ObjectiveDiscriminating atypical hepatocellular carcinoma (HCC) from other malignancies in liver nodules classified as Liver Imaging Reporting and Data System category M (LR-M) remains a significant diagnostic challenge on conventional ultrasound examination. The LR-M category, originally intended to capture non-HCC malignancies, paradoxically contains up to 63% of atypical HCCs that deviate from classic enhancement patterns, leading to potential misdiagnosis and suboptimal treatment planning. While deep learning has shown promise in HCC diagnosis, most existing models rely exclusively on single-modality ultrasound, overlooking the diagnostic benefits of integrating complementary information from multiple imaging sources. To address this gap, we propose a novel attention-weighted tri-modal ultrasound network (TUS-Net) that integrates contrast-enhanced ultrasound (CEUS), B-mode ultrasound (BUS), and time-intensity curves (TICs) to improve diagnostic accuracy for these clinically challenging lesions. MethodsOur framework incorporates a three-dimensional convolutional neural network (C3D) backbone to extract spatiotemporal features from CEUS videos, capturing dynamic vascular patterns critical for lesion characterization. To effectively fuse complementary modalities, we introduce a dual-channel feature fusion module (DCFFM) that adaptively combines features from CEUS and BUS through channel-wise attention mechanisms, allowing the model to dynamically weigh the contribution of each modality based on diagnostic relevance. Additionally, we propose a temporal intensity feature fusion module (TIFFM) that leverages quantitative hemodynamic information from TICs to guide the model’s attention toward diagnostically critical temporal phases, such as arterial wash-in and portal venous washout. The model is further enhanced by automated lesion localization using YOLOX and class activation mapping for interpretability, ensuring that predictions align with clinically meaningful imaging features. ResultsEvaluated on a tri-modal ultrasound dataset comprising 161 patients with pathologically confirmed LR-M nodules (131 atypical HCC and 30 non-HCC malignancies), our model achieved an accuracy of 86.83%, a sensitivity of 92.50%, a specificity of 75.50%, and an AUC of 89.32% in screening atypical HCC. Compared to single-modality baselines, TUS-Net demonstrated superior specificity, a clinically critical metric given the higher risk associated with misclassifying non-HCC malignancies. Ablation studies confirmed the contribution of each module, with the full model outperforming both standard C3D and 3D ResNet backbones integrated with attention mechanisms. A reader study involving junior and senior radiologists further validated the clinical utility of AI assistance, showing consistent improvements in specificity and inter-reader consistency, particularly for less experienced clinicians. ConclusionThese results surpass existing benchmark models and demonstrate the potential of our approach to enhance diagnostic precision in clinically specific cases. By intelligently fusing multi-modal ultrasound data with attention-guided mechanisms, TUS-Net offers a reliable and interpretable tool that holds promise for improving the non-invasive diagnosis of atypical HCC in challenging LR-M liver nodules.

ObjectiveDiscriminating atypical hepatocellular carcinoma (HCC) from other malignancies in liver nodules classified as Liver Imaging Reporting and Data System category M (LR-M) remains a significant diagnostic challenge on conventional ultrasound examination. The LR-M category, originally intended to capture non-HCC malignancies, paradoxically contains up to 63% of atypical HCCs that deviate from classic enhancement patterns, leading to potential misdiagnosis and suboptimal treatment planning. While deep learning has shown promise in HCC diagnosis, most existing models rely exclusively on single-modality ultrasound, overlooking the diagnostic benefits of integrating complementary information from multiple imaging sources. To address this gap, we propose a novel attention-weighted tri-modal ultrasound network (TUS-Net) that integrates contrast-enhanced ultrasound (CEUS), B-mode ultrasound (BUS), and time-intensity curves (TICs) to improve diagnostic accuracy for these clinically challenging lesions. MethodsOur framework incorporates a three-dimensional convolutional neural network (C3D) backbone to extract spatiotemporal features from CEUS videos, capturing dynamic vascular patterns critical for lesion characterization. To effectively fuse complementary modalities, we introduce a dual-channel feature fusion module (DCFFM) that adaptively combines features from CEUS and BUS through channel-wise attention mechanisms, allowing the model to dynamically weigh the contribution of each modality based on diagnostic relevance. Additionally, we propose a temporal intensity feature fusion module (TIFFM) that leverages quantitative hemodynamic information from TICs to guide the model’s attention toward diagnostically critical temporal phases, such as arterial wash-in and portal venous washout. The model is further enhanced by automated lesion localization using YOLOX and class activation mapping for interpretability, ensuring that predictions align with clinically meaningful imaging features. ResultsEvaluated on a tri-modal ultrasound dataset comprising 161 patients with pathologically confirmed LR-M nodules (131 atypical HCC and 30 non-HCC malignancies), our model achieved an accuracy of 86.83%, a sensitivity of 92.50%, a specificity of 75.50%, and an AUC of 89.32% in screening atypical HCC. Compared to single-modality baselines, TUS-Net demonstrated superior specificity, a clinically critical metric given the higher risk associated with misclassifying non-HCC malignancies. Ablation studies confirmed the contribution of each module, with the full model outperforming both standard C3D and 3D ResNet backbones integrated with attention mechanisms. A reader study involving junior and senior radiologists further validated the clinical utility of AI assistance, showing consistent improvements in specificity and inter-reader consistency, particularly for less experienced clinicians. ConclusionThese results surpass existing benchmark models and demonstrate the potential of our approach to enhance diagnostic precision in clinically specific cases. By intelligently fusing multi-modal ultrasound data with attention-guided mechanisms, TUS-Net offers a reliable and interpretable tool that holds promise for improving the non-invasive diagnosis of atypical HCC in challenging LR-M liver nodules.

Objective:To investigate the nutritional risk status of children in PICU and analyze its influencing factors.Methods:From July 2021 to February 2023,all children aged 1 to 18 years admitted to PICU of Beijing Children's Hospital were investigated by using the pediatric Yorkhill Malnutrition Scoring tool(PYMS)and the clinical data questionnaire.Results:A total of 492 children in PICU were enrolled.The first nutritional risk screening results showed that there were 32 cases of no/low nutritional risk(6.5%),76 cases of medium risk(15.4%),and 384 cases of high risk(78.1%).The incidence of medium/high nutritional risk was as high as 93.5%.The PYMS score of nutritional risk in PICU was(2.61±1.42).The results of multiple linear regression analysis showed that weight,fever time before admission,white blood cells,body mass index,primary diagnosis,father's education,and diet before illness were the main influencing factors of nutritional risk of children in PICU(P<0.05).Conclusion:Children in PICU are in a state of high nutritional risk.It is suggested that children in PICU should carry out nutritional screening in a standardized manner,identify children with high nutritional risk and its influencing factors early.To actively conduct nutritional assessment and nutritional intervention could improve the clinical outcome of children in PICU.

Bismuth iodide(BiOI)with different crystal plane ratios of(110)to(001)was synthesized,and typeⅡheterojunction formed between(001)and(110)crystal planes of BiOI was used to improve the separation efficiency of photogenerated electrons and holes.Then the BiOI(001)/(110)was composited with nitrogen-doped graphene quantum dots(N-GQDs)to prepare a ternary composites,which could enhance the range and intensity of light absorption,and prolonged the lifetime of photogenerated electrons due to the formation of Z-scheme heterojunctions between BiOI and N-GQDs,thereby leading to the excellent photoelectric performance of the BiOI/N-GQDs for generating sensitive photoelectric response signals.A photoelectrochemical sensor for sensitive detection of chlorpyrifos(CPF)was designed with BiOI/N-GQDs-modified FTO electrode as a photocathode.The S and N atoms contained in CPF were coordinated with Bi(Ⅲ)on the surface of BiOI,which reduced the photocurrent of BiOI/N-GQDs.The photocurrent change was linear with logarithm of concentration of CPF in the range of 1.5×10-12-5.0×10-9 mol/L,and the detection limit was 1.5×10-12 mol/L.The sensor was highly sensitive,selective and stable,and could be used for determination of trace CPF in environmental and food samples.

ObjectivePrimary liver cancer, predominantly hepatocellular carcinoma (HCC), is a significant global health issue, ranking as the sixth most diagnosed cancer and the third leading cause of cancer-related mortality. Accurate and early diagnosis of HCC is crucial for effective treatment, as HCC and non-HCC malignancies like intrahepatic cholangiocarcinoma (ICC) exhibit different prognoses and treatment responses. Traditional diagnostic methods, including liver biopsy and contrast-enhanced ultrasound (CEUS), face limitations in applicability and objectivity. The primary objective of this study was to develop an advanced, light-weighted classification network capable of distinguishing HCC from other non-HCC malignancies by leveraging the automatic analysis of brightness changes in CEUS images. The ultimate goal was to create a user-friendly and cost-efficient computer-aided diagnostic tool that could assist radiologists in making more accurate and efficient clinical decisions. MethodsThis retrospective study encompassed a total of 161 patients, comprising 131 diagnosed with HCC and 30 with non-HCC malignancies. To achieve accurate tumor detection, the YOLOX network was employed to identify the region of interest (ROI) on both B-mode ultrasound and CEUS images. A custom-developed algorithm was then utilized to extract brightness change curves from the tumor and adjacent liver parenchyma regions within the CEUS images. These curves provided critical data for the subsequent analysis and classification process. To analyze the extracted brightness change curves and classify the malignancies, we developed and compared several models. These included one-dimensional convolutional neural networks (1D-ResNet, 1D-ConvNeXt, and 1D-CNN), as well as traditional machine-learning methods such as support vector machine (SVM), ensemble learning (EL), k-nearest neighbor (KNN), and decision tree (DT). The diagnostic performance of each method in distinguishing HCC from non-HCC malignancies was rigorously evaluated using four key metrics: area under the receiver operating characteristic (AUC), accuracy (ACC), sensitivity (SE), and specificity (SP). ResultsThe evaluation of the machine-learning methods revealed AUC values of 0.70 for SVM, 0.56 for ensemble learning, 0.63 for KNN, and 0.72 for the decision tree. These results indicated moderate to fair performance in classifying the malignancies based on the brightness change curves. In contrast, the deep learning models demonstrated significantly higher AUCs, with 1D-ResNet achieving an AUC of 0.72, 1D-ConvNeXt reaching 0.82, and 1D-CNN obtaining the highest AUC of 0.84. Moreover, under the five-fold cross-validation scheme, the 1D-CNN model outperformed other models in both accuracy and specificity. Specifically, it achieved accuracy improvements of 3.8% to 10.0% and specificity enhancements of 6.6% to 43.3% over competing approaches. The superior performance of the 1D-CNN model highlighted its potential as a powerful tool for accurate classification. ConclusionThe 1D-CNN model proved to be the most effective in differentiating HCC from non-HCC malignancies, surpassing both traditional machine-learning methods and other deep learning models. This study successfully developed a user-friendly and cost-efficient computer-aided diagnostic solution that would significantly enhances radiologists’ diagnostic capabilities. By improving the accuracy and efficiency of clinical decision-making, this tool has the potential to positively impact patient care and outcomes. Future work may focus on further refining the model and exploring its integration with multimodal ultrasound data to maximize its accuracy and applicability.

Objective To analyze the epidemiological distribution,microbiological characteristics,drug-resistance status,and risk factors for mortality in adult intensive care unit(ICU)patients with Klebsiella pneumoniae infection.Methods This multi-center prospective cohort study included ICU patients with suspected infection from 67 hospitals across 16 Chinese provinces/municipalities between July 1,2021 and December 31,2022.Clinical data and microbiological results were collected,and patients were divided into survival and non-survival groups according to their survival status and drug-resistance situation.Risk factors for mortality and drug resistance in ICU patients with Klebsiella pneumonia infection were determined through univariate and multivariate logistic regression analyses.Results A total of 2964 ICU-infected patients were enrolled,with 12 175 microbial specimens submitted for testing.Among these,487 specimens tested positive for Klebsiella pneumoniae.Ultimately,314 patients with Klebsiella pneumoniae infection were identified,primarily from lung infections,with a drug-resistance rate of 78.3%.The in-hospital mortality rate of ICU patients infected with Klebsiella pneumoniae was 19.8%.Univariate and multivariate logistic regression analyses revealed that older age(P=0.027),high drug-resistance rate(P=0.028),and low clinical-effectiveness rate(P<0.001)were independent risk factors for mortality in ICU patients infected with Klebsiella pneumoniae.Drug-resistance analysis showed that,compared with non-resistant cases,ICU patients with drug-resistant Klebsiella pneumoniae infection had lower pathogen-clearance rates(P=0.003),clinical-effectiveness rates(P=0.004),and antibiotic-effectiveness rates(P<0.010),and higher mortality rates(P=0.006).Patients with Klebsiella pneumoniae abdominal infection(P=0.003)and urinary tract infection(P=0.007)had higher drug-resistance incidences.There were no statistically significant differences in clinical-effectiveness rate,Klebsiella pneumoniae clearance,drug-resistance incidence,mortality rate,or hospital-stay length between patients with lung infection and those with non-lung infection of Klebsiella pneumoniae(P>0.05).Compared with patients with non-bloodstream infection,patients with bloodstream infection of Klebsiella pneumoniae had lower clinical-effectiveness rates(P=0.027)and higher mortality rates(P=0.021).Conclusions Older age,high drug-resistance rate,and low clinical-effectiveness rate are independent risk factors for mortality in ICU patients infected with Klebsiella pneumoniae.ICU patients with bloodstream infection of Klebsiella pneumoniae may have lower clinical-effectiveness rates and higher mortality rates.ICU patients with abdominal and urinary tract infections caused by Klebsiella pneumoniae are more likely to develop drug resistance.

Objective To explore the integration value of mobile virtual reality devices in the classroom teaching of human anatomy,and to evaluate their potential impact on the in-depth construction of human anatomy knowledge,the cultivation of spatial cognitive ability,and the transformation of teaching paradigms from the perspectives of cognitive load theory and situated learning.Methods The undergraduate students majoring in clinical medicine in Peking University were selected as the research objects.Among them,students in grade 2019 were the control group,and students in grade 2022 were the experimental group,introducing movable virtual anatomy equipment and other teaching auxiliary method in theory and practice courses.The final exam scores of the two groups of students were compared,and a questionnaire survey was conducted for the experimental group after the course,and the survey result were statistically analyzed.Results The final examination result showed that the average score of the experimental group was 82.47±10.19,and the average score of the control group was 74.82±16.56,which was significantly higher in the experimental group than in the control group,with statistical significance(P＜0.05).The questionnaire survey result showed that compared with traditional classroom teaching,94.62％of students preferred the new auxiliary teaching mode such as VR,96.77％of students believed that VR assisted teaching could achieve the traditional teaching effect or better,95.7％of them think that it improved students' interest in learning human anatomy,and 98.92％thought that it improved students' knowledge of anatomy.Conclusion The application of mobile virtual reality devices in anatomy classroom teaching provides immersive and interactive 3D visualization teaching scenarios,effectively reducing students' cognitive load on abstract and complex anatomical structures,promoting spatial understanding and knowledge internalization,significantly improving teaching effectiveness and self-learning ability,thus changing the traditional anatomy teaching mode and laying a solid foundation for the development of future medical education and the cultivation of medical talents.

BACKGROUND: Atrial fibrillation (AF) is a common cardiac arrhythmia in the elderly. This study aimed to evaluate urban-rural disparities in its prevalence and management in elderly Chinese. METHODS: Consecutive participants aged ≥ 65 years attending outpatient clinics were enrolled for AF screening using handheld single-lead electrocardiogram (ECG) from April 2017 to December 2022. Each ECG rhythm strip was reviewed from the research team. AF or uninterpretable single-lead ECGs were referred for 12-lead ECG. Primary study outcome comparison was between rural and urban areas for the prevalence of AF. The Student's t-test was used to compare mean values of clinical characteristics between rural and urban participants, while the Pearson's chi-square test was used to compare between-group proportions. Multivariate stepwise logistic regression analysis was performed to estimate the association between AF and various patient characteristics. RESULTS: The 29,166 study participants included 13,253 men (45.4%) and had a mean age of 72.2 years. The 7073 rural participants differed significantly (P ≤ 0.02) from the 22,093 urban participants in several major characteristics, such as older age, greater body mass index, and so on. The overall prevalence of AF was 4.6% (n = 1347). AF was more prevalent in 7073 rural participants than 22,093 urban participants (5.6% vs. 4.3%, P < 0.01), before and after adjustment for age, body mass index, blood pressure, pulse rate, cigarette smoking, alcohol consumption and prior medical history. Multivariate logistic regression analysis identified overweight/obesity (OR = 1.35, 95% CI: 1.17-1.54) in urban areas and cigarette smoking (OR = 1.62, 95% CI: 1.20-2.17) and alcohol consumption (OR = 1.42, 95% CI: 1.04-1.93) in rural areas as specific risk factors for prevalent AF. In patients with known AF in urban areas (n = 781) and rural areas (n = 338), 60.6% and 45.9%, respectively, received AF treatment (P < 0.01), and only 22.4% and 17.2%, respectively, received anticoagulation therapy (P = 0.05). CONCLUSIONS In China, there are urban-rural disparities in AF in the elderly, with a higher prevalence and worse management in rural areas than urban areas. Our study findings provide insight for health policymakers to consider urban-rural disparity in the prevention and treatment of AF.

OBJECTIVE: High-altitude hypoxia exposure often damages hippocampus-dependent learning and memory. Nogo-A is an important axonal growth inhibitory factor. However, its function in high-altitude hypoxia and its mechanism of action remain unclear. METHODS: In an in vivo study, a low-pressure oxygen chamber was used to simulate high-altitude hypoxia, and genetic or pharmacological intervention was used to block the Nogo-A/NgR1 signaling pathway. Contextual fear conditioning and Morris water maze behavioral tests were used to assess learning and memory in rats, and synaptic damage in the hippocampus and changes in oxidative stress levels were observed. In vitro, SH-SY5Y cells were used to assess oxidative stress and mitochondrial function with or without Nogo-A knockdown in Oxygen Glucose-Deprivation/Reperfusion (OGD/R) models. RESULTS: Exposure to acute high-altitude hypoxia for 3 or 7 days impaired learning and memory in rats, triggered oxidative stress in the hippocampal tissue, and reduced the dendritic spine density of hippocampal neurons. Blocking the Nogo-A/NgR1 pathway ameliorated oxidative stress, synaptic damage, and the learning and memory impairment induced by high-altitude exposure. CONCLUSION: Our results demonstrate the detrimental role of Nogo-A protein in mediating learning and memory impairment under high-altitude hypoxia and suggest the potential of the Nogo-A/NgR1 signaling pathway as a crucial therapeutic target for alleviating learning and memory dysfunction induced by high-altitude exposure. GRAPHICAL ABSTRACT available in www.besjournal.com.
Animals ; Oxidative Stress ; Hippocampus/metabolism* ; Rats ; Nogo Proteins/genetics* ; Male ; Rats, Sprague-Dawley ; Hypoxia/metabolism* ; Altitude ; Synapses ; Humans ; Altitude Sickness/metabolism*

Abelmoschi Corolla, the dried corolla of Abelmoschus manihot, has anti-inflammatory, antioxidant, and anti-fibrosis activities. Its chemical constituents mainly include flavonoids, organic acids, steroids, and polysaccharides. This study reviewed the research progress in the chemical constituents and pharmacological activities of Abelmoschi Corolla in recent 20 years. According to the concept of quality marker(Q-marker), the Q-markers of Abelmoschi Corolla were predicted from plant phylogeny, chemical constituent specificity, traditional efficacy, chemical constituent measurability, and absorbed constituents. The primary Q-markers for Abelmoschi Corolla were anticipated to include quercetin-3'-O-β-D-glucopyranoside, gossypetin-8-O-β-D-glucuronide, isoquercetin, myricetin,quercetin, and hyperoside, with the aim of providing reference data for improving the quality evaluation system of Abelmoschi Corolla.
Abelmoschus/chemistry* ; Drugs, Chinese Herbal/pharmacology* ; Flowers/chemistry* ; Humans ; Animals ; Quality Control ; Flavonoids/chemistry*