Impaired arginine/ornithine metabolism drives severe HFMD by promoting cytokine storm.

Introduction: The Hand, Foot and Mouth Disease (HFMD), caused by enterovirus 71 infection, is a global public health emergency. Severe HFMD poses a significant threat to the life and well-being of children. Numerous studies have indicated that the occurrence of severe HFMD is associated with cytokine storm. However, the precise molecular mechanism underlying cytokine storm development remains elusive, and there are currently no safe and effective treatments available for severe HFMD in children. Methods: In this study, we established a mouse model of severe HFMD to investigate the molecular mechanisms driving cytokine storm. We specifically analyzed metabolic disturbances, focusing on arginine/ornithine metabolism, and assessed the potential therapeutic effects of spermine, an ornithine metabolite. Results: Our results identified disturbances in arginine/ornithine metabolism as a pivotal factor driving cytokine storm onset in severe HFMD cases. Additionally, we discovered that spermine effectively mitigated the inflammatory injury phenotype observed in mice with severe HFMD. Discussion: In conclusion, our findings provide novel insights into the molecular mechanisms underlying severe HFMD from a metabolic perspective while offering a promising new strategy for its safe and effective treatment.

A novel approach to managing facility airborne diseases: suppressing air pathogens with smoke aerosols generated from fungicide phase transition.

BACKGROUND: Environmental microorganisms are major contributors to the development and spread of disease. Chemical disinfection can inhibit pathogens and play a preventive role against diseases. In agriculture, prolonging the floating time of chemical pesticides in the air has a positive effect on the control of airborne diseases. However, the interaction of chemical pesticides with airborne pathogens is not yet known. RESULTS: Here, triazole fungicide was transformed into stable smoke aerosols in order to assess the feasibility of employing phase transition release pesticides for air disinfection. The phase transition had a minimal impact on hexaconazole (Hexa) and myclobutanil (Mycl), with their smoke formation rates remaining consistently >90%. In microscopic morphology, triadimenol (Tria) and epoxiconazole (Epox) are solid, and tebuconazole (Tebu), Hexa, Mycl and difenoconazole (Dife) are liquid. Liquid smoke has advantages over solid smoke in the inhibition of environmental pathogens. The floatability and spatial distribution of fungicide aerosol were optimized by the combination of smoke particles with different properties, so that the fungicide aerosol could meet the conditions of practical application. In practical applications, smoke exhibits a gentler deposition process at the target interface compared to spray, along with a more homogeneous distribution of fungicides. Moreover, fungicide smoke demonstrates superior control efficacy and leaves behind lower residual amounts on fruit. CONCLUSION: In conclusion, the implementation of fungicide phase transition as a smoke aerosol offers a viable approach to effectively suppress pathogen aerosols and enhance the control of airborne diseases. © 2024 Society of Chemical Industry.

A Hybrid Structure to Improve Electrochemical Performance of SiO Anode Materials in Lithium-Ion Battery.

The wide utilization of lithium-ion batteries (LIBs) prompts extensive research on the anode materials with large capacity and excellent stability. Despite the attractive electrochemical properties of pure Si anodes outperforming other Si-based materials, its unsafety caused by huge volumetric expansion is commonly admitted. Silicon monoxide (SiO) anode is advantageous in mild volume fluctuation, and would be a proper alternative if the low initial columbic efficiency and conductivity can be ameliorated. Herein, a hybrid structure composed of active material SiO particles and carbon nanofibers (SiO/CNFs) is proposed as a solution. CNFs, through electrospun processes, serve as a conductive skeleton for SiO nanoparticles and enable SiO nanoparticles to be uniformly embedded in. As a result, the SiO/CNF electrochemical performance reaches a peak at 20% the mass ratio of SiO, where the retention rate reaches 73.9% after 400 cycles at a current density of 100 mA g-1, and the discharge capacity after stabilization and 100 cycles are 1.47 and 1.84 times higher than that of pure SiO, respectively. A fast lithium-ion transport rate during cycling is also demonstrated as the corresponding diffusion coefficient of the SiO/CNF reaches ~8 × 10-15 cm2 s-1. This SiO/CNF hybrid structure provides a flexible and cost-effective solution for LIBs and sheds light on alternative anode choices for industrial battery assembly.

Genome analysis through image processing with deep learning models.

Genomic sequences are traditionally represented as strings of characters: A (adenine), C (cytosine), G (guanine), and T (thymine). However, an alternative approach involves depicting sequence-related information through image representations, such as Chaos Game Representation (CGR) and read pileup images. With rapid advancements in deep learning (DL) methods within computer vision and natural language processing, there is growing interest in applying image-based DL methods to genomic sequence analysis. These methods involve encoding genomic information as images or integrating spatial information from images into the analytical process. In this review, we summarize three typical applications that use image processing with DL models for genome analysis. We examine the utilization and advantages of these image-based approaches.

Immune-related adverse events and their effects on survival outcomes in patients with non-small cell lung cancer treated with immune checkpoint inhibitors: a systematic review and meta-analysis.

Background: The use of immune checkpoint inhibitors (ICIs) has become the standard of care for non-small cell lung cancer. The purpose of this study was to systematically review the literature to determine whether the occurrence of immune-related adverse events (irAEs) following the use of ICIs predicts different clinical outcomes in non-small cell lung cancer (NSCLC). Methods: Relevant studies from the time of database creation to July 20, 2023, were systematically searched to explore the differences in clinical outcomes in patients with advanced NSCLC with or without irAEs. The outcome indicators included the occurrence of irAEs, progression-free survival (PFS), and overall survival (OS). Results: 25 studies met the inclusion criteria. Of these studies, 22 reported the effect on OS, and 19 reported the effect on PFS. The results showed that for patients with NSCLC, the occurrence of irAEs after receiving immunotherapy showed a statistically significant benefit over the absence of irAEs for OS (HR=0.55,95% CI=0.46-0.65) and PFS (HR=0.55 95% CI=0.48-0.64), but severe irAEs (grades 3-5) were associated with worse OS (HR=1.05, 95% CI=0.87-1.27). Compared with gastrointestinal, lung, and hepatitis, irAEs of the skin and endocrine system tend to predict better OS and PFS. Conclusion: The occurrence of irAEs, especially mild and early irAEs, indicates better OS and PFS in patients with NSCLC treated with ICIs, irrespective of patient characteristics, type of ICIs, and irAEs. However, Grade 3 or higher toxicities resulted in worse OS. Systematic review registration: https://www.crd.york.ac.uk/prospero/, identifier CRD42023409444.

biotextgraph: graphical summarization of functional similarities from textual information.

SUMMARY: Functional interpretation of biological entities such as differentially expressed genes is one of the fundamental analyses in bioinformatics. The task can be addressed by using biological pathway databases with enrichment analysis (EA). However, textual description of biological entities in public databases is less explored and integrated in existing tools and it has a potential to reveal new mechanisms. Here, we present a new R package biotextgraph for graphical summarization of omics' textual description data which enables assessment of functional similarities of the lists of biological entities. We illustrate application examples of annotating gene identifiers in addition to EA. The results suggest that the visualization based on words and inspection of biological entities with text can reveal a set of biologically meaningful terms that could not be obtained by using biological pathway databases alone. The results suggest the usefulness of the package in the routine analysis of omics-related data. The package also offers a web-based application for convenient querying. AVAILABILITY AND IMPLEMENTATION: The package, documentation, and web server are available at: https://github.com/noriakis/biotextgraph.

Exploration of the Existence Forms and Patterns of Dissolved Oxygen Molecules in Water.

The structure of liquid water is primarily composed of three-dimensional networks of water clusters formed by hydrogen bonds, and dissolved oxygen is one of the most important indicators for assessing water quality. In this work, distilled water with different concentration of dissolved oxygen were prepared, and a clear negative correlation between the size of water clusters and dissolved oxygen concentration was observed. Besides, a phenomenon of rapid absorption and release of oxygen at the water interfaces was unveiled, suggesting that oxygen molecules predominantly exist at the interfaces of water clusters. Oxygen molecules can move rapidly through the interfaces among water clusters, allowing dissolved oxygen to quickly reach a saturation level at certain partial pressure of oxygen and temperature. Further exploration into the mechanism by molecular dynamics simulations of oxygen and water clusters found that oxygen molecules can only exist stably at the interfaces among water clusters. A semi-empirical formula relating the average number of water molecules in a cluster (n) to 17O NMR half-peak width (W) was summarized: n = 0.1 W + 0.85. These findings provide a foundation for exploring the structure and properties of water.

Exploring the immune landscape and drug prediction of an M2 tumor-associated macrophage-related gene signature in EGFR-negative lung adenocarcinoma.

BACKGROUND: Improving immunotherapy efficacy for EGFR-negative lung adenocarcinoma (LUAD) patients remains a critical challenge, and the therapeutic effect of immunotherapy is largely determined by the tumor microenvironment (TME). Tumor-associated macrophages (TAMs) are the top-ranked immune infiltrating cells in the TME, and M2-TAMs exert potent roles in tumor promotion and chemotherapy resistance. An M2-TAM-based prognostic signature was constructed by integrative analysis of single-cell RNA-seq (scRNA-seq) and bulk RNA-seq data to reveal the immune landscape and select drugs in EGFR-negative LUAD. METHODS: M2-TAM-based biomarkers were obtained from the intersection of bulk RNA-seq data and scRNA-seq data. After consensus clustering of EGFR-negative LUAD into different clusters based on M2-TAM-based genes, we compared the prognosis, clinical features, estimate scores, immune infiltration, and checkpoint genes among the clusters. Next, we combined univariate Cox and LASSO regression analyses to establish an M2-TAM-based prognostic signature. RESULTS: CCL20, HLA-DMA, HLA-DRB5, KLF4, and TMSB4X were verified as prognostic M2-like TAM-related genes by univariate Cox and LASSO regression analyses. IPS and TMB analyses revealed that the high-risk group responded better to common immunotherapy. CONCLUSION: The study shows the potential of the M2-like TAM-related gene signature in EGFR-negative LUAD, explores the immune landscape based on M2-like TAM-related genes, and predict immunotherapy response of patients with EGFR-negative LUAD, providing a new insight for individualized treatment.

Diamond-Like Carbon Depositing on the Surface of Polylactide Membrane for Prevention of Adhesion Formation During Tendon Repair.

Post-traumatic peritendinous adhesion presents a significant challenge in clinical medicine. This study proposes the use of diamond-like carbon (DLC) deposited on polylactic acid (PLA) membranes as a biophysical mechanism for anti-adhesion barrier to encase ruptured tendons in tendon-injured rats. The results indicate that PLA/DLC composite membrane exhibits more efficient anti-adhesion effect than PLA membrane, with histological score decreasing from 3.12 ± 0.27 to 2.20 ± 0.22 and anti-adhesion effectiveness increasing from 21.61% to 44.72%. Mechanistically, the abundant C=O bond functional groups on the surface of DLC can reduce reactive oxygen species level effectively; thus, the phosphorylation of NF-κB and M1 polarization of macrophages are inhibited. Consequently, excessive inflammatory response augmented by M1 macrophage-originated cytokines including interleukin-6 (IL-6), interleukin-1ß (IL-1ß), and tumor necrosis factor-α (TNF-α) is largely reduced. For biocompatibility evaluation, PLA/DLC membrane is slowly absorbed within tissue and displays prolonged barrier effects compared to traditional PLA membranes. Further studies show the DLC depositing decelerates the release of degradation product lactic acid and its induction of macrophage M2 polarization by interfering esterase and PLA ester bonds, which further delays the fibrosis process. It was found that the PLA/DLC membrane possess an efficient biophysical mechanism for treatment of peritendinous adhesion.

CFViSA: A comprehensive and free platform for visualization and statistics in omics-data.

INTRODUCTION: The rapid growth of omics technologies has led to the use of bioinformatics as a powerful tool for unravelling scientific puzzles. However, the obstacles of bioinformatics are compounded by the complexity of data processing and the distinct nature of omics data types, particularly in terms of visualization and statistics. OBJECTIVES: We developed a comprehensive and free platform, CFViSA, to facilitate effortless visualization and statistical analysis of omics data by the scientific community. METHODS: CFViSA was constructed using the Scala programming language and utilizes the AKKA toolkit for the web server and MySQL for the database server. The visualization and statistical analysis were performed with the R program. RESULTS: CFViSA integrates two omics data analysis pipelines (microbiome and transcriptome analysis) and an extensive array of 79 analysis tools spanning simple sequence processing, visualization, and statistics available for various omics data, including microbiome and transcriptome data. CFViSA starts from an analysis interface, paralleling a demonstration full course to help users understand operating principles and scientifically set the analysis parameters. Once analysis is conducted, users can enter the task history interface for figure adjustments, and then a complete series of results, including statistics, feature tables and figures. All the graphic layouts were printed with necessary statistics and a traceback function recording the options for analysis and visualization; these statistics were excluded from the five competing methods. CONCLUSION: CFViSA is a user-friendly bioinformatics cloud platform with detailed guidelines for integrating functions in multi-omics analysis with real-time visualization adjustment and complete series of results provision. CFViSA is available at http://www.cloud.biomicroclass.com/en/CFViSA/.

Employing Atomic Force Microscopy (AFM) for Microscale Investigation of Interfaces and Interactions in Membrane Fouling Processes: New Perspectives and Prospects.

Membrane fouling presents a significant challenge in the treatment of wastewater. Several detection methods have been used to interpret membrane fouling processes. Compared with other analysis and detection methods, atomic force microscopy (AFM) is widely used because of its advantages in liquid-phase in situ 3D imaging, ability to measure interactive forces, and mild testing conditions. Although AFM has been widely used in the study of membrane fouling, the current literature has not fully explored its potential. This review aims to uncover and provide a new perspective on the application of AFM technology in future studies on membrane fouling. Initially, a rigorous review was conducted on the morphology, roughness, and interaction forces of AFM in situ characterization of membranes and foulants. Then, the application of AFM in the process of changing membrane fouling factors was reviewed based on its in situ measurement capability, and it was found that changes in ionic conditions, pH, voltage, and even time can cause changes in membrane fouling morphology and forces. Existing membrane fouling models are then discussed, and the role of AFM in predicting and testing these models is presented. Finally, the potential of the improved AFM techniques to be applied in the field of membrane fouling has been underestimated. In this paper, we have fully elucidated the potentials of the improved AFM techniques to be applied in the process of membrane fouling, and we have presented the current challenges and the directions for the future development in an attempt to provide new insights into this field.

Predicting cell types with supervised contrastive learning on cells and their types.

Single-cell RNA-sequencing (scRNA-seq) is a powerful technique that provides high-resolution expression profiling of individual cells. It significantly advances our understanding of cellular diversity and function. Despite its potential, the analysis of scRNA-seq data poses considerable challenges related to multicollinearity, data imbalance, and batch effect. One of the pivotal tasks in single-cell data analysis is cell type annotation, which classifies cells into discrete types based on their gene expression profiles. In this work, we propose a novel modeling formalism for cell type annotation with a supervised contrastive learning method, named SCLSC (Supervised Contrastive Learning for Single Cell). Different from the previous usage of contrastive learning in single cell data analysis, we employed the contrastive learning for instance-type pairs instead of instance-instance pairs. More specifically, in the cell type annotation task, the contrastive learning is applied to learn cell and cell type representation that render cells of the same type to be clustered in the new embedding space. Through this approach, the knowledge derived from annotated cells is transferred to the feature representation for scRNA-seq data. The whole training process becomes more efficient when conducting contrastive learning for cell and their types. Our experiment results demonstrate that the proposed SCLSC method consistently achieves superior accuracy in predicting cell types compared to five state-of-the-art methods. SCLSC also performs well in identifying cell types in different batch groups. The simplicity of our method allows for scalability, making it suitable for analyzing datasets with a large number of cells. In a real-world application of SCLSC to monitor the dynamics of immune cell subpopulations over time, SCLSC demonstrates a capability to discriminate cell subtypes of CD19+ B cells that were not present in the training dataset.

An update of clinical value of circulating tumor DNA in esophageal cancer: a systematic review and meta-analysis.

BACKGROUND: Esophageal cancer (EC) is a deadly disease with limited therapeutic options. Although circulating tumor DNA (ctDNA) could be a promising tool in this regard, the availiable evidence is limited. We performed a systematic review and meta-analysis to summarize the clinical applicability of the next-generation sequencing (NGS) and droplet digital polymerase chain reaction (ddPCR) technology on the ctDNA detection of the EC and listed the current challenges. METHODS: We systematically searched MEDLINE (via PubMed), Embase (via OVID), ISI Web of Science database and Cochrane Library from January, 2000 to April, 2023. Progression-free survival (PFS) and overall survival (OS) were set as primary outcome endpoints. Pathologic response was evaluated by tumor regression grade (TRG), according to the eighth edition of the American Joint Committee on Cancer (AJCC). Major pathologic regression (MPR) was defined as TRG 1 and 2. The MPR was set as secondary endpoint. Hazard rate (HR) and associated 95% CI were used as the effect indicators the association between ctDNA and prognosis of EC. MPR rates were also calculated. Fixed-effect model (Inverse Variance) or random-effect model (Mantel-Haenszel method) was performed depending on the statistically heterogeneity. RESULTS: Twenty-two studies, containing 1144 patients with EC, were included in this meta-analysis. The results showed that OS (HR = 3.87; 95% CI, 2.86-5.23) and PFS (HR = 4.28; 95% CI, 3.34-5.48) were shorter in ctDNA-positive patients. In the neoadjuvant therapy, the sensitivity analysis showed the clarified HR of ctDNA-positive was 1.13(95% CI, 1.01-1.28). We also found that TP53, NOTCH1, CCND1 and CNKN2A are the most frequent mutation genes. CONCLUSIONS: Positive ctDNA is associated with poor prognosis, which demonstrated clinical value of ctDNA. Longitudinal ctDNA monitoring showed potential prognostic value in the neoadjuvant therapy. In an era of precision medicine, ctDNA could be a promising tool to individualize treatment planning and to improve outcomes in EC. PROSPERO REGISTRATION NUMBER: CRD42023412465.

Correction: Microbial Gene Ontology informed deep neural network for microbe functionality discovery in human diseases.

[This corrects the article DOI: 10.1371/journal.pone.0290307.].

Investigation of the BERT model on nucleotide sequences with non-standard pre-training and evaluation of different k-mer embeddings.

MOTIVATION: In recent years, pre-training with the transformer architecture has gained significant attention. While this approach has led to notable performance improvements across a variety of downstream tasks, the underlying mechanisms by which pre-training models influence these tasks, particularly in the context of biological data, are not yet fully elucidated. RESULTS: In this study, focusing on the pre-training on nucleotide sequences, we decompose a pre-training model of Bidirectional Encoder Representations from Transformers (BERT) into its embedding and encoding modules to analyze what a pre-trained model learns from nucleotide sequences. Through a comparative study of non-standard pre-training at both the data and model levels, we find that a typical BERT model learns to capture overlapping-consistent k-mer embeddings for its token representation within its embedding module. Interestingly, using the k-mer embeddings pre-trained on random data can yield similar performance in downstream tasks, when compared with those using the k-mer embeddings pre-trained on real biological sequences. We further compare the learned k-mer embeddings with other established k-mer representations in downstream tasks of sequence-based functional prediction. Our experimental results demonstrate that the dense representation of k-mers learned from pre-training can be used as a viable alternative to one-hot encoding for representing nucleotide sequences. Furthermore, integrating the pre-trained k-mer embeddings with simpler models can achieve competitive performance in two typical downstream tasks. AVAILABILITY AND IMPLEMENTATION: The source code and associated data can be accessed at https://github.com/yaozhong/bert_investigation.

Interfacial Polymerization Depth Mediated by the Shuttle Effect Regulating the Application Performance of Pesticide-Loaded Microcapsules.

The highly water-soluble nematicide fosthiazate is anticipated to undergo microencapsulation in order to enhance its retention around plant roots and mitigate leaching into groundwater. However, the underlying mechanism governing the influence of hydrophilicity of the microcapsule (MC) core on the evolution of the microcapsule shell remains unclear, posing challenges for encapsulating water-soluble core materials. This study elucidates the microlevel formation mechanism of microcapsules by investigating the impact of interfacial mass transfer on shell formation and proposes a method for regulating the structure of shells. The study reveals that enhancing the hydrophilicity of the core enhances the shuttle effect between the oil and aqueous phase, expands the region of polymerization reactions, and forms a loose and thick shell. The thickness of the microcapsule shell prepared using solvent oil 150# (MCs-SOL) measures only 264 nm, while that of the microcapsules prepared using propylene glycol diacetate and solvent oil 150# at a ratio of 2:1 (MCs-P2S1) is 5.2 times greater. The enhanced compactness of the shell reduced the release rate of microcapsules and the leaching distance of fosthiazate in soil, thereby mitigating the risk of leaching loss and facilitating the distribution of active ingredients within crop roots. The MCs-SOL had a limited leaching distance measurement of 8 cm and exhibited a satisfactory efficacy of 87.3% in controlling root galling nematodes. The thickness and compactness of the MCs shell can be regulated by manipulating the interfacial shuttle effect, providing a promising approach to enhancing utilization efficiency while mitigating potential environmental risks.

Development and validation of a predictive model for the diagnosis of bladder tumors using narrow band imaging.

PURPOSE: At present, the prediction of bladder tumor nature during cystoscopy is partially dependent on the clinician's own experience. Subjective factors may lead to excessive biopsy or delayed treatment. The purpose of our study is to establish a reliable model for predicting the nature of bladder tumors using narrow band imaging. METHODS: From November 2021 to November 2022, the clinical data of 231 patients who required a cystoscopy were prospectively collected at our center. Cystoscopy was performed in 219 eligible patients, in which both tumor and vascular morphology characteristics were recorded. Pathological results were used as the diagnostic standard. A logistic regression analysis was used to screen out factors related to tumor pathology. Bootstrap resampling was used for internal validation. A total of 71 patients from four other centers served as an external validation cohort. RESULTS: The following diagnostic factors were identified: tumor morphology (cauliflower-like or algae-like lesions), vascular morphology (dotted or circumferential vessels), tumor boundary (clear or unclear), and patients' symptoms (gross hematuria) and were included in the prediction model. The internal validation results showed that the area under the curve was 0.94 (95% CI 0.92-0.97), and the P value from the goodness-of-fit test was 0.97. After external validation, the results showed the area under the curve was 0.89 (95% CI 0.82-0.97) and the P value of the goodness-of-fit test was 0.24. CONCLUSION: A diagnostic prediction nomogram was established for bladder cancer. The verification results showed that the prediction model has good prediction performance.

Microbial Gene Ontology informed deep neural network for microbe functionality discovery in human diseases.

The human microbiome plays a crucial role in human health and is associated with a number of human diseases. Determining microbiome functional roles in human diseases remains a biological challenge due to the high dimensionality of metagenome gene features. However, existing models were limited in providing biological interpretability, where the functional role of microbes in human diseases is unexplored. Here we propose to utilize a neural network-based model incorporating Gene Ontology (GO) relationship network to discover the microbe functionality in human diseases. We use four benchmark datasets, including diabetes, liver cirrhosis, inflammatory bowel disease, and colorectal cancer, to explore the microbe functionality in the human diseases. Our model discovered and visualized the novel candidates' important microbiome genes and their functions by calculating the important score of each gene and GO term in the network. Furthermore, we demonstrate that our model achieves a competitive performance in predicting the disease by comparison with other non-Gene Ontology informed models. The discovered candidates' important microbiome genes and their functions provide novel insights into microbe functional contribution.

Zero-shot-capable identification of phage-host relationships with whole-genome sequence representation by contrastive learning.

Accurately identifying phage-host relationships from their genome sequences is still challenging, especially for those phages and hosts with less homologous sequences. In this work, focusing on identifying the phage-host relationships at the species and genus level, we propose a contrastive learning based approach to learn whole-genome sequence embeddings that can take account of phage-host interactions (PHIs). Contrastive learning is used to make phages infecting the same hosts close to each other in the new representation space. Specifically, we rephrase whole-genome sequences with frequency chaos game representation (FCGR) and learn latent embeddings that 'encapsulate' phages and host relationships through contrastive learning. The contrastive learning method works well on the imbalanced dataset. Based on the learned embeddings, a proposed pipeline named CL4PHI can predict known hosts and unseen hosts in training. We compare our method with two recently proposed state-of-the-art learning-based methods on their benchmark datasets. The experiment results demonstrate that the proposed method using contrastive learning improves the prediction accuracy on known hosts and demonstrates a zero-shot prediction capability on unseen hosts. In terms of potential applications, the rapid pace of genome sequencing across different species has resulted in a vast amount of whole-genome sequencing data that require efficient computational methods for identifying phage-host interactions. The proposed approach is expected to address this need by efficiently processing whole-genome sequences of phages and prokaryotic hosts and capturing features related to phage-host relationships for genome sequence representation. This approach can be used to accelerate the discovery of phage-host interactions and aid in the development of phage-based therapies for infectious diseases.

A novel approach to quantify calcifications of thyroid nodules in US images based on deep learning: predicting the risk of cervical lymph node metastasis in papillary thyroid cancer patients.

OBJECTIVE: Based on ultrasound (US) images, this study aimed to detect and quantify calcifications of thyroid nodules, which are regarded as one of the most important features in US diagnosis of thyroid cancer, and to further investigate the value of US calcifications in predicting the risk of lymph node metastasis (LNM) in papillary thyroid cancer (PTC). METHODS: Based on the DeepLabv3+ networks, 2992 thyroid nodules in US images were used to train a model to detect thyroid nodules, of which 998 were used to train a model to detect and quantify calcifications. A total of 225 and 146 thyroid nodules obtained from two centers, respectively, were used to test the performance of these models. A logistic regression method was used to construct the predictive models for LNM in PTCs. RESULTS: Calcifications detected by the network model and experienced radiologists had an agreement degree of above 90%. The novel quantitative parameters of US calcification defined in this study showed a significant difference between PTC patients with and without cervical LNM (p < 0.05). The calcification parameters were beneficial to predicting the LNM risk in PTC patients. The LNM prediction model using these calcification parameters combined with patient age and other US nodular features showed a higher specificity and accuracy than the calcification parameters alone. CONCLUSIONS: Our models not only detect the calcifications automatically, but also have value in predicting cervical LNM risk of PTC patients, thereby making it possible to investigate the relationship between calcifications and highly invasive PTC in detail. CLINICAL RELEVANCE STATEMENT: Due to the high association of US microcalcifications with thyroid cancers, our model will contribute to the differential diagnosis of thyroid nodules in daily practice. KEY POINTS: • We developed an ML-based network model for automatically detecting and quantifying calcifications within thyroid nodules in US images. • Three novel parameters for quantifying US calcifications were defined and verified. • These US calcification parameters showed value in predicting the risk of cervical LNM in PTC patients.