentas de bioinformática para a identificação de características típicas da família sirtuin na sequência do candidato; e experimentais, como a obtenção da proteína recombinante 6xHis-GdSir2.3 que demonstrou atividade desacetilase dependente de NAD; e que serviu como antígeno na produção de IgY - α - 6xHis-GdSir2.3 para a localização subcelular da proteína endógena em G. duodenalis. O exposto é consistente com outros estudos em que o GdSir2.3 é apontado como um importante regulador de encisto devido à sua expressão aumentada durante esta fase do ciclo de vida, constituindo-se como um alvo promissor para o controle dessa parasitemia.

Quiste aracnoideo (Galassi Tipo II) y síntomas neuropsiquiátricos: a propósito de un caso / Arachnoid Cyst (Galassi Type II) and neuropsychiatry symptoms: about a case

INTRODUCCIÓN: un quiste aracnoideo es originado de la alteración de una de las membranas de la meninges (aracnoides), de predominio en la fosa craneal media; es una patología poco común que cause síntomas y si ocurren, se pueden presentar manifestaciones neuropsiquiatrías. OBJETIVO: exponer los datos clínicos; y la metodología diagnóstica y terapéutica. CASO: presentamos una paciente femenina de 24 años de edad; sin antecedentes personales médicos psiquiátricos y médicos no psiquiátricos conocidos, quien presenta una historia con un mes de evolución de síntomas psicóticos y cambios conductuales. Se le realizo CAT cerebral simple y luego una resonancia magnética cerebral contrastada para definir el tamaño del quiste aracnoideo, por los posibles síntomas neuropsiquiátricos encontrados. Por medio de exámenes de laboratorios, estudios electrofisiológicos (electroencefalograma), neuroimágenes y evaluación clínica. Se decide presentar las características clínicas encontradas de la paciente quien requirió manejo con antipsicóticos, benzodiacepinas y estabilizador del humor con gradual mejoría de sus sintomatologías de ingreso (agitación psicomotora y psicosis).

INTRODUCTION: an arachnoid cyst is caused by the alteration of one of the membranes of the meninges (arachnoid), predominantly in the middle cranial fossa; It is an uncommon pathology that causes symptoms and if they occur, neuropsychiatric manifestations may take place. OBJECTIVE: expose the clinical data; and the diagnostic and therapeutic methodology. CASE: we present a 24-year-old female patient; with no personal history of psychiatric and known non-psychiatric medical records, whom presents a story with a month of evolution of psychotic symptoms and behavioral changes. A simple cerebral CAT was performed and then a cerebral magnetic resonance imaging with contrast to define the size of the arachnoid cyst, due to the possible neuropsychiatric symptoms found. Through laboratory tests, electrophysiological studies (electroencephalogram), neuroimaging and clinical evaluation. It was decided to present the clinical characteristics of the patient who required management with antipsychotics, benzodiazepines and mood stabilizer with gradual improvement of her admission symptoms (psychomotor agitation and psychosis)

Outcomes of peroral endoscopic myotomy in patients with spastic esophageal motility disorders that do not fulfill Chicago Classification criteria.

INTRODUCTION: per-oral endoscopic myotomy (POEM) has become a mainstream treatment for achalasia and is a promising therapy in spastic disorders. METHODS: this is a retrospective study of prospectively collected data (case series). We present the first results of the use of POEM in patients with atypical spastic esophageal motor disorders that do not satisfy current Chicago Classification criteria. Seven consecutive patients with troublesome and persistent symptoms (12-180 months) related to atypical spastic esophageal motor dysfunction were systematically assessed before and after POEM, the extent of which was tailored by manometric findings. In five of the patients, other endoscopic or surgical procedures had failed. RESULTS: high-resolution manometry (HRM) showed a spastic esophageal body contractile segment in varying positions and lengths along the esophageal body which did not meet Chicago Classification criteria. After POEM, dysphagia and/or chest pain had either resolved or was greatly reduced. HRM 3-6 months after myotomy showed that the regions of spastic contraction targeted by myotomy had been ablated. There were no major complications. The clinical responses were fully maintained up to the most recent assessments after POEM (range 7-44 months). CONCLUSION: in our seven patients, POEM was a highly effective treatment for patients with troublesome symptoms related to atypical spastic esophageal motility disorders.

Trustworthy AI: Closing the gap between development and integration of AI systems in ophthalmic practice.

An increasing number of artificial intelligence (AI) systems are being proposed in ophthalmology, motivated by the variety and amount of clinical and imaging data, as well as their potential benefits at the different stages of patient care. Despite achieving close or even superior performance to that of experts, there is a critical gap between development and integration of AI systems in ophthalmic practice. This work focuses on the importance of trustworthy AI to close that gap. We identify the main aspects or challenges that need to be considered along the AI design pipeline so as to generate systems that meet the requirements to be deemed trustworthy, including those concerning accuracy, resiliency, reliability, safety, and accountability. We elaborate on mechanisms and considerations to address those aspects or challenges, and define the roles and responsibilities of the different stakeholders involved in AI for ophthalmic care, i.e., AI developers, reading centers, healthcare providers, healthcare institutions, ophthalmological societies and working groups or committees, patients, regulatory bodies, and payers. Generating trustworthy AI is not a responsibility of a sole stakeholder. There is an impending necessity for a collaborative approach where the different stakeholders are represented along the AI design pipeline, from the definition of the intended use to post-market surveillance after regulatory approval. This work contributes to establish such multi-stakeholder interaction and the main action points to be taken so that the potential benefits of AI reach real-world ophthalmic settings.

Adversarial attack vulnerability of medical image analysis systems: Unexplored factors.

Adversarial attacks are considered a potentially serious security threat for machine learning systems. Medical image analysis (MedIA) systems have recently been argued to be vulnerable to adversarial attacks due to strong financial incentives and the associated technological infrastructure. In this paper, we study previously unexplored factors affecting adversarial attack vulnerability of deep learning MedIA systems in three medical domains: ophthalmology, radiology, and pathology. We focus on adversarial black-box settings, in which the attacker does not have full access to the target model and usually uses another model, commonly referred to as surrogate model, to craft adversarial examples that are then transferred to the target model. We consider this to be the most realistic scenario for MedIA systems. Firstly, we study the effect of weight initialization (pre-training on ImageNet or random initialization) on the transferability of adversarial attacks from the surrogate model to the target model, i.e., how effective attacks crafted using the surrogate model are on the target model. Secondly, we study the influence of differences in development (training and validation) data between target and surrogate models. We further study the interaction of weight initialization and data differences with differences in model architecture. All experiments were done with a perturbation degree tuned to ensure maximal transferability at minimal visual perceptibility of the attacks. Our experiments show that pre-training may dramatically increase the transferability of adversarial examples, even when the target and surrogate's architectures are different: the larger the performance gain using pre-training, the larger the transferability. Differences in the development data between target and surrogate models considerably decrease the performance of the attack; this decrease is further amplified by difference in the model architecture. We believe these factors should be considered when developing security-critical MedIA systems planned to be deployed in clinical practice. We recommend avoiding using only standard components, such as pre-trained architectures and publicly available datasets, as well as disclosure of design specifications, in addition to using adversarial defense methods. When evaluating the vulnerability of MedIA systems to adversarial attacks, various attack scenarios and target-surrogate differences should be simulated to achieve realistic robustness estimates. The code and all trained models used in our experiments are publicly available.3.

Case-Control Analysis of the Impact of Anemia on Quality of Life in Patients with Cancer: A Qca Study Analysis.

Anemia is a common condition in cancer patients and is associated with a wide variety of symptoms that impair quality of life (QoL). However, exactly how anemia affects QoL in cancer patients is unclear because of the inconsistencies in its definition in previous reports. We aimed to examine the clinical impact of anemia on the QoL of cancer patients using specific questionnaires. We performed a post-hoc analysis of a multicenter, prospective, case-control study. We included patients with cancer with (cases) or without (controls) anemia. Participants completed the European Organization for Research and Treatment of Cancer Quality of Life questionnaire version 3.0 (EORTC QLQ-C30) and Euro QoL 5-dimension 3-level (EQ-5D-3L) questionnaire. Statistically significant and clinically relevant differences in the global health status were examined. From 2015 to 2018, 365 patients were included (90 cases and 275 controls). We found minimally important differences in global health status according to the EORTC QLQ-C30 questionnaire (case vs. controls: 45.6 vs. 58%, respectively; mean difference: -12.4, p < 0.001). Regarding symptoms, cancer patients with anemia had more pronounced symptoms in six out of nine scales in comparison with those without anemia. In conclusion, cancer patients with anemia had a worse QoL both clinically and statistically.

Iterative Augmentation of Visual Evidence for Weakly-Supervised Lesion Localization in Deep Interpretability Frameworks: Application to Color Fundus Images.

Interpretability of deep learning (DL) systems is gaining attention in medical imaging to increase experts' trust in the obtained predictions and facilitate their integration in clinical settings. We propose a deep visualization method to generate interpretability of DL classification tasks in medical imaging by means of visual evidence augmentation. The proposed method iteratively unveils abnormalities based on the prediction of a classifier trained only with image-level labels. For each image, initial visual evidence of the prediction is extracted with a given visual attribution technique. This provides localization of abnormalities that are then removed through selective inpainting. We iteratively apply this procedure until the system considers the image as normal. This yields augmented visual evidence, including less discriminative lesions which were not detected at first but should be considered for final diagnosis. We apply the method to grading of two retinal diseases in color fundus images: diabetic retinopathy (DR) and age-related macular degeneration (AMD). We evaluate the generated visual evidence and the performance of weakly-supervised localization of different types of DR and AMD abnormalities, both qualitatively and quantitatively. We show that the augmented visual evidence of the predictions highlights the biomarkers considered by experts for diagnosis and improves the final localization performance. It results in a relative increase of 11.2± 2.0% per image regarding sensitivity averaged at 10 false positives/image on average, when applied to different classification tasks, visual attribution techniques and network architectures. This makes the proposed method a useful tool for exhaustive visual support of DL classifiers in medical imaging.

Initial and Repeated Point Prevalence Surveys to Inform SARS-CoV-2 Infection Prevention in 26 Skilled Nursing Facilities - Detroit, Michigan, March-May 2020.

Skilled nursing facilities (SNFs) are focal points of the coronavirus disease 2019 (COVID-19) pandemic, and asymptomatic infections with SARS-CoV-2, the virus that causes COVID-19, among SNF residents and health care personnel have been described (1-3). Repeated point prevalence surveys (serial testing of all residents and health care personnel at a health care facility irrespective of symptoms) have been used to identify asymptomatic infections and have reduced SARS-CoV-2 transmission during SNF outbreaks (1,3). During March 2020, the Detroit Health Department and area hospitals detected a sharp increase in COVID-19 diagnoses, hospitalizations, and associated deaths among SNF residents. The Detroit Health Department collaborated with local government, academic, and health care system partners and a CDC field team to rapidly expand SARS-CoV-2 testing and implement infection prevention and control (IPC) activities in all Detroit-area SNFs. During March 7-May 8, among 2,773 residents of 26 Detroit SNFs, 1,207 laboratory-confirmed cases of COVID-19 were identified during three periods: before (March 7-April 7) and after two point prevalence surveys (April 8-25 and April 30-May 8): the overall attack rate was 44%. Within 21 days of receiving their first positive test results, 446 (37%) of 1,207 COVID-19 patients were hospitalized, and 287 (24%) died. Among facilities participating in both surveys (n = 12), the percentage of positive test results declined from 35% to 18%. Repeated point prevalence surveys in SNFs identified asymptomatic COVID-19 cases, informed cohorting and IPC practices aimed at reducing transmission, and guided prioritization of health department resources for facilities experiencing high levels of SARS-CoV-2 transmission. With the increased availability of SARS-CoV-2 testing, repeated point prevalence surveys and enhanced and expanded IPC support should be standard tools for interrupting and preventing COVID-19 outbreaks in SNFs.

A Deep Learning Model for Segmentation of Geographic Atrophy to Study Its Long-Term Natural History.

PURPOSE: To develop and validate a deep learning model for the automatic segmentation of geographic atrophy (GA) using color fundus images (CFIs) and its application to study the growth rate of GA. DESIGN: Prospective, multicenter, natural history study with up to 15 years of follow-up. PARTICIPANTS: Four hundred nine CFIs of 238 eyes with GA from the Rotterdam Study (RS) and Blue Mountain Eye Study (BMES) for model development, and 3589 CFIs of 376 eyes from the Age-Related Eye Disease Study (AREDS) for analysis of GA growth rate. METHODS: A deep learning model based on an ensemble of encoder-decoder architectures was implemented and optimized for the segmentation of GA in CFIs. Four experienced graders delineated, in consensus, GA in CFIs from the RS and BMES. These manual delineations were used to evaluate the segmentation model using 5-fold cross-validation. The model was applied further to CFIs from the AREDS to study the growth rate of GA. Linear regression analysis was used to study associations between structural biomarkers at baseline and the GA growth rate. A general estimate of the progression of GA area over time was made by combining growth rates of all eyes with GA from the AREDS set. MAIN OUTCOME MEASURES: Automatically segmented GA and GA growth rate. RESULTS: The model obtained an average Dice coefficient of 0.72±0.26 on the BMES and RS set while comparing the automatically segmented GA area with the graders' manual delineations. An intraclass correlation coefficient of 0.83 was reached between the automatically estimated GA area and the graders' consensus measures. Nine automatically calculated structural biomarkers (area, filled area, convex area, convex solidity, eccentricity, roundness, foveal involvement, perimeter, and circularity) were significantly associated with growth rate. Combining all growth rates indicated that GA area grows quadratically up to an area of approximately 12 mm2, after which growth rate stabilizes or decreases. CONCLUSIONS: The deep learning model allowed for fully automatic and robust segmentation of GA on CFIs. These segmentations can be used to extract structural characteristics of GA that predict its growth rate.