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 distribution and antimicrobial resistance profiles of clinically isolated Haemophilus influenzae and Moraxella catarrhalis in hospitals across China from 2015 to 2021,and provide evidence for rational use of antimicrobial agents.Methods Data of H.influenzae and M.catarrhalis strains isolated from 2015 to 2021 in CHINET program were collected for analysis,and antimicrobial susceptibility testing was performed by disc diffusion method or automated systems according to the uniform protocol of CHINET.The results were interpreted according to the CLSI breakpoints in 2022.Beta-lactamases was detected by using nitrocefin disk.Results From 2015 to 2021,a total of 43 642 strains of Haemophilus species were isolated,accounting for 2.91％of the total clinical isolates and 4.07％of Gram-negative bacteria in CHINET program.Among the 40 437 strains of H.influenzae,66.89％were isolated from children and 33.11％were isolated from adults.More than 90％of the H.influenzae strains were isolated from respiratory tract specimens.The prevalence of β-lactamase was 53.79％in H.influenzae strains.The H.influenzae strains isolated from children showed higher resistance rate than the strains isolated from adults.Overall,779 strains of H.influenzae did not produce β-lactamase but were resistant to ampicillin(BLNAR).Beta-lactamase-producing strains showed significantly higher resistance rates to these antimicrobial agents than the β-lactamase-nonproducing strains.Of the 16 191 M.catarrhalis strains,80.06％were isolated from children and 19.94％isolated from adults.M.catarrhalis strains were mostly susceptible to both amoxicillin-clavulanic acid and cefuroxime,evidenced by resistance rate lower than 2.0％.Conclusions The emergence of antibiotic-resistant H.influenzae due to β-lactamase production poses a challenge for clinical anti-infective treatment.Therefore,it is very important to implement antibiotic resistance surveillance for H.influenzae and guide rational antibiotic use.All local clinical microbiology laboratories should actively improve antibiotic susceptibility testing and strengthen antibiotic resistance surveillance for H.influenzae.

Objective To investigate the changing antibiotic resistance profiles of E.coli isolated from patients in the 52 hospitals participating in the CHINET program from 2015 to 2021.Methods Antimicrobial susceptibility was tested for clinical isolates of E.coli according to the unified protocol of CHINET program.WHONET 5.6 and SPSS 20.0 software were used for data analysis.Results Atotal of 289 760 nonduplicate clinical strains ofE.coli were isolated from 2015 to 2021,mainly from urine samples(44.7±3.2)％.The proportion of E.coli strains isolated from urine samples was higher in females than in males(59.0％vs 29.5％).The proportion of E.coli strains isolated from respiratory tract and cerebrospinal fluid samples was significantly higher in children than in adults(16.7％vs 7.8％,0.8％vs 0.1％,both P＜0.05).The isolates from internal medicine department accounted for the largest proportion(28.9±2.8)％with an increasing trend over years.Overall,the prevalence of ESBLs-producing E.coli and carbapenem resistant E.coli(CREco)was 55.9％and 1.8％,respectively during the 7-year period.The prevalence of ESBLs-producing E.coli was the highest in tertiary hospitals each year from 2015 to 2021 compared to secondary hospitals.The prevalence of CREco was higher in children's hospitals compared to secondary and tertiary hospitals each year from 2015 to 2021.The prevalence of ESBLs-producing E.coli in tertiary hospitals and children's hospitals and the prevalence of CREco in children's hospitals showed a decreasing trend over the 7-year period.The prevalence of CREco in secondary and tertiary hospitals increased slowly.Antibiotic resistance rates changed slowly from 2015 to 2021.Carbapenem drugs(imipenem,meropenem)were the most active drugs amongβ-lactams against E.coli(resistance rate≤2.1％).The resistance rates of E.coli to β-lactam/β-lactam inhibitor combinations(piperacillin-tazobactam,cefoperazone-sulbactam),aminoglycosides(amikacin),nitrofurantoin and fosfomycin(for urinary isolates only)were all less than 10％.The resistance rate of E.coli strains to antibiotics varied with the level of hospitals and the departments where the strains were isolated,especially for cefazolin and ciprofloxacin,to which the resistance rate of E.coli strains from children in non-ICU departments was significantly lower than that of the strains isolated from other departments(P＜0.05).The E.coli isolates from ICU showed higher resistance rate to most antimicrobial agents tested(excluding tigecycline)than the strains isolated from other departments.The E.coli strains isolated from tertiary hospitals showed higher resistance rates to the antimicrobial agents tested(excluding tigecycline,polymyxin B,cefepime and carbapenems)than the strains from secondary hospitals and children's hospitals.Conclusions E.coli is an important pathogen causing clinical infection.More than half of the clinical isolates produced ESBL.The prevalence of CREco is increasing in secondary and tertiary hospitals over the 7-year period even though the overall prevalence is still low.This is an issue of concern.

Objective: To explore glymphatic impairment in pediatric refractory epilepsy (RE) using multi-parameter magnetic resonance imaging (MRI), assess its relationship with white-matter (WM) abnormalities and clinical indicators, and preliminarily evaluate the performance of multi-parameter MRI in discriminating RE from drug-sensitive epilepsy (DSE). Materials and Methods: We retrospectively included 70 patients with DSE (mean age, 9.7 ± 3.5 years; male:female, 37:33) and 26 patients with RE (9.0 ± 2.9 years; male:female, 12:14). The diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) index as well as fractional anisotropy (FA), mean diffusivity (MD), and nodal efficiency values were measured and compared between patients with RE and DSE. With sex and age as covariables, differences in the FA and MD values were analyzed using tract-based spatial statistics, and nodal efficiency was analyzed using a linear model. Pearson’s partial correlation was analyzed. Receiver operating characteristic (ROC) curves were used to evaluate the discrimination performance of the MRI-based machine-learning models through five-fold cross-validation. Results: In the RE group, FA decreased and MD increased in comparison with the corresponding values in the DSE group, and these differences mainly involved the callosum, right and left corona radiata, inferior and superior longitudinal fasciculus, and posterior thalamic radiation (threshold-free cluster enhancement, P < 0.05). The RE group also showed reduced nodal efficiency, which mainly involved the limbic system, default mode network, and visual network (false discovery rate, P < 0.05), and significantly lower DTI-ALPS index (F = 2.0, P = 0.049). The DTI-ALPS index was positively correlated with FA (0.25 ≤ r ≤ 0.32) and nodal efficiency (0.22 ≤ r ≤ 0.37), and was negatively correlated with the MD (-0.24 ≤ r≤ -0.34) and seizure frequency (r = -0.47). A machine-learning model combining DTI-ALPS, FA, MD, and nodal efficiency achieved a cross-validated ROC curve area of 0.83 (sensitivity, 78.2%; specificity, 84.8%). Conclusion Pediatric patients with RE showed impaired glymphatic function in comparison with patients with DSE, which was correlated with WM abnormalities and seizure frequency. Multi-parameter MRI may be feasible for distinguishing RE from DSE.

Objective: To explore glymphatic impairment in pediatric refractory epilepsy (RE) using multi-parameter magnetic resonance imaging (MRI), assess its relationship with white-matter (WM) abnormalities and clinical indicators, and preliminarily evaluate the performance of multi-parameter MRI in discriminating RE from drug-sensitive epilepsy (DSE). Materials and Methods: We retrospectively included 70 patients with DSE (mean age, 9.7 ± 3.5 years; male:female, 37:33) and 26 patients with RE (9.0 ± 2.9 years; male:female, 12:14). The diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) index as well as fractional anisotropy (FA), mean diffusivity (MD), and nodal efficiency values were measured and compared between patients with RE and DSE. With sex and age as covariables, differences in the FA and MD values were analyzed using tract-based spatial statistics, and nodal efficiency was analyzed using a linear model. Pearson’s partial correlation was analyzed. Receiver operating characteristic (ROC) curves were used to evaluate the discrimination performance of the MRI-based machine-learning models through five-fold cross-validation. Results: In the RE group, FA decreased and MD increased in comparison with the corresponding values in the DSE group, and these differences mainly involved the callosum, right and left corona radiata, inferior and superior longitudinal fasciculus, and posterior thalamic radiation (threshold-free cluster enhancement, P < 0.05). The RE group also showed reduced nodal efficiency, which mainly involved the limbic system, default mode network, and visual network (false discovery rate, P < 0.05), and significantly lower DTI-ALPS index (F = 2.0, P = 0.049). The DTI-ALPS index was positively correlated with FA (0.25 ≤ r ≤ 0.32) and nodal efficiency (0.22 ≤ r ≤ 0.37), and was negatively correlated with the MD (-0.24 ≤ r≤ -0.34) and seizure frequency (r = -0.47). A machine-learning model combining DTI-ALPS, FA, MD, and nodal efficiency achieved a cross-validated ROC curve area of 0.83 (sensitivity, 78.2%; specificity, 84.8%). Conclusion Pediatric patients with RE showed impaired glymphatic function in comparison with patients with DSE, which was correlated with WM abnormalities and seizure frequency. Multi-parameter MRI may be feasible for distinguishing RE from DSE.

Objective: To explore glymphatic impairment in pediatric refractory epilepsy (RE) using multi-parameter magnetic resonance imaging (MRI), assess its relationship with white-matter (WM) abnormalities and clinical indicators, and preliminarily evaluate the performance of multi-parameter MRI in discriminating RE from drug-sensitive epilepsy (DSE). Materials and Methods: We retrospectively included 70 patients with DSE (mean age, 9.7 ± 3.5 years; male:female, 37:33) and 26 patients with RE (9.0 ± 2.9 years; male:female, 12:14). The diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) index as well as fractional anisotropy (FA), mean diffusivity (MD), and nodal efficiency values were measured and compared between patients with RE and DSE. With sex and age as covariables, differences in the FA and MD values were analyzed using tract-based spatial statistics, and nodal efficiency was analyzed using a linear model. Pearson’s partial correlation was analyzed. Receiver operating characteristic (ROC) curves were used to evaluate the discrimination performance of the MRI-based machine-learning models through five-fold cross-validation. Results: In the RE group, FA decreased and MD increased in comparison with the corresponding values in the DSE group, and these differences mainly involved the callosum, right and left corona radiata, inferior and superior longitudinal fasciculus, and posterior thalamic radiation (threshold-free cluster enhancement, P < 0.05). The RE group also showed reduced nodal efficiency, which mainly involved the limbic system, default mode network, and visual network (false discovery rate, P < 0.05), and significantly lower DTI-ALPS index (F = 2.0, P = 0.049). The DTI-ALPS index was positively correlated with FA (0.25 ≤ r ≤ 0.32) and nodal efficiency (0.22 ≤ r ≤ 0.37), and was negatively correlated with the MD (-0.24 ≤ r≤ -0.34) and seizure frequency (r = -0.47). A machine-learning model combining DTI-ALPS, FA, MD, and nodal efficiency achieved a cross-validated ROC curve area of 0.83 (sensitivity, 78.2%; specificity, 84.8%). Conclusion Pediatric patients with RE showed impaired glymphatic function in comparison with patients with DSE, which was correlated with WM abnormalities and seizure frequency. Multi-parameter MRI may be feasible for distinguishing RE from DSE.

Objective: To explore glymphatic impairment in pediatric refractory epilepsy (RE) using multi-parameter magnetic resonance imaging (MRI), assess its relationship with white-matter (WM) abnormalities and clinical indicators, and preliminarily evaluate the performance of multi-parameter MRI in discriminating RE from drug-sensitive epilepsy (DSE). Materials and Methods: We retrospectively included 70 patients with DSE (mean age, 9.7 ± 3.5 years; male:female, 37:33) and 26 patients with RE (9.0 ± 2.9 years; male:female, 12:14). The diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) index as well as fractional anisotropy (FA), mean diffusivity (MD), and nodal efficiency values were measured and compared between patients with RE and DSE. With sex and age as covariables, differences in the FA and MD values were analyzed using tract-based spatial statistics, and nodal efficiency was analyzed using a linear model. Pearson’s partial correlation was analyzed. Receiver operating characteristic (ROC) curves were used to evaluate the discrimination performance of the MRI-based machine-learning models through five-fold cross-validation. Results: In the RE group, FA decreased and MD increased in comparison with the corresponding values in the DSE group, and these differences mainly involved the callosum, right and left corona radiata, inferior and superior longitudinal fasciculus, and posterior thalamic radiation (threshold-free cluster enhancement, P < 0.05). The RE group also showed reduced nodal efficiency, which mainly involved the limbic system, default mode network, and visual network (false discovery rate, P < 0.05), and significantly lower DTI-ALPS index (F = 2.0, P = 0.049). The DTI-ALPS index was positively correlated with FA (0.25 ≤ r ≤ 0.32) and nodal efficiency (0.22 ≤ r ≤ 0.37), and was negatively correlated with the MD (-0.24 ≤ r≤ -0.34) and seizure frequency (r = -0.47). A machine-learning model combining DTI-ALPS, FA, MD, and nodal efficiency achieved a cross-validated ROC curve area of 0.83 (sensitivity, 78.2%; specificity, 84.8%). Conclusion Pediatric patients with RE showed impaired glymphatic function in comparison with patients with DSE, which was correlated with WM abnormalities and seizure frequency. Multi-parameter MRI may be feasible for distinguishing RE from DSE.

Objective: To explore glymphatic impairment in pediatric refractory epilepsy (RE) using multi-parameter magnetic resonance imaging (MRI), assess its relationship with white-matter (WM) abnormalities and clinical indicators, and preliminarily evaluate the performance of multi-parameter MRI in discriminating RE from drug-sensitive epilepsy (DSE). Materials and Methods: We retrospectively included 70 patients with DSE (mean age, 9.7 ± 3.5 years; male:female, 37:33) and 26 patients with RE (9.0 ± 2.9 years; male:female, 12:14). The diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) index as well as fractional anisotropy (FA), mean diffusivity (MD), and nodal efficiency values were measured and compared between patients with RE and DSE. With sex and age as covariables, differences in the FA and MD values were analyzed using tract-based spatial statistics, and nodal efficiency was analyzed using a linear model. Pearson’s partial correlation was analyzed. Receiver operating characteristic (ROC) curves were used to evaluate the discrimination performance of the MRI-based machine-learning models through five-fold cross-validation. Results: In the RE group, FA decreased and MD increased in comparison with the corresponding values in the DSE group, and these differences mainly involved the callosum, right and left corona radiata, inferior and superior longitudinal fasciculus, and posterior thalamic radiation (threshold-free cluster enhancement, P < 0.05). The RE group also showed reduced nodal efficiency, which mainly involved the limbic system, default mode network, and visual network (false discovery rate, P < 0.05), and significantly lower DTI-ALPS index (F = 2.0, P = 0.049). The DTI-ALPS index was positively correlated with FA (0.25 ≤ r ≤ 0.32) and nodal efficiency (0.22 ≤ r ≤ 0.37), and was negatively correlated with the MD (-0.24 ≤ r≤ -0.34) and seizure frequency (r = -0.47). A machine-learning model combining DTI-ALPS, FA, MD, and nodal efficiency achieved a cross-validated ROC curve area of 0.83 (sensitivity, 78.2%; specificity, 84.8%). Conclusion Pediatric patients with RE showed impaired glymphatic function in comparison with patients with DSE, which was correlated with WM abnormalities and seizure frequency. Multi-parameter MRI may be feasible for distinguishing RE from DSE.

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.