Marine-derived bioactive materials as antibiofilm and antivirulence agents.

Microbial infections are major human health issues, and, recently, the mortality rate owing to bacterial and fungal infections has been increasing. In addition to intrinsic and extrinsic antimicrobial resistance mechanisms, biofilm formation is a key adaptive resistance mechanism. Several bioactive compounds from marine organisms have been identified for use in biofilm therapy owing to their structural complexity, biocompatibility, and economic viability. In this review, we discuss recent trends in the application of marine natural compounds, marine-bioinspired nanomaterials, and marine polymer conjugates as possible therapeutic agents for controlling biofilms and virulence factors. We also comprehensively discuss the mechanisms underlying biofilm formation and inhibition of virulence factors by marine-derived materials and propose possible applications of novel and effective antibiofilm and antivirulence agents.

Updated Primer on Generative Artificial Intelligence and Large Language Models in Medical Imaging for Medical Professionals.

The emergence of Chat Generative Pre-trained Transformer (ChatGPT), a chatbot developed by OpenAI, has garnered interest in the application of generative artificial intelligence (AI) models in the medical field. This review summarizes different generative AI models and their potential applications in the field of medicine and explores the evolving landscape of Generative Adversarial Networks and diffusion models since the introduction of generative AI models. These models have made valuable contributions to the field of radiology. Furthermore, this review also explores the significance of synthetic data in addressing privacy concerns and augmenting data diversity and quality within the medical domain, in addition to emphasizing the role of inversion in the investigation of generative models and outlining an approach to replicate this process. We provide an overview of Large Language Models, such as GPTs and bidirectional encoder representations (BERTs), that focus on prominent representatives and discuss recent initiatives involving language-vision models in radiology, including innovative large language and vision assistant for biomedicine (LLaVa-Med), to illustrate their practical application. This comprehensive review offers insights into the wide-ranging applications of generative AI models in clinical research and emphasizes their transformative potential.

Aerobic and anoxic utilization of organic matter for flexible nitrite supply in nutrient conversion pathways based on anaerobic ammonium oxidation: Microbial interactive mechanism.

This study investigated nutrient conversion pathways and corresponding interactive mechanisms in a mainstream partial-nitritation (PN)/anaerobic ammonium oxidation (anammox)/partial-denitrification-(PD)-enhanced biological phosphorus-removal (EBPR) (PN/A/PD-EBPR) process. A laboratory-scale sequencing batch reactor was operated for 301 days under different operational strategies. Mainstream PN/A/PD-EBPR was successfully operated with aerobic and anoxic utilization of organic matter. Aerobic utilization of organic matter was an effective strategy for conversion to denitrifying polyphosphate-accumulating organism-based phosphorus removal, referring to a biological reaction that outperformed nitrite-oxidizing bacteria. Aerobically adsorbed organic matter could be used as a carbon source for PD, which further enhanced nitrogen removal by PN/A. Ultimately, the interaction between complex nutrient conversion pathways served to achieve stable performance. High-throughput sequencing results elucidated the core microbe functioning in the mainstream PN/A/PD-EBPR process with respect to various nutrients. The outcomes of this study will be beneficial to those attempting to implement mainstream PN/A/PD-EBPR.

Bacterial extracellular vesicles: Modulation of biofilm and virulence properties.

Microbial pathogens cause persistent infections by forming biofilms and producing numerous virulence factors. Bacterial extracellular vesicles (BEVs) are nanostructures produced by various bacterial species vital for molecular transport. BEVs include various components, including lipids (glycolipids, LPS, and phospholipids), nucleic acids (genomic DNA, plasmids, and short RNA), proteins (membrane proteins, enzymes, and toxins), and quorum-sensing signaling molecules. BEVs play a major role in forming extracellular polymeric substances (EPS) in biofilms by transporting EPS components such as extracellular polysaccharides, proteins, and extracellular DNA. BEVs have been observed to carry various secretory virulence factors. Thus, BEVs play critical roles in cell-to-cell communication, biofilm formation, virulence, disease progression, and resistance to antimicrobial treatment. In contrast, BEVs have been shown to impede early-stage biofilm formation, disseminate mature biofilms, and reduce virulence. This review summarizes the current status in the literature regarding the composition and role of BEVs in microbial infections. Furthermore, the dual functions of BEVs in eliciting and suppressing biofilm formation and virulence in various microbial pathogens are thoroughly discussed. This review is expected to improve our understanding of the use of BEVs in determining the mechanism of biofilm development in pathogenic bacteria and in developing drugs to inhibit biofilm formation by microbial pathogens. STATEMENT OF SIGNIFICANCE: Bacterial extracellular vesicles (BEVs) are nanostructures formed by membrane blebbing and explosive cell lysis. It is essential for transporting lipids, nucleic acids, proteins, and quorum-sensing signaling molecules. BEVs play an important role in the formation of the biofilm's extracellular polymeric substances (EPS) by transporting its components, such as extracellular polysaccharides, proteins, and extracellular DNA. Furthermore, BEVs shield genetic material from nucleases and thermodegradation by packaging it during horizontal gene transfer, contributing to the transmission of bacterial adaptation determinants like antibiotic resistance. Thus, BEVs play a critical role in cell-to-cell communication, biofilm formation, virulence enhancement, disease progression, and drug resistance. In contrast, BEVs have been shown to prevent early-stage biofilm, disperse mature biofilm, and reduce virulence characteristics.

Optimization of Protease Treatment Conditions for Chlorella pyrenoidosa Protein Extraction and Investigation of Its Potential as an Alternative Protein Source.

This study aimed to determine enzymes that effectively extract Chlorella pyrenoidosa proteins and optimize the processing conditions using response surface methods. Furthermore, the potential of enzymatically hydrolyzed C. pyrenoidosa protein extract (CPE) as a substitute protein source was investigated. The enzymatic hydrolysis conditions for protein extraction were optimized using single-factor analysis and a response surface methodology-Box-Behnken design. The R2 value of the optimized model was 0.9270, indicating the reliability of the model, and the optimal conditions were as follows: a hydrolysis temperature of 45.56 °C, pH 9.1, and a hydrolysis time of 49.85 min. The amino acid composition of CPE was compared to that of C. pyrenoidosa powder (CP), which was found to have a higher content of essential amino acids (EAA). The electrophoretic profiles of CP and CPE confirmed that CPE has a low molecular weight. Furthermore, CPE showed higher antioxidant activity and phenol content than CP, with ABTS and DPPH radical scavenging abilities of 69.40 ± 1.61% and 19.27 ± 3.16%, respectively. CPE had high EAA content, antioxidant activity, and phenol content, indicating its potential as an alternative protein source. Overall, in this study, we developed an innovative, ecofriendly, and gentle enzymatic hydrolysis strategy for the extraction and refinement of Chlorella proteins.

Approximating Intermediate Feature Maps of Self-Supervised Convolution Neural Network to Learn Hard Positive Representations in Chest Radiography.

Recent advances in contrastive learning have significantly improved the performance of deep learning models. In contrastive learning of medical images, dealing with positive representation is sometimes difficult because some strong augmentation techniques can disrupt contrastive learning owing to the subtle differences between other standardized CXRs compared to augmented positive pairs; therefore, additional efforts are required. In this study, we propose intermediate feature approximation (IFA) loss, which improves the performance of contrastive convolutional neural networks by focusing more on positive representations of CXRs without additional augmentations. The IFA loss encourages the feature maps of a query image and its positive pair to resemble each other by maximizing the cosine similarity between the intermediate feature outputs of the original data and the positive pairs. Therefore, we used the InfoNCE loss, which is commonly used loss to address negative representations, and the IFA loss, which addresses positive representations, together to improve the contrastive network. We evaluated the performance of the network using various downstream tasks, including classification, object detection, and a generative adversarial network (GAN) inversion task. The downstream task results demonstrated that IFA loss can improve the performance of effectively overcoming data imbalance and data scarcity; furthermore, it can serve as a perceptual loss encoder for GAN inversion. In addition, we have made our model publicly available to facilitate access and encourage further research and collaboration in the field.

Poly(vinyl alcohol)/chitosan hydrogel incorporating chitooligosaccharide-gentisic acid conjugate with antioxidant and antibacterial properties as a potential wound dressing.

The design and development of wound dressing with antioxidant and antibacterial properties to accelerate wound healing remain challenging. In this study, we synthesize a chitooligosaccharide-gentisic acid (COS-GSA) conjugate using the free-radical grafting method, and fabricate a poly(vinyl alcohol) (PVA)/chitosan (CH)/COS-GSA (PVA/CH/CG) hydrogel using a freeze-thaw method. We characterize the synthesized COS-GSA conjugates using through polyphenol assay, absorbance, and 1H NMR spectroscopy and evaluate their antioxidant properties. The COS-GSA conjugates are successfully synthesized and exhibit better antioxidant properties than pristine COSs. Subsequently, the fabricated hydrogel is characterized based on its morphological analysis, rheological properties, water contact angle, swelling, degradation, water retention properties, and COS-GSA release profiles. Finally, the biocompatibility of the fabricated hydrogel is evaluated on HDF and HaCaT cells through indirect and direct cytotoxicity. The PVA/CH/CG hydrogel exhibited significantly higher antioxidant properties (DPPH, 2,2'-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), and hydrogen peroxide (H2O2) scavenging activities) and antibacterial activities (Staphylococcus aureus and Pseudomonas aeruginosa) compared to other fabricated hydrogels such as PVA, PVA/CH, and PVA/CH/COS (PVA/CH/C). These results provide evidence that PVA/CH/CG hydrogels with antioxidant, antibacterial, and non-cytotoxic properties have great potential for wound-dressing applications.

Synthesis and evaluation of tirbanibulin derivatives: a detailed exploration of the structure-activity relationship for anticancer activity.

Tirbanibulin, an FDA-approved microtubule-targeting agent (MTA) introduced in 2020, represents a pioneering treatment for precancerous actinic keratosis. Despite its failure to gain approval as an anticancer agent due to insufficient efficacy, there remains potential value in extending its application into malignancy treatment through tirbanibulin-based derivatives. Tirbanibulin possesses a distinctive dual mechanism of action involving microtubule and Src inhibition, distinguishing it from other MTAs. In spite of its unique profile, exploration of tirbanibulin's structure-activity relationship (SAR) and the development of its derivatives are significantly limited in the current literature. This study addresses this gap by synthesizing various tirbanibulin derivatives and exploring their SAR through modifications in the core amide motif and the eastern benzylamine part. Our results underscore the critical role of the pyridinyl acetamide core structure for optimal cellular potency, with favorable tolerance observed for modifications at the para position of the benzylamine moiety. Particularly noteworthy is the analogue modified with p-fluorine benzylamine, which exhibited favorable in vivo PK profiles. These findings provide crucial insights into the potential advancement of tirbanibulin-based compounds as promising anticancer agents.

Deep Learning for Automated Triaging of Stable Chest Radiographs in a Follow-up Setting.

Background Most artificial intelligence algorithms that interpret chest radiographs are restricted to an image from a single time point. However, in clinical practice, multiple radiographs are used for longitudinal follow-up, especially in intensive care units (ICUs). Purpose To develop and validate a deep learning algorithm using thoracic cage registration and subtraction to triage pairs of chest radiographs showing no change by using longitudinal follow-up data. Materials and Methods A deep learning algorithm was retrospectively developed using baseline and follow-up chest radiographs in adults from January 2011 to December 2018 at a tertiary referral hospital. Two thoracic radiologists reviewed randomly selected pairs of "change" and "no change" images to establish the ground truth, including normal or abnormal status. Algorithm performance was evaluated using area under the receiver operating characteristic curve (AUC) analysis in a validation set and temporally separated internal test sets (January 2019 to August 2021) from the emergency department (ED) and ICU. Threshold calibration for the test sets was conducted, and performance with 40% and 60% triage thresholds was assessed. Results This study included 3 304 996 chest radiographs in 329 036 patients (mean age, 59 years ± 14 [SD]; 170 433 male patients). The training set included 550 779 pairs of radiographs. The validation set included 1620 pairs (810 no change, 810 change). The test sets included 533 pairs (ED; 265 no change, 268 change) and 600 pairs (ICU; 310 no change, 290 change). The algorithm had AUCs of 0.77 (validation), 0.80 (ED), and 0.80 (ICU). With a 40% triage threshold, specificity was 88.4% (237 of 268 pairs) and 90.0% (261 of 290 pairs) in the ED and ICU, respectively. With a 60% triage threshold, specificity was 79.9% (214 of 268 pairs) and 79.3% (230 of 290 pairs) in the ED and ICU, respectively. For urgent findings (consolidation, pleural effusion, pneumothorax), specificity was 78.6%-100% (ED) and 85.5%-93.9% (ICU) with a 40% triage threshold. Conclusion The deep learning algorithm could triage pairs of chest radiographs showing no change while detecting urgent interval changes during longitudinal follow-up. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Czum in this issue.

Improved regression model for anaerobic ammonium oxidation by repeated and prolonged batch assay under stressful salinity and pH conditions.

The aim of this study was to propose repeated and prolonged batch (RPB) assay as a promising specific anammox activity (SAA) methodology assessing the anammox activity under stressed salinity and pH conditions. Response surface analysis (RSA) was used as a regression tool to evaluate statistical significance. The feasibility of RPB was investigated at 0 to 15 g-NaCl/L of salinity and pH 6 to 8 with reflecting the results of preliminary SAA. As a result, conventional SAA was statistically insignificant. In addition, the RSA results obtained from repeated batch did not meet the statistical significance despite ten times iterative reaction. Interestingly, the RPB assay (i.e., applied both repeated and prolonged reaction) was effective to obtain the reliable results. Candidadus Brocadia and Candidadus Jettenia were functional anammox microbiome during RPB. Outcomes of this study suggest that RPB assay can be applied to accurately determine the anammox activity under various stressful conditions.

Overcoming the Challenges in the Development and Implementation of Artificial Intelligence in Radiology: A Comprehensive Review of Solutions Beyond Supervised Learning.

Artificial intelligence (AI) in radiology is a rapidly developing field with several prospective clinical studies demonstrating its benefits in clinical practice. In 2022, the Korean Society of Radiology held a forum to discuss the challenges and drawbacks in AI development and implementation. Various barriers hinder the successful application and widespread adoption of AI in radiology, such as limited annotated data, data privacy and security, data heterogeneity, imbalanced data, model interpretability, overfitting, and integration with clinical workflows. In this review, some of the various possible solutions to these challenges are presented and discussed; these include training with longitudinal and multimodal datasets, dense training with multitask learning and multimodal learning, self-supervised contrastive learning, various image modifications and syntheses using generative models, explainable AI, causal learning, federated learning with large data models, and digital twins.

Controlling biofilm and virulence properties of Gram-positive bacteria by targeting wall teichoic acid and lipoteichoic acid.

Wall teichoic acid (WTA) and lipoteichoic acid (LTA) are structural components of Gram-positive bacteria's peptidoglycan and cell membrane, which are mostly anionic glycopolymers. WTA confers numerous physiological, virulence, and pathogenic features to bacterial pathogens. It controls cell shape, cell division, and the localisation of autolytic enzymes and ion homeostasis. In the context of virulence and pathogenicity, it aids bacterial cell attachment and colonisation and protects against the host defence system and antibiotics. Having such a broad function in pathogenic bacteria's lifecycle, WTA/LTA become one of the potential targets for antibacterial agents to reduce bacterial infection in the host. The number of reports for targeting the WTA/LTA pathway has risen, mostly by focusing on three distinct targets: antivirulence targets, ß-lactam potentiator targets, and essential targets. The current review looked at the role of WTA/LTA in biofilm development and virulence in a range of Gram-positive pathogenic bacteria. Furthermore, alternate strategies, such as the application of natural and synthetic compounds that target the WTA/LTA pathway, have been thoroughly discussed. Moreover, the application of nanomaterials and a combination of drugs have also been discussed as a viable method for targeting the WTA/LTA in numerous Gram-positive bacteria. In addition, a future perspective for controlling bacterial infection by targeting the WTA/LTA is proposed.

MuSiC-ViT: A multi-task Siamese convolutional vision transformer for differentiating change from no-change in follow-up chest radiographs.

A major responsibility of radiologists in routine clinical practice is to read follow-up chest radiographs (CXRs) to identify changes in a patient's condition. Diagnosing meaningful changes in follow-up CXRs is challenging because radiologists must differentiate disease changes from natural or benign variations. Here, we suggest using a multi-task Siamese convolutional vision transformer (MuSiC-ViT) with an anatomy-matching module (AMM) to mimic the radiologist's cognitive process for differentiating baseline change from no-change. MuSiC-ViT uses the convolutional neural networks (CNNs) meet vision transformers model that combines CNN and transformer architecture. It has three major components: a Siamese network architecture, an AMM, and multi-task learning. Because the input is a pair of CXRs, a Siamese network was adopted for the encoder. The AMM is an attention module that focuses on related regions in the CXR pairs. To mimic a radiologist's cognitive process, MuSiC-ViT was trained using multi-task learning, normal/abnormal and change/no-change classification, and anatomy-matching. Among 406 K CXRs studied, 88 K change and 115 K no-change pairs were acquired for the training dataset. The internal validation dataset consisted of 1,620 pairs. To demonstrate the robustness of MuSiC-ViT, we verified the results with two other validation datasets. MuSiC-ViT respectively achieved accuracies and area under the receiver operating characteristic curves of 0.728 and 0.797 on the internal validation dataset, 0.614 and 0.784 on the first external validation dataset, and 0.745 and 0.858 on a second temporally separated validation dataset. All code is available at https://github.com/chokyungjin/MuSiC-ViT.

Mesoporous Polydopamine-Encapsulated Fluorescent Nanodiamonds: A Versatile Platform for Biomedical Applications.

Fluorescent nanodiamonds (FNDs) are versatile nanomaterials with promising properties. However, efficient functionalization of FNDs for biomedical applications remains challenging. In this study, we demonstrate mesoporous polydopamine (mPDA) encapsulation of FNDs. The mPDA shell is generated by sequential formation of micelles via self-assembly of Pluronic F127 (F127) with 1,3,5-trimethyl benzene (TMB) and composite micelles via oxidation and self-polymerization of dopamine hydrochloride (DA). The surface of the mPDA shell can be readily functionalized with thiol-terminated methoxy polyethylene glycol (mPEG-SH), hyperbranched polyglycerol (HPG), and d-α-tocopheryl polyethylene glycol 1000 succinate (TPGS). The PEGylated FND@mPDA particles are efficiently taken up by, and employed as a fluorescent imaging probe for, HeLa cells. HPG-functionalized FND@mPDA is conjugated with an amino-terminated oligonucleotide to detect microRNA via hybridization. Finally, the increased surface area of the mPDA shell permits efficient loading of doxorubicin hydrochloride. Further modification with TPGS increases drug delivery efficiency, resulting in high toxicity to cancer cells.

Inhibition of Biofilm and Virulence Properties of Pathogenic Bacteria by Silver and Gold Nanoparticles Synthesized from Lactiplantibacillus sp. Strain C1.

The emergence of antibiotic resistance in microbial pathogens necessitates the development of alternative ways to combat the infections that arise. The current study used nanotechnology as an alternate technique to control virulence characteristics and biofilm development in Pseudomonas aeruginosa and Staphylococcus aureus. Furthermore, based on the acceptance and biocompatibility of the probiotic bacteria, we chose a lactic acid bacteria (LAB) for synthesizing two types of metallic nanoparticles (NPs) in this study. Using molecular techniques, the LAB strain C1 was isolated from Kimchi food samples and identified as Lactiplantibacillus sp. strain C1. The prepared supernatant from strain C1 was used to produce gold nanoparticles (AuNPs) and silver nanoparticles (AgNPs). C1-AuNPs and C1-AgNPs were characterized physiochemically using a variety of instruments. C1-AuNPs and C1-AgNPs had spherical shapes and sizes of 100.54 ± 14.07 nm (AuNPs) and 129.51 ± 12.31 nm (AgNPs), respectively. C1-AuNPs and C1-AgNPs were discovered to have high zeta potentials of -23.29 ± 1.17 and -30.57 ± 0.29 mV, respectively. These nanoparticles have antibacterial properties against several bacterial pathogens. C1-AuNPs and C1-AgNPs significantly inhibited the initial stage biofilm formation and effectively eradicated established mature biofilms of P. aeruginosa and S. aureus. Furthermore, when P. aeruginosa was treated with sub-MIC levels of C1-AuNPs and C1-AgNPs, their different virulence features were significantly reduced. Both NPs greatly inhibited the hemolytic activity of S. aureus. The inhibition of P. aeruginosa and S. aureus biofilms and virulence features by C1-AuNPs and C1-AgNPs can be regarded as viable therapeutic strategies for preventing infections caused by these bacteria.

Integrative Analyses Reveal the Anticancer Mechanisms and Sensitivity Markers of the Next-Generation Hypomethylating Agent NTX-301.

Epigenetic dysregulation characterized by aberrant DNA hypermethylation is a hallmark of cancer, and it can be targeted by hypomethylating agents (HMAs). Recently, we described the superior therapeutic efficacy of a novel HMA, namely, NTX-301, when used as a monotherapy and in combination with venetoclax in the treatment of acute myeloid leukemia. Following a previous study, we further explored the therapeutic properties of NTX-301 based on experimental investigations and integrative data analyses. Comprehensive sensitivity profiling revealed that NTX-301 primarily exerted anticancer effects against blood cancers and exhibited improved potency against a wide range of solid cancers. Subsequent assays showed that the superior efficacy of NTX-301 depended on its strong effects on cell cycle arrest, apoptosis, and differentiation. Due to its superior efficacy, low doses of NTX-301 achieved sufficiently substantial tumor regression in vivo. Multiomics analyses revealed the mechanisms of action (MoAs) of NTX-301 and linked these MoAs to markers of sensitivity to NTX-301 and to the demethylation activity of NTX-301 with high concordance. In conclusion, our findings provide a rationale for currently ongoing clinical trials of NTX-301 and will help guide the development of novel therapeutic options for cancer patients.

Elucidating prioritized factor for mainstream partial nitritation between C/N ratio and dissolved oxygen: Response surface methodology and microbial community shifts.

Recently, C/N ratio is suggested as a promising control factor with dissolved oxygen (DO) achieving mainstream partial nitritation (PN); however, their combined effects on mainstream PN are still limited. This study evaluated the mainstream PN with respect to the combined factors, and investigated the prioritized factor affecting the community of aerobic functional microbes competing with NOB. Response surface methodology was performed to assess the combined effects of C/N ratio and DO on the activity of functional microbes. Aerobic heterotrophic bacteria (AHB) played the greatest role in oxygen competition among functional microbes, which resulted in relative inhibition of nitrite-oxidizing bacteria (NOB). The combination of high C/N ratio and low DO had a positive role in the relative inhibition of NOB. In bioreactor operation, the PN was successfully achieved at ≥ 1.5 of C/N ratio for 0.5-2.0 mg/L DO conditions. Interestingly, aerobic functional microbes outcompeting NOB were shifted with C/N ratio rather than DO, suggesting C/N ratio is more prioritized factor achieving mainstream PN. These findings will provide insights into how combined aerobic conditions contribute to achieve mainstream PN.

Inhibition of Polymicrobial Biofilms of Candida albicans-Staphylococcus aureus/Streptococcus mutans by Fucoidan-Gold Nanoparticles.

The polymicrobial proliferation and development of complex biofilm morphologies by bacterial and fungal pathogens in the host are some of the key factors contributing to the failure of antimicrobial treatments. The polymicrobial interaction of Candida albicans and some bacterial species has been extensively studied in both in vitro and in vivo model systems. Alternative strategies for disrupting polymicrobial interaction and biofilm formation are constantly needed. Among several alternative strategies, the use of nanoparticles synthesized using a natural product in the treatment of microbial infection has been considered a promising approach. The current study aimed to synthesize gold nanoparticles (AuNPs) using a natural product, fucoidan, and to test their efficacy against mono and duo combinations of fungal (Candida albicans) and bacterial (Staphylococcus aureus/Streptococcus mutans) biofilms. Several methods were used to characterize and study Fu-AuNPs, including UV-vis absorption spectroscopy, FTIR, FE-TEM, EDS, DLS, zeta potential, and XRD. The concentration-dependent inhibition of early-stage biofilms and the eradication of mature biofilms of single species of C. albicans, S. aureus, and S. mutans have been observed. Early biofilms of a dual-species combination of C. albicans and S. aureus/S. mutans were also suppressed at an increasing concentration of Fu-AuNPs. Furthermore, Fu-AuNPs significantly eradicated the established mature biofilm of mixed species. The treatment method proposed in this study, which involves the use of marine-bioinspired nanoparticles, is a promising and biocompatible agent for preventing the growth of polymicrobial biofilms of bacterial and fungal pathogens.

CheSS: Chest X-Ray Pre-trained Model via Self-supervised Contrastive Learning.

Training deep learning models on medical images heavily depends on experts' expensive and laborious manual labels. In addition, these images, labels, and even models themselves are not widely publicly accessible and suffer from various kinds of bias and imbalances. In this paper, chest X-ray pre-trained model via self-supervised contrastive learning (CheSS) was proposed to learn models with various representations in chest radiographs (CXRs). Our contribution is a publicly accessible pretrained model trained with a 4.8-M CXR dataset using self-supervised learning with a contrastive learning and its validation with various kinds of downstream tasks including classification on the 6-class diseases in internal dataset, diseases classification in CheXpert, bone suppression, and nodule generation. When compared to a scratch model, on the 6-class classification test dataset, we achieved 28.5% increase in accuracy. On the CheXpert dataset, we achieved 1.3% increase in mean area under the receiver operating characteristic curve on the full dataset and 11.4% increase only using 1% data in stress test manner. On bone suppression with perceptual loss, we achieved improvement in peak signal to noise ratio from 34.99 to 37.77, structural similarity index measure from 0.976 to 0.977, and root-square-mean error from 4.410 to 3.301 when compared to ImageNet pretrained model. Finally, on nodule generation, we achieved improvement in Fréchet inception distance from 24.06 to 17.07. Our study showed the decent transferability of CheSS weights. CheSS weights can help researchers overcome data imbalance, data shortage, and inaccessibility of medical image datasets. CheSS weight is available at https://github.com/mi2rl/CheSS .

Denitrification dynamics in unsaturated soils with different porous structures and water saturation degrees: A focus on the shift in microbial community structures.

Despite its environmental significance, little is known about denitrification in vadose zones owing to the complexity of such environments. Here, we investigated denitrification in unsaturated soils with different pore distributions. To this end, we performed batch-type denitrification experiments and analyzed microbial community shifts before and after possible reactions with nitrates to clarify the relevant denitrifying mechanism in the microcosms. For quantitative comparison, pore distribution in the test soil samples was characterized based on the uniformity coefficient (Cu) and water saturation degree (SD). Micro-CT analysis of the soil pore distribution confirmed that the proportion of bigger-sized pores increased with decreasing Cu. However, oxygen diffusion into the system was controlled by SD rather than Cu. Within a certain SD range (51-67%), the pore condition changed abruptly from an oxic to an anoxic state. Consequently, denitrification occurred even under unsaturated soil conditions when the SD increased beyond 51-67%. High throughput sequencing revealed that the same microbial species were potentially responsible for denitrification under both partially (SD 67%), and fully saturated (SD of 100%) conditions, implying that the mechanism of denitrification in a vadose zone, if it exists, might be possibly similar under varying conditions.