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.

Machine learning for real-time aggregated prediction of hospital admission for emergency patients.

Machine learning for hospital operations is under-studied. We present a prediction pipeline that uses live electronic health-records for patients in a UK teaching hospital's emergency department (ED) to generate short-term, probabilistic forecasts of emergency admissions. A set of XGBoost classifiers applied to 109,465 ED visits yielded AUROCs from 0.82 to 0.90 depending on elapsed visit-time at the point of prediction. Patient-level probabilities of admission were aggregated to forecast the number of admissions among current ED patients and, incorporating patients yet to arrive, total emergency admissions within specified time-windows. The pipeline gave a mean absolute error (MAE) of 4.0 admissions (mean percentage error of 17%) versus 6.5 (32%) for a benchmark metric. Models developed with 104,504 later visits during the Covid-19 pandemic gave AUROCs of 0.68-0.90 and MAE of 4.2 (30%) versus a 4.9 (33%) benchmark. We discuss how we surmounted challenges of designing and implementing models for real-time use, including temporal framing, data preparation, and changing operational conditions.

An interaction of inorganic arsenic exposure with body weight and composition on type 2 diabetes indicators in Diversity Outbred mice.

Type 2 diabetes (T2D) is a complex metabolic disorder with no cure and high morbidity. Exposure to inorganic arsenic (iAs), a ubiquitous environmental contaminant, is associated with increased T2D risk. Despite growing evidence linking iAs exposure to T2D, the factors underlying inter-individual differences in susceptibility remain unclear. This study examined the interaction between chronic iAs exposure and body composition in a cohort of 75 Diversity Outbred mice. The study design mimics that of an exposed human population where the genetic diversity of the mice provides the variation in response, in contrast to a design that includes untreated mice. Male mice were exposed to iAs in drinking water (100 ppb) for 26 weeks. Metabolic indicators used as diabetes surrogates included fasting blood glucose and plasma insulin (FBG, FPI), blood glucose and plasma insulin 15 min after glucose challenge (BG15, PI15), homeostatic model assessment for [Formula: see text]-cell function and insulin resistance (HOMA-B, HOMA-IR), and insulinogenic index. Body composition was determined using magnetic resonance imaging, and the concentrations of iAs and its methylated metabolites were measured in liver and urine. Associations between cumulative iAs consumption and FPI, PI15, HOMA-B, and HOMA-IR manifested as significant interactions between iAs and body weight/composition. Arsenic speciation analyses in liver and urine suggest little variation in the mice's ability to metabolize iAs. The observed interactions accord with current research aiming to disentangle the effects of multiple complex factors on T2D risk, highlighting the need for further research on iAs metabolism and its consequences in genetically diverse mouse strains.

Cocaine-Induced Locomotor Activation Differs Across Inbred Mouse Substrains.

Cocaine use disorders (CUD) are devastating for affected individuals and impose a significant societal burden, but there are currently no FDA-approved therapies. The development of novel and effective treatments has been hindered by substantial gaps in our knowledge about the etiology of these disorders. The risk for developing a CUD is influenced by genetics, the environment and complex interactions between the two. Identifying specific genes and environmental risk factors that increase CUD risk would provide an avenue for the development of novel treatments. Rodent models of addiction-relevant behaviors have been a valuable tool for studying the genetics of behavioral responses to drugs of abuse. Traditional genetic mapping using genetically and phenotypically divergent inbred mice has been successful in identifying numerous chromosomal regions that influence addiction-relevant behaviors, but these strategies rarely result in identification of the causal gene or genetic variant. To overcome this challenge, reduced complexity crosses (RCC) between closely related inbred mouse strains have been proposed as a method for rapidly identifying and validating functional variants. The RCC approach is dependent on identifying phenotypic differences between substrains. To date, however, the study of addiction-relevant behaviors has been limited to very few sets of substrains, mostly comprising the C57BL/6 lineage. The present study expands upon the current literature to assess cocaine-induced locomotor activation in 20 inbred mouse substrains representing six inbred strain lineages (A/J, BALB/c, FVB/N, C3H/He, DBA/2 and NOD) that were either bred in-house or supplied directly by a commercial vendor. To our knowledge, we are the first to identify significant differences in cocaine-induced locomotor response in several of these inbred substrains. The identification of substrain differences allows for the initiation of RCC populations to more rapidly identify specific genetic variants associated with acute cocaine response. The observation of behavioral profiles that differ between mice generated in-house and those that are vendor-supplied also presents an opportunity to investigate the influence of environmental factors on cocaine-induced locomotor activity.

Behavioral characterization of a novel Cisd2 mutant mouse.

Wolfram syndrome (WFS) is a rare autosomal recessive disorder characterized by diabetes mellitus and insipidus, progressive optic atrophy and sensorineural deafness. An increased incidence of psychiatric disorders has also been reported in WFS patients. There are two subtypes of WFS. Type 1 (WFS1) is caused by mutations in the WFS1 gene and type 2 (WFS2) results from mutations in the CISD2 gene. Existing Wfs1 knockout mice exhibit many WFS1 cardinal symptoms including diabetic nephropathy, metabolic disruptions and optic atrophy. Far fewer studies have examined loss of Cisd2 function in mice. We identified B6.DDY-Cisd2m1Lmt, a mouse model with a spontaneous mutation in the Cisd2 gene. B6.DDY-Cisd2m1Lmt mice were initially identified based on the presence of audible sonic vocalizations as well as decreased body size and weight compared to unaffected wildtype littermates. Although Wfs1 knockout mice have been characterized for numerous behavioral phenotypes, similar studies have been lacking for Cisd2 mutant mice. We tested B6.DDY-Cisd2m1Lmt mice in a battery of behavioral assays that model phenotypes related to neurological and psychiatric disorders including anxiety, sensorimotor gating, stress response, social interaction and learning and memory. B6.DDY-Cisd2m1Lmt mice displayed hypoactivity across several behavioral tests, exhibited increased stress response and had deficits in spatial learning and memory and sensorimotor gating compared to wildtype littermates. Our data indicate that the B6.DDY-Cisd2m1Lmt mouse strain is a useful model to investigate potential mechanisms underlying the neurological and psychiatric symptoms observed in WFS.

Characterization of genetically complex Collaborative Cross mouse strains that model divergent locomotor activating and reinforcing properties of cocaine.

RATIONALE: Few effective treatments exist for cocaine use disorders due to gaps in knowledge about its complex etiology. Genetically defined animal models provide a useful tool for advancing our understanding of the biological and genetic underpinnings of addiction-related behavior and evaluating potential treatments. However, many attempts at developing mouse models of behavioral disorders were based on overly simplified single gene perturbations, often leading to inconsistent and misleading results in pre-clinical pharmacology studies. A genetically complex mouse model may better reflect disease-related behaviors. OBJECTIVES: Screening defined, yet genetically complex, intercrosses of the Collaborative Cross (CC) mice revealed two lines, RIX04/17 and RIX41/51, with extreme high and low behavioral responses to cocaine. We characterized these lines as well as their CC parents, CC004/TauUnc and CC041/TauUnc, to evaluate their utility as novel model systems for studying the biological and genetic mechanisms underlying behavioral responses to cocaine. METHODS: Behavioral responses to acute (initial locomotor sensitivity) and repeated (behavioral sensitization, conditioned place preference, intravenous self-administration) exposures to cocaine were assessed. We also examined the monoaminergic system (striatal tissue content and in vivo fast scan cyclic voltammetry), HPA axis reactivity, and circadian rhythms as potential mechanisms for the divergent phenotypic behaviors observed in the two strains, as these systems have a previously known role in mediating addiction-related behaviors. RESULTS: RIX04/17 and 41/51 show strikingly divergent initial locomotor sensitivity to cocaine with RIX04/17 exhibiting very high and RIX41/51 almost no response. The lines also differ in the emergence of behavioral sensitization with RIX41/51 requiring more exposures to exhibit a sensitized response. Both lines show conditioned place preference for cocaine. We determined that the cocaine sensitivity phenotype in each RIX line was largely driven by the genetic influence of one CC parental strain, CC004/TauUnc and CC041/TauUnc. CC004 demonstrates active operant cocaine self-administration and CC041 is unable to acquire under the same testing conditions, a deficit which is specific to cocaine as both strains show operant response for a natural food reward. Examination of potential mechanisms driving differential responses to cocaine show strain differences in molecular and behavioral circadian rhythms. Additionally, while there is no difference in striatal dopamine tissue content or dynamics, there are selective differences in striatal norepinephrine and serotonergic tissue content. CONCLUSIONS: These CC strains offer a complex polygenic model system to study underlying mechanisms of cocaine response. We propose that CC041/TauUnc and CC004/TauUnc will be useful for studying genetic and biological mechanisms underlying resistance or vulnerability to the stimulatory and reinforcing effects of cocaine.

Reciprocal F1 Hybrids of Two Inbred Mouse Strains Reveal Parent-of-Origin and Perinatal Diet Effects on Behavior and Expression.

Parent-of-origin effects (POE) in mammals typically arise from maternal effects or imprinting. In some instances, such POE have been associated with psychiatric disorders, as well as with changes in a handful of animal behaviors. However, POE on complex traits such as behavior remain largely uncharacterized. Moreover, although both behavior and epigenetic effects are known to be modified by perinatal environmental exposures such as nutrient deficiency, the architecture of such environment-by-POE is mostly unexplored. To study POE and environment-by-POE, we employ a relatively neglected but especially powerful experimental system for POE-detection: reciprocal F1 hybrids (RF1s). We exposed female NOD/ShiLtJ×C57Bl/6J and C57Bl/6J×NOD/ShiLtJ mice, perinatally, to one of four different diets, then after weaning recorded a set of behaviors that model psychiatric disease. Whole-brain microarray expression data revealed an imprinting-enriched set of 15 genes subject to POE. The most-significant expression POE, on the non-imprinted gene Carmil1 (a.k.a. Lrrc16a), was validated using qPCR in the same and in a new set of mice. Several behaviors, especially locomotor behaviors, also showed POE. Bayesian mediation analysis suggested Carmil1 expression suppresses behavioral POE, and that the imprinted gene Airn suppresses POE on Carmil1 expression. A suggestive diet-by-POE was observed on percent center time in the open field test, and a significant diet-by-POE was observed on one imprinted gene, Mir341, and on 16 non-imprinted genes. The relatively small, tractable set of POE and diet-by-POE detected on behavior and expression here motivates further studies examining such effects across RF1s on multiple genetic backgrounds.

Neural activation deficits in a mouse genetic model of NMDA receptor hypofunction in tests of social aggression and swim stress.

Mice with reduced expression of the NR1 subunit of the NMDA receptor (NR1 hypomorphic mice) display altered behavioral phenotypes that may relate to behavioral characteristics of schizophrenia. Altered phenotypes in the NR1 hypomorphs include marked deficits in species-typical behavioral interactions in tests of social aggression and social affiliation. To gain insight into neuroanatomical circuits disrupted by reduced NMDA receptor function, the present work compared regional brain activation in NR1 hypomorphic mice and their wild type controls after a resident-intruder test. Induction of Fos protein was used as an index of neuronal activation. Wild type mice exhibited robust induction of Fos in select brain regions, including specific nuclei of the hypothalamus and amygdala, lateral septum, and widespread regions of the cerebral cortex. Although the behavioral patterns were different for male and female mice, neuroanatomical patterns of Fos induction were remarkably similar for the two sexes. To determine socially specific components of Fos induction by the resident-intruder test, responses were compared for mice assessed in a test of general arousal and stress involving forced swim. Some common brain regions were activated by both tests but regionally specific differences were also found. The NR1 hypomorphic mice tested in the resident-intruder procedure displayed distinctly different behavioral interactions compared to the wild type mice and exhibited a significantly blunted Fos response in almost all brain regions. The mutant mice also exhibited reduced Fos in response to swim stress in specific brain regions. These data suggest that the NR1 hypomorphic mice have functional activation deficits in response to social challenge and swim stress.

Seizure responses and induction of Fos by the NMDA agonist (tetrazol-5-yl)glycine in a genetic model of NMDA receptor hypofunction.

Effects of the direct NMDA agonist (tetrazol-5-yl)glycine (TZG) were examined in a genetic mouse model of reduced NMDA receptor function. In this model, expression of the NR1 subunit is reduced but not eliminated and the mice are therefore designated as NR1 hypomorphic. Previous work suggested that the reduced NR1 subunit expression produced a functional subsensitivity as judged by a blunted Fos induction response to a sub-seizure dose of TZG. In the present study seizure threshold doses of TZG were tested in the wild type and mutant mice. Surprisingly, there was no difference in the seizure sensitivity between the wild type mice and mice presumed to express very low levels of the NR1 subunit. An extensive neuroanatomical analysis of Fos induction was conducted after the threshold seizure doses of TZG. The results demonstrate that some brain regions of the NR1 -/- mice exhibit much lower Fos induction in comparison to the NR1 +/+ mice. These regions include hippocampus, amygdala, and cerebral cortical regions. However, in other regions, similar induction of Fos was observed in both genotypes in response to the NMDA agonist. Regions showing similar Fos induction in the NR1 +/+ and NR1 -/- mice include the lateral septum, nucleus of the solitary tract, and medial hypothalamic regions. The results suggest that the NMDA receptor hypofunction in the NR1 -/- mice is not global but regionally specific and that subcortical structures are responsible for the seizure-inducing effects of TZG.