ObjectiveTo investigate the present situation of input, output, outcome and impact of all registered community-based rehabilitation stations in Inner Mongolia in China, and analyze how the input predict the output, outcome and impact. MethodsFrom March 1st to April 30th, 2025, a questionnaire survey was conducted on all registered community-based rehabilitation stations in Inner Mongolia, covering four dimensions: input, output, outcome and impact. A total of 1 365 questionnaires were distributed. The input included four items: laws and policies, human resources, equipment and facilities, and rehabilitation information management. The output included two items: technical paths and benefits/effectiveness. The outcome included three items: coverage rates, rehabilitation interventions and functional results. The impact included two items: health and sustainability. Each item contained several questions, all of which were described in a positive way. Each question was scored from one to five. A lower score indicated that the situation of the community-based rehabilitation station was more in line with the content described in the question. Regression analysis was performed using the total score of each item of input dimension as independent variables, and the total scores of the output, outcome and impact dimensions as dependent variables. ResultsA total of 1 262 valid questionnaires were collected. The mean values of input, output, outcome and impact of community-based rehabilitation stations were 1.827 to 1.904, with coefficient of variation of 45.892% to 49.239%. The regression analysis showed that, rehabilitation information management, human resources, and laws and policies significantly predicted the output dimension (R² = 0.910, P < 0.001). Meanwhile, all four items in the input dimension predicted both the outcome (R² = 0.850, P < 0.001) and impact dimensions (R² = 0.833, P < 0.001). ConclusionInput, output, outcome and impact of the community-based rehabilitation stations in Inner Mongolia were generally in line with the content of the questions, although some imbalances were observed. Additionally, the input of community-based rehabilitation stations could significantly predict their output, outcome and impact.

Objective To investigate the causal relationship between metabolic syndrome and breast cancer by using a bidirectional two-sample Mendelian randomization (MR) approach. Methods Genome-wide association study (GWAS) summary statistics for metabolic syndrome and breast cancer were acquired from the Integrative Epidemiology Unit GWAS database and the GWAS Catalog, with populations encompassing the United States and East Asia. A bidirectional causal design was employed: a forward analysis with metabolic syndrome as the exposure and breast cancer as the outcome, followed by a reverse analysis wherein their roles were interchanged. The inverse-variance weighting (IVW) method was primarily used for effect estimation, supplemented by MR-Egger regression, the weighted median method, the simple mode method, and the weighted mode method. Instrument variable strength was screened using the F-statistic (F>10). Robustness of the results was assessed through heterogeneity tests, horizontal pleiotropy tests, forest plots, and leave-one-out sensitivity analyses. Results The IVW analysis indicated no significant causal relationship between metabolic syndrome and breast cancer (OR=1.00, 95%CI: 0.97-1.03), P>0.05). Sensitivity analyses yielded consistent results, suggesting the good robustness of the study findings. Conclusion This study found no evidence to support a causal relationship, either positive or negative, between metabolic syndrome and breast cancer.

Hypertension ranks among the most prevalent chronic non-communicable diseases worldwide, with its prevalence rising steadily and exerting substantial impacts on the demographic composition of voluntary blood donors and the safety of blood supply. The eligibility of hypertensive individuals for safe blood donation and the relevant screening and management strategies have long remained controversial in transfusion medicine. This article systematically reviews the epidemiological characteristics of hypertension, donation-related physiological alterations and underlying risk mechanisms. By synthesizing evidence from international multicenter studies and regulatory practices across major nations, it conducts a comprehensive analysis of blood donation safety and donor admission criteria for hypertensive patients. Available evidence demonstrates that for hypertensives with well-stabilized blood pressure and no severe cardiac, cerebral or renal complications or target organ damage, the overall incidence of adverse reactions following blood donation is comparable to that in healthy donors, and no definitive correlation has been identified between blood pressure levels and vasovagal responses. Current mainstream international management frameworks are gradually shifting from screening reliant on a single blood pressure cutoff value toward an integrated strategy combining stratified risk classification and individualized dynamic assessment, with core focus on long-term blood pressure control, consistency of medication regimens and complication status. Meanwhile, emerging technologies including artificial intelligence, ambulatory blood pressure monitoring and wearable devices have opened innovative avenues for blood donation risk evaluation and full-cycle donor management. In light of existing clinical evidence, future efforts should further refine the multi-dimensional risk stratification system, facilitate the implementation of intelligent screening and continuous monitoring technologies, and scale up health education and follow-up management for hypertensive populations. Such measures aim to appropriately expand the pool of eligible blood donors while safeguarding donation safety, ultimately achieving coordinated improvement in blood safety and blood supply capacity.

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.

ObjectiveTreadmill exercise intervention was evaluated for its effects on addictive behavior, learning and memory, and hippocampal neuroinflammation in methamphetamine (MA)-addicted mice, and transcriptomic screening was performed to identify candidate molecules involved in exercise-mediated attenuation of MA-related neuronal injury. MethodsTwo-month-old male C57BL/6 mice were randomly assigned to four groups: control (C), exercise control (CE), methamphetamine (M), and methamphetamine plus exercise (ME). Mice in the M and ME groups received intraperitoneal injections of MA (1 mg/kg) once daily for 7 consecutive days to establish the addiction model, whereas mice in the C and CE groups received equal volumes of saline. Subsequently, mice in the CE and ME groups underwent treadmill exercise at 12 m/min for 60 min/d for 2 weeks. Conditioned place preference (CPP), the Y-maze, and the Morris water maze (MWM) were used to assess addictive behavior, spatial recognition memory, and spatial learning and memory, respectively. Hippocampal neuronal ultrastructure was examined by transmission electron microscopy (TEM), and microglia-related responses were evaluated by immunofluorescence staining for ionized calcium-binding adapter molecule 1 (Iba-1) in the hippocampal CA1 region. Previously generated and publicly available RNA-seq data from brain tissue of the C, M, and ME groups were reanalyzed, with three biological samples per group. Differentially expressed genes (DEGs) were re-screened using the criteria of P<0.05 and fold change>1.5 or <0.67, followed by gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. NLR family CARD domain containing 5 (Nlrc5) was further validated by reverse transcription-quantitative polymerase chain reaction (RT-qPCR), and Western blot was used to determine the protein expression of NLRC5, nuclear factor kappa B (NF‑κB), interleukin-1β (IL-1β), postsynaptic density-95 (PSD-95), and synaptophysin. ResultsTreadmill exercise significantly attenuated MA-induced conditioned place preference and improved learning and memory performance in both the Y-maze and MWM tests. Exercise increased exploration time and distance in the novel arm of the Y-maze, shortened escape latency, and increased the number of platform crossings in the MWM. Average swimming speed during the probe trial did not differ significantly among groups, suggesting that the MWM differences were not attributable to obvious differences in swimming performance. TEM revealed that MA exposure induced mitochondrial swelling, disrupted cristae, reduced organelle abundance, and nuclear envelope abnormalities in hippocampal neurons, whereas these ultrastructural changes were alleviated after exercise. Immunofluorescence staining showed that MA increased Iba-1 fluorescence intensity in the hippocampal CA1 region, while treadmill exercise significantly reduced this change. Transcriptomic analysis identified 160 DEGs associated with MA exposure and 58 DEGs associated with exercise intervention, with 9 overlapping DEGs. KEGG enrichment analysis showed that MA-related DEGs were enriched in the NOD-like receptor signaling pathway. Among the overlapping DEGs, Nlrc5, which was downregulated after MA exposure and restored following exercise intervention, was selected as a candidate gene for further validation. RT-qPCR confirmed that treadmill exercise reversed the MA-induced decrease in Nlrc5 mRNA expression. Western blot analysis further showed that treadmill exercise increased NLRC5, PSD-95, and synaptophysin protein expression while reducing NF-κB and IL-1β expression. ConclusionTwo weeks of treadmill exercise attenuated MA-induced conditioned place preference and ameliorated learning and memory impairment, alleviated hippocampal neuroinflammation and neuronal ultrastructural damage, and restored the expression of synapse-related proteins. Transcriptomic screening combined with molecular validation demonstrated that exercise intervention was accompanied by restoration of NLRC5 expression, reduced NF‑κB and IL-1β expression, and increased PSD-95 and synaptophysin expression. These findings suggest that NLRC5 may represent a candidate molecule involved in exercise-mediated amelioration of MA-related neuronal injury. However, the present results mainly demonstrate coordinated changes in NLRC5 expression, inflammatory indicators, and cognitive outcomes rather than a direct causal relationship, and the specific role of NLRC5 requires further functional validation.

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 explore the efficacy and influencing factors of intravitreal injection of ranibizumab in the treatment of macular edema （ME） secondary to retinal vein obstruction （RVO） with different optical coherence tomography （OCT） subtypes. METHODS A retrospective study was conducted on 150 patients with ME secondary to RVO treated at Dept. of Ocular Trauma of Tianjin Eye Hospital between January 1， 2021 and January 1， 2024. According to OCT findings， patients were classified into the diffuse retinal thickening （DRT） group （48 cases）， cystoid macular edema （CME） group （83 cases）， and serous retinal detachment （SRD） group （19 cases）. The best corrected visual acuity （BCVA） and central macular thickness （CMT） were compared before and at 1， 3 and 6 months after treatment. Clinical efficacies of 3 groups were compared based on CMT and fluorescein fundus angiography （FFA） findings before and after treatment. Adverse events and the number of additional injections of ranibizumab during treatment were compared among 3 groups. Using “ineffectiveness” in clinical outcomes at 6 months post- treatment as the dependent variable and patients’ baseline data as the independent variables， a multivariate Logistic regression analysis was conducted to identify risk factors influencing the clinical efficacy of ranibizumab. RESULTS The proportion of branch RVO was significantly higher in the CME and SRD groups than in the DRT group （P＜0.05）， while central RVO （CRVO） was more frequent in the DRT group than in the CME and SRD groups （P＜0.05）. The proportion of patients with ischemia was highest in the SRD group， followed by the CME and DRT groups （P＜0.05）， while the proportion of patients with ischemia in the CME group was significantly higher than that in the DRT group （P＜0.05）. Before treatment， the BCVA and CMT showed no significant differences among the 3 groups （P＞0.05）. After treatment， BCVA and CMT in all 3 groups were significantly reduced compared to those before treatment （P＜0.05）. At different treatment time points， patients in the CME group and SRD group consistently showed significantly higher BCVA and CMT values compared to those in the DRT group （P＜0.05）. Six months after treatment， the differences in clinical efficacy among the 3 groups were statistically significant （P＜0.05）， with the proportion of non-responders in the SRD group being significantly higher than that in the DRT group and the CME group （P＜0.05）. The number of additional injections of ranibizumab in patients from the CME group and the SRD group was significantly more than that in the DRT group （P＜0.05）. The incidence of adverse reactions did not differ significantly among 3 groups （P＞0.05）. Multivariate Logistic regression revealed that CRVO and ischemic type were common risk factors affecting the clinical efficacy of ranibizumab in all 3 groups， while longer disease duration was an independent risk factor for the clinical efficacy of ranibizumab in patients from the DRT group. CONCLUSIONS The therapeutic efficacy of ranibizumab varies among different OCT phenotypes of ME secondary to RVO. DRT patients achieve the best visual improvement， SRD patients have the highest non-response rate， and CME/SRD patients require more additional injections of ranibizumab. CRVO and ischemia are shared adverse prognostic factors for poor prognosis in various subtypes of ME secondary to RVO. Individualized treatment and follow-up strategies should be developed based on OCT patterns and risk factors.