C-DARL: Contrastive diffusion adversarial representation learning for label-free blood vessel segmentation.

Blood vessel segmentation in medical imaging is one of the essential steps for vascular disease diagnosis and interventional planning in a broad spectrum of clinical scenarios in image-based medicine and interventional medicine. Unfortunately, manual annotation of the vessel masks is challenging and resource-intensive due to subtle branches and complex structures. To overcome this issue, this paper presents a self-supervised vessel segmentation method, dubbed the contrastive diffusion adversarial representation learning (C-DARL) model. Our model is composed of a diffusion module and a generation module that learns the distribution of multi-domain blood vessel data by generating synthetic vessel images from diffusion latent. Moreover, we employ contrastive learning through a mask-based contrastive loss so that the model can learn more realistic vessel representations. To validate the efficacy, C-DARL is trained using various vessel datasets, including coronary angiograms, abdominal digital subtraction angiograms, and retinal imaging. Experimental results confirm that our model achieves performance improvement over baseline methods with noise robustness, suggesting the effectiveness of C-DARL for vessel segmentation.Our source code is available at https://github.com/boahK/MEDIA_CDARL.2.

C-DARL: Contrastive diffusion adversarial representation learning for label-free blood vessel segmentation.

Blood vessel segmentation in medical imaging is one of the essential steps for vascular disease diagnosis and interventional planning in a broad spectrum of clinical scenarios in image-based medicine and interventional medicine. Unfortunately, manual annotation of the vessel masks is challenging and resource-intensive due to subtle branches and complex structures. To overcome this issue, this paper presents a self-supervised vessel segmentation method, dubbed the contrastive diffusion adversarial representation learning (C-DARL) model. Our model is composed of a diffusion module and a generation module that learns the distribution of multi-domain blood vessel data by generating synthetic vessel images from diffusion latent. Moreover, we employ contrastive learning through a mask-based contrastive loss so that the model can learn more realistic vessel representations. To validate the efficacy, C-DARL is trained using various vessel datasets, including coronary angiograms, abdominal digital subtraction angiograms, and retinal imaging. Experimental results confirm that our model achieves performance improvement over baseline methods with noise robustness, suggesting the effectiveness of C-DARL for vessel segmentation.

Multi-Scale Hybrid Vision Transformer for Learning Gastric Histology: AI-Based Decision Support System for Gastric Cancer Treatment.

Gastric endoscopic screening is an effective way to decide appropriate gastric cancer treatment at an early stage, reducing gastric cancer-associated mortality rate. Although artificial intelligence has brought a great promise to assist pathologist to screen digitalized endoscopic biopsies, existing artificial intelligence systems are limited to be utilized in planning gastric cancer treatment. We propose a practical artificial intelligence-based decision support system that enables five subclassifications of gastric cancer pathology, which can be directly matched to general gastric cancer treatment guidance. The proposed framework is designed to efficiently differentiate multi-classes of gastric cancer through multiscale self-attention mechanism using 2-stage hybrid vision transformer networks, by mimicking the way how human pathologists understand histology. The proposed system demonstrates its reliable diagnostic performance by achieving class-average sensitivity of above 0.85 for multicentric cohort tests. Moreover, the proposed system demonstrates its great generalization capability on gastrointestinal track organ cancer by achieving the best class-average sensitivity among contemporary networks. Furthermore, in the observational study, artificial intelligence-assisted pathologists show significantly improved diagnostic sensitivity within saved screening time compared to human pathologists. Our results demonstrate that the proposed artificial intelligence system has a great potential for providing presumptive pathologic opinion and supporting decision of appropriate gastric cancer treatment in practical clinical settings.

Teaching with collective resilience during COVID-19: Korean teachers and collaborative professionalism.

This study applies system-focused resilience and collaborative professionalism to examine how teachers in Korea collectively developed resilience and transformed teaching during COVID-19. Using qualitative data from seven individual interviews and four focus groups, we found Korean teachers navigated complex challenges (rapidly changing policies, online teaching, exacerbated learning gaps, and excessive social pressure) and utilized contextual resources (collective autonomy and flexibility, solidity and solidarity, and collective responsibility) to develop strategies (collaborative inquiry, timely communication, and envisioning the future of schooling). The study extends teacher resilience toward more collective and communal, from the individual level, by linking resilience to collaborative systemic changes.

A waterproof cellulose nanofibril sheet prepared by the deposition of an alkyl ketene dimer on a controlled porous structure.

This study prepared a waterproof cellulose nanofibril (CNF) sheet via the deposition of an alkyl ketene dimer (AKD) on the sheet's controlled porous structure. The porosity of the CNF sheet was controlled by drying under different conditions, which included hot-press drying (HD) and solvent-exchange drying (SD), and the effect on the hydrophobization and water-related barrier performance of the sheet were investigated. When the SD sheet was immersed in an AKD wax solution, the sheet exhibited super-hydrophobicity and a lower water vapor transmission rate, compared with the HD sheet. This indicated that the porous structure of the SD sheet enabled AKD to be adsorbed on both the surface and the inner surface and it filled in the pores of the sheet, thereby giving rise to excellent waterproofing properties. The performance of a hydrophobized SD sheet as a water barrier material was comparable to a linear low-density polyethylene film. This study confirms the possibility for AKD wax to be immersed in a porous CNF sheet and used as a potential barrier material in hydrogel packaging.

Self-evolving vision transformer for chest X-ray diagnosis through knowledge distillation.

Although deep learning-based computer-aided diagnosis systems have recently achieved expert-level performance, developing a robust model requires large, high-quality data with annotations that are expensive to obtain. This situation poses a conundrum that annually-collected chest x-rays cannot be utilized due to the absence of labels, especially in deprived areas. In this study, we present a framework named distillation for self-supervision and self-train learning (DISTL) inspired by the learning process of the radiologists, which can improve the performance of vision transformer simultaneously with self-supervision and self-training through knowledge distillation. In external validation from three hospitals for diagnosis of tuberculosis, pneumothorax, and COVID-19, DISTL offers gradually improved performance as the amount of unlabeled data increase, even better than the fully supervised model with the same amount of labeled data. We additionally show that the model obtained with DISTL is robust to various real-world nuisances, offering better applicability in clinical setting.

Multi-task vision transformer using low-level chest X-ray feature corpus for COVID-19 diagnosis and severity quantification.

Developing a robust algorithm to diagnose and quantify the severity of the novel coronavirus disease 2019 (COVID-19) using Chest X-ray (CXR) requires a large number of well-curated COVID-19 datasets, which is difficult to collect under the global COVID-19 pandemic. On the other hand, CXR data with other findings are abundant. This situation is ideally suited for the Vision Transformer (ViT) architecture, where a lot of unlabeled data can be used through structural modeling by the self-attention mechanism. However, the use of existing ViT may not be optimal, as the feature embedding by direct patch flattening or ResNet backbone in the standard ViT is not intended for CXR. To address this problem, here we propose a novel Multi-task ViT that leverages low-level CXR feature corpus obtained from a backbone network that extracts common CXR findings. Specifically, the backbone network is first trained with large public datasets to detect common abnormal findings such as consolidation, opacity, edema, etc. Then, the embedded features from the backbone network are used as corpora for a versatile Transformer model for both the diagnosis and the severity quantification of COVID-19. We evaluate our model on various external test datasets from totally different institutions to evaluate the generalization capability. The experimental results confirm that our model can achieve state-of-the-art performance in both diagnosis and severity quantification tasks with outstanding generalization capability, which are sine qua non of widespread deployment.

Immobilization of laccase via cross-linked enzyme aggregates prepared using genipin as a natural cross-linker.

Genipin is a nontoxic natural cross-linker that was successfully used to prepare cross-linked enzyme aggregates (CLEAs) of Trametes versicolor laccase. The recovered activity of CLEAs was influenced by the co-solvent type, genipin concentration, cross-linking time, preparation pH, and bovine serum albumin (BSA; amino group feeder) concentration. The characteristics of CLEAs prepared using genipin under optimal conditions (genipin-BSA-CLEAs) were compared with those of typical CLEAs prepared using glutaraldehyde or dextran polyaldehyde. Genipin-BSA-CLEAs were nano-sized (average diameter, approximately 700 nm), had a ball-like shape, showed a narrow size distribution, and exhibited the highest substrate affinity among the prepared CLEAs. The thermal stability of genipin-BSA-CLEAs was 6.8-fold higher than that of free laccase, and their pH stability was also much higher than that of free laccase in the tested range. Additionally, genipin-BSA-CLEAs retained 85% of their initial activity after 10 cycles of reuse. Particularly, genipin-BSA-CLEAs showed higher thermal and pH stability than CLEAs that were cross-linked using glutaraldehyde. Therefore, genipin represents an alternative to toxic compounds such as glutaraldehyde during cross-linking to prepare CLEAs.

Effect of hydrogen bond donor on the choline chloride-based deep eutectic solvent-mediated extraction of lignin from pine wood.

In this work, twenty-five kinds of choline chloride (ChCl)-based deep eutectic solvents (DESs) containing acid, hydroxyl, amide, and binary hydrogen bond donors (HBDs) were prepared and successfully used to pretreat pine wood powder. As a result of the pretreatment, the glucan content in the pretreated biomass was increased, whereas the contents of hemicellulose and lignin were significantly decreased. The biomass pretreatment efficiency of the DESs had improved with increasing the polarity and hydrogen bond acidity (α) of the DESs. Among the studied DESs, ChCl:lactic acid:formic acid (1:1:1) with the highest α value was the most efficient DES in extracting lignin from biomass. The pretreated biomass also showed an enhanced enzymatic saccharification yield owing to the decreased particle size of the biomass and reduced content of hemicellulose and lignin. During the pretreatment process of biomass using DESs, the extracted lignin could be recovered successfully, with a yield of up to 60% and purity of over 90%. The molecular weight of the extracted lignin was much lower than that of the native cellulolytic enzyme lignin. The DES used for pretreatment process could be also successfully reused with high recovery yield of DES and high retention of delignification capacity.

Effect of Cellulose Solvents on the Characteristics of Cellulose/Fe2O3 Hydrogel Microspheres as Enzyme Supports.

Cellulose hydrogels are considered useful biocompatible and biodegradable materials. However, as few cellulose-dissolving solvents can be used to prepare cellulose hydrogel microspheres, the use of unmodified cellulose-based hydrogel microspheres for enzyme immobilization remains limited. Here, we prepared cellulose/Fe2O3 hydrogel microspheres as enzyme supports through sol-gel transition using a solvent-in-oil emulsion. Cellulose-dissolving solvents including 1-ethyl-3-methylimidazolium ([Emim][Ac]), an aqueous mixture of NaOH and thiourea, tetrabutylammonium hydroxide, and tetrabutylphosphonium hydroxide were used to prepare regular shaped cellulose/Fe2O3 microspheres. The solvent affected microsphere characteristics like crystallinity, hydrophobicity, surface morphology, size distribution, and swelling properties. The immobilization efficiency of the microspheres for lipase was also significantly influenced by the type of cellulose solvent used. In particular, the lipase immobilized on cellulose/Fe2O3 microspheres prepared using [Emim][Ac] showed the highest protein loading, and its specific activity was 3.1-fold higher than that of free lipase. The immobilized lipase could be simply recovered by a magnet and continuously reused.

Deep Learning COVID-19 Features on CXR Using Limited Training Data Sets.

Under the global pandemic of COVID-19, the use of artificial intelligence to analyze chest X-ray (CXR) image for COVID-19 diagnosis and patient triage is becoming important. Unfortunately, due to the emergent nature of the COVID-19 pandemic, a systematic collection of CXR data set for deep neural network training is difficult. To address this problem, here we propose a patch-based convolutional neural network approach with a relatively small number of trainable parameters for COVID-19 diagnosis. The proposed method is inspired by our statistical analysis of the potential imaging biomarkers of the CXR radiographs. Experimental results show that our method achieves state-of-the-art performance and provides clinically interpretable saliency maps, which are useful for COVID-19 diagnosis and patient triage.

The profiles of microRNAs from urinary extracellular vesicles (EVs) prepared by various isolation methods and their correlation with serum EV microRNAs.

AIMS: MicroRNAs (miRNAs) that circulate in biological fluids are frequently enclosed in extracellular vesicles (EVs). However, urinary EVs and their cargo miRNAs have not been systematically studied according to their EV isolation methods. METHODS: In type 2 diabetes mellitus persons with diabetic nephropathy (n = 4), we compared miRNA species in urine EVs prepared by ultracentrifugation (UC), qEV original size exclusion column (qEV), ExoQuick-TC Plus (ExoQuick), and ultrafiltration using Amicon Ultra centrifugal filter devices (Amicons) 10 K and 100 K. EV miRNAs were profiled by next-generation sequencing (NGS). Additionally, we evaluated the correlations of EV miRNA expression between the urine and serum samples isolated by UC. RESULTS: From each of 100 ml of urine, the UC method yielded the highest number of EV miRNA species (233 ± 37.3), with the ExoQuick yielded the lowest (103 ± 17.4). Urine EV miRNA profiles were highly correlated between UC, qEV, ExoQuick and Amicon 10 K methods. EV miRNA profiles between the urine and serum samples showed variable correlations between the patients (paired sample number = 3, r = 0.39-0.72). CONCLUSIONS: UC, qEV, ExoQuick, and Amicon 10 K are acceptable for urinary EV isolation to profile miRNAs. Urine- and serum-derived EV miRNA profiles have variable correlations depending on specific patients.

Impact of Changes in Maternal Age and Parity Distribution on the Increasing Trends in the Low Birth Weight and Very Low Birth Weight Rates in South Korea, 2005-2015.

OBJECTIVES: The aim of this study was to evaluate the impact of shifts in maternal age and parity on the increasing trends in the low birth weight (LBW) and very low birth weight (VLBW) rates from 2005 to 2015 in South Korea. METHODS: Data from 4 993 041 live births registered with Statistics Korea during the period between 2005 and 2015 were analyzed. Applying a modified standardization method, we partitioned the total increment in the LBW and VLBW rates into (1) the increase in the LBW and VLBW rates due to changes in the maternal age and parity distribution (AP-dis) and (2) the increase due to changes in the age-specific and parity-specific rates (AP-spe) of LBW and VLBW for singleton and multiple births, respectively. RESULTS: During the study period, the total increment in the LBW and VLBW rates was 1.43%p and 0.25%p, respectively. Among singleton births, changes in the AP-dis accounted for 79% (0.34%p) and 50% (0.06%p) of the total increment in the LBW and VLBW rates, respectively. Meanwhile, among multiple births, changes in the AP-dis did not contribute to the increase in the LBW and VLBW rates, with 100% of the increase in the LBW (1.00%p) and VLBW (0.13%p) rates being attributed to changes in the AP-spe. CONCLUSIONS: This study demonstrated that shifts in maternal age and parity were prominent contributors to the increase in the LBW and VLBW rates among singleton births between 2005 and 2015 in South Korea.

Indoor levels of volatile organic compounds and formaldehyde from emission sources at elderly care centers in Korea.

The objective of this study is to characterize indoor and outdoor levels of volatile organic compounds (VOCs) and formaldehyde (HCHO) and identify indoor emission sources in thirty elderly care centers (ECCs) located in the Seoul metropolitan city and Gyeonggi province in Korea. Air monitoring samples from indoor and outdoor environments were collected from January to December in 2007. Statistical analyses of indoor and outdoor VOCs and HCHO levels in three rooms (a bedroom, living, and dining rooms) of each ECC were performed, and these were compared to identify environmental factors associated with an increase of indoor pollution levels. Total volatile organic compounds (TVOC) levels were significantly (p<0.05) different between indoor (230.7±1.7 µg/m3) and outdoor (137.8±1.9 µg/m3) environments, with an I/O ratio of 1.67. The indoor HCHO level (20.1±1.6 µg/m3) was significantly (p<0.05) higher than the outdoor level (8.1±1.9 µg/m3), with an I/O ratio of 2.48. Indoor VOCs and HCHO levels in the bedrooms were significantly (p<0.05) higher than those in the living and dining rooms. Furthermore, indoor levels of VOCs and HCHO at ECCs were significantly (p<0.05) different depending on environmental factors such as the use of carpet, paint, and wooden furniture. In multiple regression analysis, indoor VOCs and HCHO levels at ECCs were significantly (p<0.05) correlated with two micro-environmental factors: the use of carpet and paint. This study confirmed that indoor VOCs and HCHO levels were significantly higher than those in outdoor environments. These air pollutants were mainly emitted from indoor sources, such as carpet, paint, and construction materials at the ECCs in Korea.

Protection of Radiation-Induced DNA Damage by Functional Cosmeceutical Poly-Gamma-Glutamate.

This study compared the radioprotective effects of high-molecular-weight poly-gamma-glutamate (γ-PGA, average molecular mass 3,000 kDa) and a reduced form of glutathione (GSH, a known radioprotector) on calf thymus DNA damage. The radiation-induced DNA damage was measured on the basis of the decreased fluorescence intensity after binding the DNA with ethidium bromide. All the experiments used 6°Co gamma radiation at 1,252 Gy, representing 50% DNA damage. When increasing the concentration of γ-PGA from 0.33 to 1.65 µM, the DNA protection from radiation-induced damage also increased, with a maximum of 87% protection. Meanwhile, the maximal DNA protection when increasing the concentration of GSH was only 70%. Therefore, γ-PGA exhibited significant radioprotective effects against gamma irradiation.

Macrophage-Targeted Indocyanine Green-Neomannosyl Human Serum Albumin for Intraoperative Sentinel Lymph Node Mapping in Porcine Esophagus.

BACKGROUND: The sentinel lymph node (SLN) concept has been proposed to avoid unnecessary invasive LN dissection in surgery for esophageal cancer. This study evaluated a new macrophage-targeting fluorescent agent, indocyanine green-neomannosyl human serum albumin (ICG:MSA), for SLN mapping using a custom-made intraoperative color and fluorescence-merged imaging system (ICFIS) in porcine esophagus. METHODS: The LN targeting ability of ICG:MSA, indocyanine green-human serum albumin (ICG:HSA), and ICG was examined in vitro using the U937 differentiated monocyte cell line and in vivo in a mouse footpad model using fluorescence imaging. SLN identification in rabbit esophagus was then performed using ICG:MSA, ICG:HSA, and ICG. Finally, intraoperative SLN detection was conducted in porcine esophagus after esophagoscopic injection of ICG:MSA. RESULTS: The fluorescence signal of U937 cells treated by ICG:MSA was significantly higher than that of ICG or ICG:HSA (ICG: 1.0 ± 0.37; ICG:HSA: 3.4 ± 0.28, ICG:MSA: 6.8 ± 1.61; ICG to ICG:HSA, p = 0.03; ICG:HSA to ICG:MSA, p = 0.04; ICG to ICG:MSA, p = 0.0009). ICG:MSA was retained in popliteal LNs as long as 3 h, while ICG rapidly diffused through the entire mouse lymphatic system within 5 min. Esophageal SLN was detected within 15 min after injection of either ICG or ICG:MSA, but ICG:MSA provided more distinguishable images of LNss than ICG in rabbit esophagus. The SLN was also successfully detected in all porcine esophagus; the mean number of SLNs identified per esophagus was 1.6 ± 0.55. CONCLUSIONS: ICG:MSA has more specific macrophage-targeting properties, which could overcome the limitation of the low SLN retention of ICG, and could provide more precise real-time SLN detection during esophageal cancer surgery.

Perceptions and practices of commensality and solo-eating among Korean and Japanese university students: A cross-cultural analysis.

BACKGROUND/OBJECTIVES: Commensality, eating together with others, is a major representation of human sociality. In recent time, environments around commensality have changed significantly due to rapid social changes, and the decline of commensality is perceived as a serious concern in many modern societies. This study employs a cross-cultural analysis of university students in two East Asian countries, and examines cross-cultural variations of perceptions and actual practices of commensality and solo-eating. SUBJECTS/METHODS: The analysis was drawn from a free-list survey and a self-administrative questionnaires of university students in urban Korea and Japan. The free-listing survey was conducted with a small cohort to explore common images and meanings of commensality and solo-eating. The self-administrative questionnaire was developed based on the result of the free-list survey, and conducted with a larger cohort to examine reasons and problems of practices and associated behaviors and food intake. RESULTS: We found that Korean subjects tended to show stronger associations between solo-eating and negative emotions while the Japanese subjects expressed mixed emotions towards the practice of solo-eating. In the questionnaire, more Korean students reported they prefer commensality and tend to eat more quantities when they eat commensally. In contrast, more Japanese reported that they do not have preference on commensality and there is no notable difference in food quantities when they eat commensally and alone. Compared to the general Korean cohort finding, more proportion of overweight and obese groups of Korean subjects reported that they tend to eat more when they are alone than normal and underweight groups. This difference was not found in the overweight Japanese subjects. CONCLUSION: Our study revealed cross-cultural variations of perceptions and practices of commensality and solo-eating in a non-western setting.

Intraoperative fluorescence image-guided pulmonary segmentectomy.

BACKGROUND: The intraoperative color and fluorescence-merged imaging system (ICFIS) is a new technology that may aid the demarcation of intersegmental borders during pulmonary segmentectomy. This study was performed to validate, for the first time, image-guided segmentectomy using ICFIS and to find the optimal dosage of fluorescent dye to ensure safe and sustained imaging during surgery. METHODS: Nine rabbits were subjected to pulmonary segmentectomy. These constituted three groups of three rabbits each. After ligation of the segmental pulmonary artery supplying the targeted segment, the rabbits were injected intravenously with indocyanine green (ICG) at a concentration of 0.3, 0.6, or 3.0 mg/kg, depending on their group assignment. The optimal dose was determined from the rabbit study and then used to guide ICFIS during pulmonary segmentectomy in five pigs. RESULTS: The fluorescent signal contrast ratios of the targeted area to the normal lung using ICG concentrations of 0.3, 0.6, or 3.0 mg/kg were 1.9 ± 0.25, 2.0 ± 0.17, and 2.1 ± 0.06, respectively. The mean ICG washout times were 1, 3, and 6 min, respectively. Proceeding with an ICG concentration of 0.6 mg/kg, the mean washout time was found to be longer in pigs (15 min). This provided adequate time for successful ICFIS-guided segmentectomy in all five pigs, without the requirement for additional procedures for intersegmental plane demarcation. CONCLUSIONS: ICG image-guided segmentectomy using ICFIS enabled immediate visualization of the intersegmental planes. The washout time using the ICG dose determined in this study was long enough to ensure that visualization was sustained throughout the surgery.

Thoracoscopic color and fluorescence imaging system for sentinel lymph node mapping in porcine lung using indocyanine green-neomannosyl human serum albumin: intraoperative image-guided sentinel nodes navigation.

PURPOSE: This study was performed to validate a newly developed sentinel lymph node (SLN) targeting tracer, indocyanine green-neomannosyl human serum albumin (ICG:MSA), and a thoracoscopic version of the intraoperative color and fluorescence imaging system (ICFIS) for lung cancer SLN mapping. METHODS: ICG alone or ICG:MSA (5 µg/kg) was injected into the rat thigh, and the results were compared. The fluorescence signal-to-background ratios of SLNs were recorded and evaluated over a 2-h period by using ICFIS. Additionally, a SLN biopsy was performed via video-assisted thoracoscopic surgery with the use of ICG:MSA in porcine lung by using thoracoscopic ICFIS. RESULTS: The newly developed ICG:MSA showed a significantly improved signal-to-background ratio compared with ICG alone throughout the trials. All SLNs were identified in both rats (ten SLNs in ten rat thighs) and pigs (ten SLNs in ten porcine lungs) under in vivo conditions. All SLNs were dissected successfully by using video-assisted thoracoscopic surgery with the help of thoracoscopic ICFIS. DISCUSSION: ICG:MSA accumulates in the SLN by uptake and retention through the mannose-specific receptors on macrophages. Thoracoscopic ICFIS successfully assisted SLN mapping despite low near-infrared light transmission in the commercial thoracoscope. On the basis of the results of the thoracoscopic SLN mapping, we anticipate that ICG:MSA and thoracoscopic ICFIS can be translated to clinical trials in the near future.