Enhancing Global Estimation of Fine Particulate Matter Concentrations by Including Geophysical a Priori Information in Deep Learning.

Global fine particulate matter (PM2.5) assessment is impeded by a paucity of monitors. We improve estimation of the global distribution of PM2.5 concentrations by developing, optimizing, and applying a convolutional neural network with information from satellite-, simulation-, and monitor-based sources to predict the local bias in monthly geophysical a priori PM2.5 concentrations over 1998-2019. We develop a loss function that incorporates geophysical a priori estimates and apply it in model training to address the unrealistic results produced by mean-square-error loss functions in regions with few monitors. We introduce novel spatial cross-validation for air quality to examine the importance of considering spatial properties. We address the sharp decline in deep learning model performance in regions distant from monitors by incorporating the geophysical a priori PM2.5. The resultant monthly PM2.5 estimates are highly consistent with spatial cross-validation PM2.5 concentrations from monitors globally and regionally. We withheld 10% to 99% of monitors for testing to evaluate the sensitivity and robustness of model performance to the density of ground-based monitors. The model incorporating the geophysical a priori PM2.5 concentrations remains highly consistent with observations globally even under extreme conditions (e.g., 1% for training, R2 = 0.73), while the model without exhibits weaker performance (1% for training, R2 = 0.51).

Self-Supervised Learning Application on COVID-19 Chest X-ray Image Classification Using Masked AutoEncoder.

The COVID-19 pandemic has underscored the urgent need for rapid and accurate diagnosis facilitated by artificial intelligence (AI), particularly in computer-aided diagnosis using medical imaging. However, this context presents two notable challenges: high diagnostic accuracy demand and limited availability of medical data for training AI models. To address these issues, we proposed the implementation of a Masked AutoEncoder (MAE), an innovative self-supervised learning approach, for classifying 2D Chest X-ray images. Our approach involved performing imaging reconstruction using a Vision Transformer (ViT) model as the feature encoder, paired with a custom-defined decoder. Additionally, we fine-tuned the pretrained ViT encoder using a labeled medical dataset, serving as the backbone. To evaluate our approach, we conducted a comparative analysis of three distinct training methods: training from scratch, transfer learning, and MAE-based training, all employing COVID-19 chest X-ray images. The results demonstrate that MAE-based training produces superior performance, achieving an accuracy of 0.985 and an AUC of 0.9957. We explored the mask ratio influence on MAE and found ratio = 0.4 shows the best performance. Furthermore, we illustrate that MAE exhibits remarkable efficiency when applied to labeled data, delivering comparable performance to utilizing only 30% of the original training dataset. Overall, our findings highlight the significant performance enhancement achieved by using MAE, particularly when working with limited datasets. This approach holds profound implications for future disease diagnosis, especially in scenarios where imaging information is scarce.

Assessment and management of secondary bacterial infections complicating Mpox (Monkeypox) using a telemedicine service. A prospective cohort study.

INTRODUCTION: During spring 2022, an outbreak of Monkeypox (mpox) emerged as an infection of concern in Europe. Due to the overlapping clinical features of mpox and bacterial infections, diagnosis of concomitant bacterial infection is challenging. In this prospective cohort study, we report the incidence, severity, and progression of patients with secondary bacterial infection complicating mpox infection. METHOD: Data were collected via a bespoke mpox telemedicine service provided by Infection services at North Manchester General Hospital, UK. A diagnosis of secondary bacterial infection was based on the history (balanitis, surrounding erythema, purulent discharge and nasal ulceration) and review of patient-collected medical photography. Patient were reviewed face-to-face where necessary. RESULTS: Secondary bacterial infection was diagnosed in 15 of 129 (11.6%) patients with mpox. Three patients with secondary bacterial infection (3/129, 2.3%) required admission to hospital and one patient underwent surgical debridement. Median healing (thus, isolation) times were longer in those with bacterial infection. DISCUSSION: In this prospective cohort study of patients with mpox, secondary bacterial infection was infrequent and predominantly mild. The virtual ward and telemedicine follow up allowed for the prompt recognition of secondary bacterial infections and timely antibiotic administration. Due to concerns regarding nosocomial transmission, mild clinical course and limited inpatient bed capacity, we believe this model of outpatient management for mpox (Clade II B.1 lineage) could be replicated in other low risk populations where suitable home isolation facilities exist.

Efficient Training on Alzheimer's Disease Diagnosis with Learnable Weighted Pooling for 3D PET Brain Image Classification.

Three-dimensional convolutional neural networks (3D CNNs) have been widely applied to analyze Alzheimer's disease (AD) brain images for a better understanding of the disease progress or predicting the conversion from cognitively impaired (CU) or mild cognitive impairment status. It is well-known that training 3D-CNN is computationally expensive and with the potential of overfitting due to the small sample size available in the medical imaging field. Here we proposed a novel 3D-2D approach by converting a 3D brain image to a 2D fused image using a Learnable Weighted Pooling (LWP) method to improve efficient training and maintain comparable model performance. By the 3D-to-2D conversion, the proposed model can easily forward the fused 2D image through a pre-trained 2D model while achieving better performance over different 3D and 2D baselines. In the implementation, we chose to use ResNet34 for feature extraction as it outperformed other 2D CNN backbones. We further showed that the weights of the slices are location-dependent and the model performance relies on the 3D-to-2D fusion view, with the best outcomes from the coronal view. With the new approach, we were able to reduce 75% of the training time and increase the accuracy to 0.88, compared with conventional 3D CNNs, for classifying amyloid-beta PET imaging from the AD patients from the CU participants using the publicly available Alzheimer's Disease Neuroimaging Initiative dataset. The novel 3D-2D model may have profound implications for timely AD diagnosis in clinical settings in the future.

Clinical features and management of individuals admitted to hospital with monkeypox and associated complications across the UK: a retrospective cohort study.

BACKGROUND: The scale of the 2022 global mpox (formerly known as monkeypox) outbreak has been unprecedented. In less than 6 months, non-endemic countries have reported more than 67 000 cases of a disease that had previously been rare outside of Africa. Mortality has been reported as rare but hospital admission has been relatively common. We aimed to describe the clinical and laboratory characteristics and outcomes of individuals admitted to hospital with mpox and associated complications, including tecovirimat recipients. METHODS: In this cohort study, we undertook retrospective review of electronic clinical records and pathology data for all individuals admitted between May 6, and Aug 3, 2022, to 16 hospitals from the Specialist and High Consequence Infectious Diseases Network for Monkeypox. The hospitals were located in ten cities in England and Northern Ireland. Inclusion criteria were clinical signs consistent with mpox and MPXV DNA detected from at least one clinical sample by PCR testing. Patients admitted solely for isolation purposes were excluded from the study. Key outcomes included admission indication, complications (including pain, secondary infection, and mortality) and use of antibiotic and anti-viral treatments. Routine biochemistry, haematology, microbiology, and virology data were also collected. Outcomes were assessed in all patients with available data. FINDINGS: 156 individuals were admitted to hospital with complicated mpox during the study period. 153 (98%) were male and three (2%) were female, with a median age of 35 years (IQR 30-44). Gender data were collected from electronic patient records, which encompassed full formal review of clincian notes. The prespecified options for data collection for gender were male, female, trans, non-binary, or unknown. 105 (71%) of 148 participants with available ethnicity data were of White ethnicity and 47 (30%) of 155 were living with HIV with a median CD4 count of 510 cells per mm3 (IQR 349-828). Rectal or perianal pain (including proctitis) was the most common indication for hospital admission (44 [28%] of 156). Severe pain was reported in 89 (57%) of 156, and secondary bacterial infection in 82 (58%) of 142 individuals with available data. Median admission duration was 5 days (IQR 2-9). Ten individuals required surgery and two cases of encephalitis were reported. 38 (24%) of the 156 individuals received tecovirimat with early cessation in four cases (two owing to hepatic transaminitis, one to rapid treatment response, and one to patient choice). No deaths occurred during the study period. INTERPRETATION: Although life-threatening mpox appears rare in hospitalised populations during the current outbreak, severe mpox and associated complications can occur in immunocompetent individuals. Analgesia and management of superimposed bacterial infection are priorities for patients admitted to hospital. FUNDING: None.

BlpC-mediated selfish program leads to rapid loss of Streptococcus pneumoniae clonal diversity during infection.

Successful colonization of a host requires bacterial adaptation through genetic and population changes that are incompletely defined. Using chromosomal barcoding and high-throughput sequencing, we investigate the population dynamics of Streptococcus pneumoniae during infant mouse colonization. Within 1 day post inoculation, diversity was reduced >35-fold with expansion of a single clonal lineage. This loss of diversity was not due to immune factors, microbiota, or exclusive genetic drift. Rather, bacteriocins induced by the BlpC-quorum sensing pheromone resulted in predation of kin cells. In this intra-strain competition, the subpopulation reaching a quorum likely eliminates others that have yet to activate the blp locus. Additionally, this reduced diversity restricts the number of unique clones that establish colonization during transmission between hosts. Genetic variation in the blp locus was also associated with altered transmissibility in a human population, further underscoring the importance of BlpC in clonal selection and its role as a selfish element.

Development of CNN models for the enteral feeding tube positioning assessment on a small scale data set.

BACKGROUND: Enteral nutrition through feeding tubes serves as the primary method of nutritional supplementation for patients unable to feed themselves. Plain radiographs are routinely used to confirm the position of the Nasoenteric feeding tubes the following insertion and before the commencement of tube feeds. Convolutional neural networks (CNNs) have shown encouraging results in assisting the tube positioning assessment. However, robust CNNs are often trained using large amounts of manually annotated data, which challenges applying CNNs on enteral feeding tube positioning assessment. METHOD: We build a CNN model for feeding tube positioning assessment by pre-training the model under a weakly supervised fashion on large quantities of radiographs. Since most of the model was pre-trained, a small amount of labeled data is needed when fine-tuning the model for tube positioning assessment. We demonstrate the proposed method using a small dataset with 175 radiographs. RESULT: The experimental result shows that the proposed model improves the area under the receiver operating characteristic curve (AUC) by up to 35.71% , from 0.56 to 0.76, and 14.49% on the accuracy, from 0.69 to 0.79 when compared with the no pre-trained method. The proposed method also has up to 40% less error when estimating its prediction confidence. CONCLUSION: Our evaluation results show that the proposed model has a high prediction accuracy and a more accurate estimated prediction confidence when compared to the no pre-trained model and other baseline models. The proposed method can be potentially used for assessing the enteral tube positioning. It also provides a strong baseline for future studies.

Contrastive Cross-Modal Pre-Training: A General Strategy for Small Sample Medical Imaging.

A key challenge in training neural networks for a given medical imaging task is the difficulty of obtaining a sufficient number of manually labeled examples. In contrast, textual imaging reports are often readily available in medical records and contain rich but unstructured interpretations written by experts as part of standard clinical practice. We propose using these textual reports as a form of weak supervision to improve the image interpretation performance of a neural network without requiring additional manually labeled examples. We use an image-text matching task to train a feature extractor and then fine-tune it in a transfer learning setting for a supervised task using a small labeled dataset. The end result is a neural network that automatically interprets imagery without requiring textual reports during inference. We evaluate our method on three classification tasks and find consistent performance improvements, reducing the need for labeled data by 67%-98%.

Mammalian Orthoreovirus Reassortment Proceeds with Little Constraint on Segment Mixing.

Segmentation of viral genomes provides the potential for genetic exchange within coinfected cells. However, for this potential to be realized, coinfecting genomes must mix during the viral life cycle. The efficiency of reassortment, in turn, dictates its potential to drive evolution. The opportunity for mixing within coinfected cells may vary greatly across virus families, such that the evolutionary implications of genome segmentation differ as a result of core features of the viral life cycle. To investigate the relationship between viral replication compartments and genetic exchange, we quantified reassortment in mammalian orthoreovirus (reovirus). Reoviruses carry a 10-segmented, double-stranded RNA genome, which is replicated within proteinaceous structures termed inclusion bodies. We hypothesized that inclusions impose a barrier to reassortment. We quantified reassortment between wild-type (wt) and variant (var) reoviruses that differ by one nucleotide per segment. Studies of wt/var systems in both T1L and T3D backgrounds revealed frequent reassortment without bias toward particular genotypes. However, reassortment was more efficient in the T3D serotype. Since T1L and T3D viruses exhibit different inclusion body morphologies, we tested the impact of this phenotype on reassortment. In both serotypes, reassortment levels did not differ by inclusion morphology. Reasoning that the merging of viral inclusions may be critical for genome mixing, we then tested the effect of blocking merging. Reassortment proceeded efficiently even under these conditions. Our findings indicate that reovirus reassortment is highly efficient despite the localization of many viral processes to inclusion bodies, and that the robustness of this genetic exchange is independent of inclusion body structure and fusion. IMPORTANCE Quantification of reassortment in diverse viral systems is critical to elucidate the implications of genome segmentation for viral evolution. In principle, genome segmentation offers a facile means of genetic exchange between coinfecting viruses. In practice, there may be physical barriers within the cell that limit the mixing of viral genomes. Here, we tested the hypothesis that localization of the various stages of the mammalian orthoreovirus life cycle within cytoplasmic inclusion bodies compartmentalizes viral replication and limits genetic exchange. Contrary to this hypothesis, our data indicate that reovirus reassortment occurs readily within coinfected cells and is not strongly affected by the structure or dynamics of viral inclusion bodies. We conclude that the potential for reassortment to contribute to reovirus evolution is high.

Alzheimer's Disease Classification Using 2D Convolutional Neural Networks.

Alzheimer's disease (AD) is a non-treatable and non-reversible disease that affects about 6% of people who are 65 and older. Brain magnetic resonance imaging (MRI) is a pseudo-3D imaging technology that is widely used for AD diagnosis. Convolutional neural networks with 3D kernels (3D CNNs) are often the default choice for deep learning based MRI analysis. However, 3D CNNs are usually computationally costly and data-hungry. Such disadvantages post a barrier of using modern deep learning techniques in the medical imaging domain, in which the number of data that can be used for training is usually limited. In this work, we propose three approaches that leverage 2D CNNs on 3D MRI data. We test the proposed methods on the Alzheimer's Disease Neuroimaging Initiative dataset across two popular 2D CNN architectures. The evaluation results show that the proposed method improves the model performance on AD diagnosis by 8.33% accuracy or 10.11% auROC compared with the ResNet-based 3D CNN model, while significantly reducing the training time by over 89%. We also discuss the potential causes for performance improvement and the limitations. We believe this work can serve as a strong baseline for future researchers.

Multi-Modal Data Analysis for Alzheimer's Disease Diagnosis: An Ensemble Model Using Imagery and Genetic Features.

Alzheimer's disease (AD) is a devastating neurological disorder primarily affecting the elderly. An estimated 6.2 million Americans age 65 and older are suffering from Alzheimer's dementia today. Brain magnetic resonance imaging (MRI) is widely used for the clinical diagnosis of AD. In the meanwhile, medical researchers have identified 40 risk locus using single-nucleotide polymorphisms (SNPs) information from Genome-wide association study (GWAS) in the past decades. However, existing studies usually treat MRI and GWAS separately. For instance, convolutional neural networks are often trained using MRI for AD diagnosis. GWAS and SNPs are frequently used to identify genomic traits. In this study, we propose a multi-modal AD diagnosis neural network that uses both MRIs and SNPs. The proposed method demonstrates a novel way to use GWAS findings by directly including SNPs in predictive models. We test the proposed methods on the Alzheimer's Disease Neuroimaging Initiative dataset. The evaluation results show that the proposed method improves the model performance on AD diagnosis and achieves 93.5% AUC and 96.1% AP, respectively, when patients have both MRI and SNP data. We believe this work brings exciting new insights to GWAS applications and sheds light on future research directions.

Role of chromosome 1q copy number variation in hepatocellular carcinoma.

Chromosome 1q often has been observed to be amplified in hepatocellular carcinoma. This review summarizes literature reports of multiple genes that have been proposed as possible 1q amplification drivers. These largely fall within 1q21-1q23. In addition, publicly available copy number alteration data from The Cancer Genome Atlas project were used to identify additional candidate genes involved in carcinogenesis. The most frequent location for gene amplification was 1q22, consistent with the results of the literature search. The genes TPM3 and NUF2 were found to be candidates whose amplification and/or mRNA up-regulation was most highly associated with poorer hepatocellular carcinoma outcomes.

Weakly-Supervised Self-Training for Breast Cancer Localization.

The use of deep learning methods has dramatically increased the state-of-the-art performance in image object localization. However, commonly used supervised learning methods require large training datasets with pixel-level or bounding box annotations. Obtaining such fine-grained annotations is extremely costly, especially in the medical imaging domain. In this work, we propose a novel weakly supervised method for breast cancer localization. The essential advantage of our approach is that the model only requires image-level labels and uses a self-training strategy to refine the predicted localization in a step-wise manner. We evaluated our approach on a large, clinically relevant mammogram dataset. The results show that our model significantly improves performance compared to other methods trained similarly.

ß-amyloid and tau drive early Alzheimer's disease decline while glucose hypometabolism drives late decline.

Clinical trials focusing on therapeutic candidates that modify ß-amyloid (Aß) have repeatedly failed to treat Alzheimer's disease (AD), suggesting that Aß may not be the optimal target for treating AD. The evaluation of Aß, tau, and neurodegenerative (A/T/N) biomarkers has been proposed for classifying AD. However, it remains unclear whether disturbances in each arm of the A/T/N framework contribute equally throughout the progression of AD. Here, using the random forest machine learning method to analyze participants in the Alzheimer's Disease Neuroimaging Initiative dataset, we show that A/T/N biomarkers show varying importance in predicting AD development, with elevated biomarkers of Aß and tau better predicting early dementia status, and biomarkers of neurodegeneration, especially glucose hypometabolism, better predicting later dementia status. Our results suggest that AD treatments may also need to be disease stage-oriented with Aß and tau as targets in early AD and glucose metabolism as a target in later AD.

Collective interactions augment influenza A virus replication in a host-dependent manner.

Infection with a single influenza A virus (IAV) is only rarely sufficient to initiate productive infection. Instead, multiple viral genomes are often required in a given cell. Here, we show that the reliance of IAV on multiple infection can form an important species barrier. Namely, we find that avian H9N2 viruses representative of those circulating widely at the poultry-human interface exhibit acute dependence on collective interactions in mammalian systems. This need for multiple infection is greatly reduced in the natural host. Quantification of incomplete viral genomes showed that their complementation accounts for the moderate reliance on multiple infection seen in avian cells but not the added reliance seen in mammalian cells. An additional form of virus-virus interaction is needed in mammals. We find that the PA gene segment is a major driver of this phenotype and that both viral replication and transcription are affected. These data indicate that multiple distinct mechanisms underlie the reliance of IAV on multiple infection and underscore the importance of virus-virus interactions in IAV infection, evolution and emergence.

Inconsistent Performance of Deep Learning Models on Mammogram Classification.

OBJECTIVES: Performance of recently developed deep learning models for image classification surpasses that of radiologists. However, there are questions about model performance consistency and generalization in unseen external data. The purpose of this study is to determine whether the high performance of deep learning on mammograms can be transferred to external data with a different data distribution. MATERIALS AND METHODS: Six deep learning models (three published models with high performance and three models designed by us) were evaluated on four different mammogram data sets, including three public (Digital Database for Screening Mammography, INbreast, and Mammographic Image Analysis Society) and one private data set (UKy). The models were trained and validated on either Digital Database for Screening Mammography alone or a combined data set that included Digital Database for Screening Mammography. The models were then tested on the three external data sets. The area under the receiver operating characteristic curve (auROC) was used to evaluate model performance. RESULTS: The three published models reported validation auROC scores between 0.88 and 0.95 on the validation data set. Our models achieved between 0.71 (95% confidence interval [CI]: 0.70-0.72) and 0.79 (95% CI: 0.78-0.80) auROC on the same validation data set. However, the same evaluation criteria of all six models on the three external test data sets were significantly decreased, only between 0.44 (95% CI: 0.43-0.45) and 0.65 (95% CI: 0.64-0.66). CONCLUSION: Our results demonstrate performance inconsistency across the data sets and models, indicating that the high performance of deep learning models on one data set cannot be readily transferred to unseen external data sets, and these models need further assessment and validation before being applied in clinical practice.

Dual Josephson Impedance Bridge: Towards a Universal Bridge for Impedance Metrology.

This paper presents a full characterization of a Dual Josephson Impedance Bridge (DJIB) at frequencies up to 80 kHz by using the DJIB to compare the best available impedance standards that are (a) directly traceable to the quantum Hall effect, (b) used as part of international impedance comparisons, or (c) believed to have calculable frequency dependence. The heart of the system is a dual Josephson Arbitrary Waveform Synthesizer (JAWS) source that offers unprecedented flexibility in high-precision impedance measurements. The JAWS sources allow a single bridge to compare impedances with arbitrary ratios and phase angles in the complex plane. The uncertainty budget shows that both the traditional METAS bridges and the DJIB have comparable uncertainties in the kilohertz range. This shows that the advantages of the DJIB, including the flexibility which allows the comparison of arbitrary impedances, the wide frequency range, and the automated balancing procedure, are obtained without compromising the measurement uncertainties. These results demonstrate that this type of instrument can considerably simplify the realization and maintenance of the various impedance scales. In addition, the DJIB is a very sensitive tool for investigating the frequency-dependent systematic-errors that can occur in impedance construction and in the voltage provided by the JAWS source at frequencies greater than 10 kHz.

Dynamic Image for 3D MRI Image Alzheimer's Disease Classification.

We propose to apply a 2D CNN architecture to 3D MRI image Alzheimer's disease classification. Training a 3D convolutional neural network (CNN) is time-consuming and computationally expensive. We make use of approximate rank pooling to transform the 3D MRI image volume into a 2D image to use as input to a 2D CNN. We show our proposed CNN model achieves 9.5% better Alzheimer's disease classification accuracy than the baseline 3D models. We also show that our method allows for efficient training, requiring only 20% of the training time compared to 3D CNN models. The code is available online: https://github.com/UkyVision/alzheimer-project.

Quantized Pulse Propagation in Josephson Junction Arrays.

We present time-domain electrical measurements and simulations of the quantized voltage pulses that are generated from series-connected Josephson junction (JJ) arrays. The transmission delay of the JJ array can lead to a broadening of the net output pulse, depending on the direction of the output pulse propagation relative to the input bias pulse. To demonstrate this, we compare time-domain measurements of output pulses from radio-frequency Josephson Arbitrary Waveform Synthesizer (RF-JAWS) circuits fabricated with two different output measurement configurations, so that the backward-propagating and forward-propagating pulses can be measured. Measurements were made on arrays with 1200 and 3600 JJs and show that the net backward-propagating output pulse is broadened by timing delays in the JJ array while the net forward-propagating output pulse is insensitive to delay effects and can theoretically be further scaled to longer JJ array lengths without significant output pulse broadening. These measurements match well with simulations and confirm the expectation that the net output pulses arise from the time-delayed superposition of individual JJ output pulses from the series array of JJs. The measurements and analysis shown here have important implications for the realization of RF-JAWS circuits to be used as quantum-based reference sources for communications metrology.

Incomplete influenza A virus genomes occur frequently but are readily complemented during localized viral spread.

Segmentation of viral genomes into multiple RNAs creates the potential for replication of incomplete viral genomes (IVGs). Here we use a single-cell approach to quantify influenza A virus IVGs and examine their fitness implications. We find that each segment of influenza A/Panama/2007/99 (H3N2) virus has a 58% probability of being replicated in a cell infected with a single virion. Theoretical methods predict that IVGs carry high costs in a well-mixed system, as 3.6 virions are required for replication of a full genome. Spatial structure is predicted to mitigate these costs, however, and experimental manipulations of spatial structure indicate that local spread facilitates complementation. A virus entirely dependent on co-infection was used to assess relevance of IVGs in vivo. This virus grows robustly in guinea pigs, but is less infectious and does not transmit. Thus, co-infection allows IVGs to contribute to within-host spread, but complete genomes may be critical for transmission.