Myosteatosis is one of the common complications in patients with end-stage liver disease, which is significantly associated with poor outcomes after liver transplantation. Currently, diagnostic criteria of myosteatosis have not been established, and CT is the most commonly used for diagnosis. The pathogenesis of myosteatosis is multifactorial, and the pathophysiological mechanisms linking it to end-stage liver disease are not fully understood. An increasing number of scholars have recognized that the severity of myosteatosis is closely related to its clinical consequences, but there are no effective treatment options available. This article reviews the pathophysiological mechanisms and diagnostic methods of myosteatosis, and its impact on the prognosis of liver transplant recipients, and discusses current treatment strategies to provide references for the perioperative management of liver transplant recipients.

Oral squamous cell carcinoma (OSCC) is a common head and neck malignancy. Approximately 50% to 60% of patients with OSCC are diagnosed at a locally advanced stage (clinical staging III-IVa). Even with comprehensive and sequential treatment primarily based on surgery, the 5-year overall survival rate remains below 50%, and patients often suffer from postoperative functional impairments such as difficulties with speaking and swallowing. Programmed death receptor-1 (PD-1) inhibitors are increasingly used in the neoadjuvant treatment of locally advanced OSCC and have shown encouraging efficacy. However, clinical practice still faces key challenges, including the definition of indications, optimization of combination regimens, and standards for efficacy evaluation. Based on the latest research advances worldwide and the clinical experience of the expert group, this expert consensus systematically evaluates the application of PD-1 inhibitors in the neoadjuvant treatment of locally advanced OSCC, covering combination strategies, treatment cycles and surgical timing, efficacy assessment, use of biomarkers, management of special populations and immune related adverse events, principles for immunotherapy rechallenge, and function preservation strategies. After multiple rounds of panel discussion and through anonymous voting using the Delphi method, the following consensus statements have been formulated: 1) Neoadjuvant therapy with PD-1 inhibitors can be used preoperatively in patients with locally advanced OSCC. The preferred regimen is a PD-1 inhibitor combined with platinum based chemotherapy, administered for 2-3 cycles. 2) During the efficacy evaluation of neoadjuvant therapy, radiographic assessment should follow the dual criteria of Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 and immune RECIST (iRECIST). After surgery, systematic pathological evaluation of both the primary lesion and regional lymph nodes is required. For combination chemotherapy regimens, PD-L1 expression and combined positive score need not be used as mandatory inclusion or exclusion criteria. 3) For special populations such as the elderly (≥ 70 years), individuals with stable HIV viral load, and carriers of chronic HBV/HCV, PD-1 inhibitors may be used cautiously under the guidance of a multidisciplinary team (MDT), with close monitoring for adverse events. 4) For patients with a poor response to neoadjuvant therapy, continuation of the original treatment regimen is not recommended; the subsequent treatment plan should be adjusted promptly after MDT assessment. Organ transplant recipients and patients with active autoimmune diseases are not recommended to receive neoadjuvant PD-1 inhibitor therapy due to the high risk of immune related activation. Rechallenge is generally not advised for patients who have experienced high risk immune related adverse events such as immune mediated myocarditis, neurotoxicity, or pneumonitis. 5) For patients with a good pathological response, individualized de escalation surgery and function preservation strategies can be explored. This consensus aims to promote the standardized, safe, and precise application of neoadjuvant PD-1 inhibitor strategies in the management of locally advanced OSCC patients.

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.

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.

BACKGROUND: Large language models (LLMs) are being explored for disease prediction and diagnosis; however, their efficacy for early sepsis identification in emergency departments (EDs) remains unexplored. This study aims to evaluate MedGo, a novel medical LLM, as a decision-support tool for clinicians managing patients with suspected sepsis. METHODS: This retrospective study included anonymized medical records of 203 patients (mean age 79.9±10.2 years) with confirmed sepsis from a tertiary hospital ED between January 2023 and January 2024. MedGo performance across nine sepsis-related assessment tasks was compared with that of two junior (<3 years of experience) and two senior (>10 years of experience) ED physicians. Assessments were scored on a 5-point Likert scale for accuracy, comprehensiveness, readability, and case-analysis skills. RESULTS: MedGo demonstrated diagnostic performance comparable to that of senior physicians across most metrics, achieving a median Likert score of 4 in accuracy, comprehensiveness, and readability. MedGo significantly outperformed junior physicians (P<0.001 for accuracy and case-analysis skills). MedGo assistance significantly enhanced both junior (P<0.001) and senior (P<0.05) physicians' diagnostic accuracy. Notably, MedGo-assisted junior physicians achieved accuracy levels comparable to those of unassisted senior physicians. MedGo maintained consistent performance across varying sepsis severities. CONCLUSION: MedGo shows significant diagnostic efficacy for sepsis and effectively supports clinicians in the ED, particularly enhancing junior physicians’ performance. Our study highlights the potential of MedGo as a valuable decision-support tool for sepsis management, paving the way for specialized sepsis AI models.

Objective To quantitatively assess the association of short-term exposure to polycyclic aromatic hydrocarbons (PAHs) in ambient fine particulate matter (PM2.5) with residents mortality. Methods A time-stratified case-crossover study was conducted from 2020 to 2022 among 10606 non-accidental residents by using the Guangzhou Cause of Death Surveillance System in Conghua District, Guangzhou. Exposure levels of PAHs in PM_2.5 and meteorological data during the study period were obtained from the Center for Disease Control and Prevention in Conghua District and the China Meteorological Administration Land Data Assimilation System (CLDAS-V2.0), respectively. Conditional Poisson regression model was used to estimate the exposure-response association between PAHs and the mortality risk. Results Fluoranthene, chrysene, benzo[k]fluoranthene, benzo[a]pyrene, and indeno[1,2,3-cd]pyrene were significantly associated with an increased risk of mortality. For every one interquartile range increase in exposure levels, the non-accidental mortality risks increased by 8.33% (95% CI: 1.80%, 15.27%), 4.67% (95% CI: 1.86%, 7.57%), 6.07% (95% CI: 2.08%, 10.21%), 4.62% (95% CI: 1.85%, 7.47%), and 4.70% (95% CI: 0.53%, 9.03%), respectively. The estimated non accidental deaths attributable to exposure to fluoranthene, chrysene, benzo[k]fluorine, benzo[a]pyrene and indine[1,2,3-cd]pyrene were 5.91%, 6.08%, 6.51%, 6.46%, and 4.21%, respectively. Conclusions Short-term exposure to PAHs in PM_2.5, including fluoranthene, chrysene, benzo[k]fluoranthene, benzo[a]pyrene and indine[1,2,3-cd]pyrene, was significantly associated with an increased risk of mortality among residents.

Objective To establish a rat model of acute antibody-mediated rejection (AMR) in kidney transplantation and investigate the preventive effect of dihydroartemisinin (DHA) on acute AMR. Methods BN rats were used as donors and Lewis rats as recipients. Kidney transplantation was performed 2 weeks after skin transplantation for sensitization. After establishing the acute AMR model in rat kidney transplantation, the recipients of experimental groups included the syngeneic kidney transplantation group (6 rats), the allogeneic kidney transplantation group (6 rats), the syngeneic skin transplantation followed by kidney transplantation group (12 rats), and the allogeneic skin transplantation followed by kidney transplantation group (24 rats). The groups for investigating the preventive effect of DHA on acute AMR included the control group (allogeneic skin transplantation followed by kidney transplantation) and the DHA group (allogeneic skin transplantation followed by kidney transplantation + DHA), with 12 rats in each group. The survival time of recipient rats, serum donor-specific antibody (DSA) levels and graft pathological changes were used to identify the acute AMR model. On this basis, DSA levels, pathological changes in the transplant kidneys and peripheral blood B-cell levels were detected to assess the preventive effect of DHA on acute AMR. Results Compared with the allogeneic kidney transplantation group, skin transplantation sensitization significantly shortened the survival time of recipient rats (P<0.01). Compared with the syngeneic skin transplantation followed by kidney transplantation group, the allogeneic skin transplantation followed by kidney transplantation group showed significantly elevated serum DSA-IgG levels from 7 days after skin transplantation to 5 days after kidney transplantation (P<0.01), and significantly elevated DSA-IgM levels at 7 and 14 days after skin transplantation(all P<0.01). The transplant kidneys in the allogeneic skin transplantation followed by kidney transplantation group showed a small number of inflammatory cell infiltrations, tubular necrosis, capillaritis, and C4d deposition starting from 1 day after kidney transplantation, with these pathological changes worsening as the post-transplantation days increased. The kidney damage became significant starting from 3 days after transplantation. The above pathology manifestations were consistent with the characteristics of acute AMR. On the basis of establishing the acute AMR model, DHA treatment significantly prolonged the survival time of recipient rats (P<0.01) , and reduced serum DSA-IgG and DSA-IgM levels. DHA treatment significantly alleviated the pathological manifestations of acute AMR, including kidney damage, inflammatory cell infiltration, capillaritis and tubular necrosis, and also reduced C4d deposition in the transplant kidneys, inflammatory cell infiltration and peripheral blood CD19⁺ B-cell levels. Conclusions An acute AMR model is established by performing kidney transplantation 2 weeks after allogeneic skin transplantation in rats. It is discovered that DHA has immunosuppressive effects and may effectively prevent acute AMR, which provides a new strategy for the management of clinical AMR.

Objective To analyze the medical resource consumption of healthcare-associated infection(HAI)in patients in different groups of disease diagnosis-related grouping(DRG)based on the DRG payment model,provide reference for optimizing prevention and control of HAI as well as resource management.Methods Medical records and DRG-related indicators of discharged patients from a municipal hospital in Sichuan Province from January 1 to December 31,2024 were analyzed retrospectively.Medical resource consumption of patients in HAI group and non-HAI group was compared.Differences in average length of hospital stay and average expense per hospitalization be-tween two groups of patients were analyzed using stratified analysis.Results In 2024,HAI incidence of discharged patients in DRG management in this hospital was 1.57％.There were statistically significant differences in age,gender,admission and discharge ways between the HAI group and the non-HAI group(all P＜0.05).The main HAI sites were lower respiratory tract,surgical site,urinary tract,and blood.The time consumption index(1.63 vs 0.85),average length of hospital stay(21.00 vs 5.00 days),expense consumption index(1.53 vs 0.92),ave-rage expense per hospitalization(44 700 vs 7 300),and multiple expense in HAI group were all higher than those in non-HAI group(all P＜0.05).The consumption of medical resources for bloodstream infection was relatively higher.Patients with HAI were mostly concentrated in the groups related to acute leukemia with major complications or co-morbidities(MCC),intracranial or craniotomy surgery with MCC,tracheotomy with mechanical ventilation for 96 hours,as well as gastric,esophageal,and duodenal surgery.The average length of hospital stay and average ex-pense per hospitalization of patients in HAI group were both higher than those in the non-HAI group,differences were statistically significant(both P＜0.05).Conclusion HAI significantly increase the consumption of medical resources.Based on DRG analysis,key disease groups for infection prevention and control can be further identified,and the consumption of medical resources can be more accurately and precisely evaluated,thereby optimizing the allocation of medical resources and improving hospital operational efficiency.

Objective To compare the accuracy differences of three-dimensional reconstruction of zygomatic-maxillary complex using cone beam computed tomography(CBCT)and multi-slice spiral computed tomography(MSCT)in the parameters related to zygomatic implantation.Methods Five adult skull specimens(10 zygomatic-maxillary complexes in total)were selected.According to the clinical application characteristics of zygomatic implantation,the parameter location for each skull specimen was conducted,and then CBCT and MSCT were used for scanning and three-dimensional reconstruction.Four parameters were measured on both skull specimens and reconstructed models:the thickness of the maxillary zygomatic process at the intersection of the zygomatic implant path and zygomaticomaxil-lary suture(LB-B1),the thickness of the zygomatic bone at the intersection of the zygomatic implant path and the midpoint of the zygomatic bone surface(LC-C1),the width of the alveolar ridge between the second premolar and the first molar(LAR),and the buccolingual width of the first molar(LM).The specimen group was measured by electronic vernier caliper,and the CBCT group and MSCT group were measured based on CBCT and MSCT.The absolute and relative errors of these four parameters were calculated,and the accuracy of the CBCT and MSCT three-dimensional reconstruction models were analyzed.Results All parameters demonstrated excellent measurement reliability and repeatability,with intra-class correlation coefficient(ICC)≥0.90.The measurement results of each parameter in the CBCT group and the MSCT group showed statistically significant differences compared with those in the specimen group(P<0.05).The measurement result of LM in the CBCT group was smaller than that in the MSCT group,and the difference was statistically significant(P<0.05).However,the measurement results of LB-B1,LC-C1 and LAR in the CBCT group showed no statistically significant difference compared with those in the MSCT group(P>0.05).The mean absolute error and relative error of the LM measurement results to the specimen group in the CBCT group were both smaller than those in the MSCT group,and the differences were statistically significant(P<0.05).There was no statistically significant difference in the mean absolute error or relative error of the LB-B1,LC-C1 and LAR measurement results to the specimen group between the CBCT group and the MSCT group(P>0.05).Conclusion CBCT and MSCT demonstrat comparable accuracy in three-dimensional reconstruction of the zygomatic-maxillary complex.However,CBCT exhibits superior accuracy for fine structures such as dental tissues,which is recommended as the primary choice for imaging data acquisition in zygomatic implantation.