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.

ObjectiveTo reveal the mechanism of Zuogui Jiangtang Jieyu prescription against damage to hippocampal synaptic microenvironment via suppressing glutamate receptor 2 （GluR2）/Parkin signal-mediated mitophagy in rats with diabetes-related depression （DD）. MethodsEighty male SD rats underwent adaptive feeding for 5 days before the study. Ten rats were randomly assigned to the normal group. The model of DD rats was established with the rest by 2-week high-fat diet + streptozotocin （STZ） tail intravenous injection + 28 days of chronic unpredictable mild stress （CUMS） combined with isolation. The rats were randomly divided into a normal group， a model group， a GluR2 blocker group （5 μg·kg^-1）， a GluR2 agonist group （10 μg·kg^-1）， a metformin + fluoxetine group （0.18 g·kg^-1 metformin + 1.8 mg·kg^-1 fluoxetine）， and high- and low-dose Zuogui Jiangtang Jieyu prescription groups （20.52 and 10.26 g·kg^-1， respectively）. The rats in the GluR2 blocker group and the GluR2 agonist group were continuously injected with CNQX and Cl-HIBO in the dentate gyrus of the hippocampus once a week starting from stress modeling， respectively， while the metformin + fluoxetine group and the high- and low-dose Zuogui Jiangtang Jieyu prescription groups were continuously given intragastric administration for 28 d at the same time of stress modeling. Depression-like behavior was evaluated by open field and forced swimming experiments. The levels of serum insulin and adenosine triphosphate （ATP） in hippocampus were detected by biochemical analysis. The levels of 5-hydroxytryptamine （5-HT） and dopamine （DA） in hippocampus were detected by enzyme-linked immunosorbent assay （ELISA）. The autophagosomes of hippocampal neurons were observed by transmission electron microscopy. The morphology and structure of dendrites and spines of hippocampal neurons were evaluated by Golgi staining. Western blot detected the expression levels of GluR2 and Parkin proteins in hippocampus. The expression levels of GluR2， Parkin， regulating synaptic membrane exocytosis protein 3 （RIMS3）， and postsynaptic density protein 95 （PSD95） in the dentate gyrus of the hippocampus were detected by immunofluorescence. ResultsCompared with the normal group， the model group exhibited reduced total activity distance in the open field and increased immobility time in forced swimming （P<0.01）， lowered levels of serum insulin and ATP， 5-HT， and DA in hippocampus （P<0.01）， increased autophagosomes of hippocampal neurons， significantly damaged morphology and structure of dendrites and spines of hippocampal neurons， decreased expression levels of GluR2， RIMS3， and PSD95 in hippocampus， and an increased Parkin expression level （P<0.05， P<0.01）. Compared with the model group， the GluR2 blocker group and the GluR2 agonist group showed aggravation and alleviation of the above abnormal changes， respectively （P<0.05， P<0.01）. The above depression-like behavior was significantly improved in the high- and low-dose Zuogui Jiangtang Jieyu prescription groups to different degrees. Specifically， the two groups saw elevated levels of serum insulin and ATP， 5-HT， and DA in hippocampus （P<0.05， P<0.01）， restrained increase in autophagosomes and damage to morphology and structure of dendrites and spines of hippocampal neurons， up-regulated protein expression levels of GluR2， RIMS3， and PSD95， and down-regulated Parkin expression level （P<0.05， P<0.01）. ConclusionZuogui Jiangtong Jieyu prescription can ameliorate the mitophagy-mediated damage to hippocampal synaptic microenvironment in DD rats， the mechanism of which might be related to the regulation of GluR2/Parkin signaling pathway.

Objective To explore the mediating role of mindful attention and awareness in depressive symptoms and insomnia severity among patients with comorbid depression and insomnia.Methods A cross-sectional study was conducted,enrolling 267 patients with comorbid depression and insomnia who were treated in the outpatient Department of Medical Psychology of Second Affiliated Hospital of Army Medical University,from March to May 2024.Basic demographic and clinical data were collected using a general information questionnaire.Depressive symptom severity was measured via the Patient Health Questionnaire-9(PHQ-9),insomnia severity via the Insomnia Severity Index(ISI),perceived stress via the Perceived Stress Scale-10(PSS-10),and mindful attention and awareness via the Mindful Attention Awareness Scale(MAAS).Pearson correlation analysis was used to examine the correlations between depressive severity,insomnia severity,perceived stress,and mindful attention and awareness.Mediation analysis was performed using Process 4.1.Results The PHQ-9 score was(13.80±5.98)and the ISI score was(17.10±5.56)in the 267 patients.Pearson correlation analysis showed that depressive severity and insomnia severity were positively correlated with perceived stress(r=0.531,0.351,P<0.001)and negatively correlated with mindful attention and awareness(r=-0.373,-0.350,P<0.001).Mediation analysis using Process 4.1 indicated that the combined mediating effect of mindful attention and awareness and insomnia between perceived stress level and depressive level was 0.157,with a 95%confidence interval(CI)of 0.102～0.217,and the total mediating effect was significant(P<0.001).Conclusion Perceived stress directly positively affects depression and indirectly exacerbates depression through insomnia as a mediator,and mindful attention and awareness can weaken the promoting effect of perceived stress on insomnia.

Purpose: Diagnosing sarcopenia necessitates the measurement of skeletal muscle mass. However, guidelines lack a standardized imaging modality with thresholds validated among Asians. This systematic review compared ultrasonography, computed tomography (CT), magnetic resonance imaging (MRI), and bioelectrical impedance analysis (BIA)/body composition monitoring in the detection of sarcopenia within Asian populations. Methods: PubMed and Embase were systematically searched for studies analyzing ultrasonography, CT, MRI, and BIA in diagnosing sarcopenia among Asians. Study quality was assessed using the Newcastle-Ottawa scale. Results: Pooled findings from 21,598 patients across 25 studies were examined. In receiver operating characteristic analysis, ultrasound displayed a pooled mean area under the curve (AUC) of 0.767 (95% confidence interval [CI], 0.709–0.806), with mean sensitivity of 81.1% (95% CI, 0.744–0.846) and specificity of 73.1% (95% CI, 0.648–0.774), for detecting sarcopenia in Asian populations. CT exhibited an AUC of 0.720 (sensitivity, 54.0%; specificity, 92.0%). MRI demonstrated an AUC of 0.839 (sensitivity, 67.0%; specificity, 66.0%). BIA displayed an AUC of 0.905 (95% CI, 0.842–0.968), 80.7% sensitivity (95% CI, 0.129–0.679), and 82.4% specificity (95% CI, 0.191–0.633). Conclusion Various modalities aid in diagnosing sarcopenia, and selection should be individualized. Although only BIA and dual-energy x-ray absorptiometry are recommended by the Asian Working Group for Sarcopenia and the European Working Group on Sarcopenia in Older People, ultrasound imaging may hold diagnostic value for sarcopenia in the Asian population. In certain groups, diagnostic use of CT and MRI is warranted. Future research can standardize and validate modality-specific thresholds and protocols within Asian populations.

Laparoscopic subtotal cholecystectomy (LSC) has been a safe and viable alternative to conversion to laparotomy in cases of severe cholecystitis. The objective of this study is to determine the utility of intraoperative choledochoscopy in LSC for the exploration of the gallbladder, cyst duct, and subsequent stone clearance of the cystic duct in cases of severe cholecystitis. A total of 72 patients diagnosed with severe cholecystitis received choledochoscopy-assisted laparoscopic subtotal cholecystectomy (CALSC). A choledochoscopy was performed to explore the gallbladder cavity and/or cystic duct, and to extract stones using a range of techniques. The clinical records, including the operative records and outcomes, were subjected to analysis. No LSC was converted to open surgery, and no bile duct or vascular injuries were sustained. All stones within the cystic duct were removed by a combination of techniques, including high-frequency needle knife electrotomy, basket, and electrohydraulic lithotripsy. A follow-up examination revealed the absence of residual bile duct stones, with the exception of one common bile duct stone, which was extracted via endoscopic retrograde cholangiopancreatography. In certain special cases, CALSC may prove to be an efficacious treatment for the management of severe cholecystitis. This technique allows for optimal comprehension of the situation within the gallbladder cavity and cystic duct, facilitating the removal of stones from the cystic duct and reducing the residue of the non-functional gallbladder remnant.

ObjectiveMagnetoencephalography (MEG), a non-invasive neuroimaging technique, meticulously captures the magnetic fields emanating from brain electrical activity. Compared with MEG based on superconducting quantum interference devices (SQUID), MEG based on optically pump magnetometer (OPM) has the advantages of higher sensitivity, better spatial resolution and lower cost. However, most of the current studies are clinical studies, and there is a lack of animal studies on MEG based on OPM technology. Pain, a multifaceted sensory and emotional phenomenon, induces intricate alterations in brain activity, exhibiting notable sex differences. Despite clinical revelations of pain-related neuronal activity through MEG, specific properties remain elusive, and comprehensive laboratory studies on pain-associated brain activity alterations are lacking. The aim of this study was to investigate the effects of inflammatory pain (induced by Complete Freund’s Adjuvant (CFA)) on brain activity in a rat model using the MEG technique, to analysis changes in brain activity during pain perception, and to explore sex differences in pain-related MEG signaling. MethodsThis study utilized adult male and female Sprague-Dawley rats. Inflammatory pain was induced via intraplantar injection of CFA (100 μl, 50% in saline) in the left hind paw, with control groups receiving saline. Pain behavior was assessed using von Frey filaments at baseline and 1 h post-injection. For MEG recording, anesthetized rats had an OPM positioned on their head within a magnetic shield, undergoing two 15-minute sessions: a 5-minute baseline followed by a 10-minute mechanical stimulation phase. Data analysis included artifact removal and time-frequency analysis of spontaneous brain activity using accumulated spectrograms, generating spectrograms focused on the 4-30 Hz frequency range. ResultsMEG recordings in anesthetized rats during resting states and hind paw mechanical stimulation were compared, before and after saline/CFA injections. Mechanical stimulation elevated alpha activity in both male and female rats pre- and post-saline/CFA injections. Saline/CFA injections augmented average power in both sexes compared to pre-injection states. Remarkably, female rats exhibited higher average spectral power 1 h after CFA injection than after saline injection during resting states. Furthermore, despite comparable pain thresholds measured by classical pain behavioral tests post-CFA treatment, female rats displayed higher average power than males in the resting state after CFA injection. ConclusionThese results imply an enhanced perception of inflammatory pain in female rats compared to their male counterparts. Our study exhibits sex differences in alpha activities following CFA injection, highlighting heightened brain alpha activity in female rats during acute inflammatory pain in the resting state. Our study provides a method for OPM-based MEG recordings to be used to study brain activity in anaesthetized animals. In addition, the findings of this study contribute to a deeper understanding of pain-related neural activity and pain sex differences.

Asthma is a chronic inflammatory disease involving multiple inflammatory cells and cytokines. Its pathogenesis is complex, involving various cells and cytokines. Traditional Chinese medicine(TCM) theory suggests that the pathogenesis of asthma is closely related to the dysfunction of internal organs such as the lungs, spleen, and kidneys. In contrast, modern immunological studies have revealed the central role of T helper 1(Th1)/T helper 2(Th2) and T helper 17(Th17)/regulatory T(Treg) cellular immune imbalance in the pathogenesis of asthma. Th1/Th2 imbalance is manifested as hyperfunction of Th2 cells, which promotes the synthesis of immunoglobulin E(IgE) and the activation of eosinophil granulocytes, leading to airway hyperresponsiveness and inflammation.Meanwhile, Th17/Treg imbalance exacerbates the inflammatory response in the airways, further contributing to asthma pathology.Currently, therapeutic strategies for asthma are actively exploring potential targets for regulating the balance of Th1/Th2 and Th17/Treg immune responses. These targets include cytokines, transcription factors, key proteins, and non-coding RNAs. Precisely regulating the expression and function of these targets can effectively modulate the activation and differentiation of immune cells. In recent years,traditional Chinese medicine active ingredients have shown unique potential and prospects in the field of asthma treatment. Based on this, the present study systematically summarizes the efficacy and specific mechanisms of TCM active ingredients in treating asthma by regulating Th1/Th2 and Th17/Treg immune balance through literature review and analysis. These active ingredients, including flavonoids, terpenoids, polysaccharides, alkaloids, and phenolic acids, exert their effects through various mechanisms, such as inhibiting the activation of inflammatory cells, reducing the release of cytokines, and promoting the normal differentiation of immune cells. This study aims to provide a solid foundation for the widespread application and in-depth development of TCM in asthma treatment and to offer new ideas for clinical research and drug development of asthma.
Asthma/genetics* ; Humans ; Drugs, Chinese Herbal/chemistry* ; Th2 Cells/drug effects* ; Th17 Cells/drug effects* ; T-Lymphocytes, Regulatory/drug effects* ; Th1 Cells/drug effects* ; Animals ; Cytokines/immunology* ; Medicine, Chinese Traditional

This study aims to investigate the mechanism of Qingrun Decoction in alleviating hepatic insulin resistance in type 2 diabetes mellitus(T2DM) rats through the reprogramming of amino acid metabolism. A T2DM rat model was established by inducing insulin resistance through a high-fat diet combined with intraperitoneal injection of streptozotocin. The model rats were randomly divided into five groups: model group, high-, medium-, and low-dose Qingrun Decoction groups, and metformin group. A normal control group was also established. The rats in the normal and model groups received 10 mL·kg~(-1) distilled water daily by gavage. The metformin group received 150 mg·kg~(-1) metformin suspension by gavage, and the Qingrun Decoction groups received 11.2, 5.6, and 2.8 g·kg~(-1) Qingrun Decoction by gavage for 8 weeks. Blood lipid levels were measured in different groups of rats. Pathological damage in rat liver tissue was assessed by hematoxylin-eosin(HE) staining and oil red O staining. Transcriptome sequencing and untargeted metabolomics were performed on rat liver and serum samples, integrated with bioinformatics analyses. Key metabolites(branched-chain amino acids, BCAAs), amino acid transporters, amino acid metabolites, critical enzymes for amino acid metabolism, resistin, adiponectin(ADPN), and mammalian target of rapamycin(mTOR) pathway-related molecules were quantified using quantitative real-time polymerase chain reaction(qRT-PCR), Western blot, and enzyme-linked immunosorbent assay(ELISA). The results showed that compared with the normal group, the model group had significantly increased serum levels of total cholesterol(TC), triglycerides(TG), low-density lipoprotein cholesterol(LDL-C), and resistin and significantly decreased ADPN levels. Hepatocytes in the model group exhibited loose arrangement, significant lipid accumulation, fatty degeneration, and pronounced inflammatory cell infiltration. In liver tissue, the mRNA transcriptional levels of solute carrier family 7 member 2(Slc7a2), solute carrier family 38 member 2(Slc38a2), solute carrier family 38 member 4(Slc38a4), and arginase(ARG) were significantly downregulated, while the mRNA transcriptional levels of solute carrier family 1 member 4(Slc1a4), solute carrier family 16 member 1(Slc16a1), and methionine adenosyltransferase(MAT) were upregulated. Furthermore, the mRNA transcription and protein expression levels of branched-chain α-keto acid dehydrogenase E1α(BCKDHA) and DEP domain-containing mTOR-interacting protein(DEPTOR) were downregulated, while mRNA transcription and protein expression levels of mTOR, as well as ribosomal protein S6 kinase 1(S6K1), were upregulated. The levels of BCAAs and S-adenosyl-L-methionine(SAM) were elevated. The serum level of 6-hydroxymelatonin was significantly reduced, while imidazole-4-one-5-propionic acid and N-(5-phospho-D-ribosyl)anthranilic acid levels were significantly increased. Compared with the model group, Qingrun Decoction significantly reduced blood lipid and resistin levels while increasing ADPN levels. Hepatocytes had improved morphology with reduced inflammatory cells, and fatty degeneration and lipid deposition were alleviated. Differentially expressed genes and differential metabolites were mainly enriched in amino acid metabolic pathways. The expression levels of Slc7a2, Slc38a2, Slc38a4, and ARG in the liver tissue were significantly upregulated, while Slc1a4, Slc16a1, and MAT expression levels were significantly downregulated. BCKDHA and DEPTOR expression levels were upregulated, while mTOR and S6K1 expression levels were downregulated. Additionally, the levels of BCAAs and SAM were significantly decreased. The serum level of 6-hydroxymelatonin was increased, while those of imidazole-4-one-5-propionic acid and N-(5-phospho-D-ribosyl)anthranilic acid were decreased. In summary, Qingrun Decoction may improve amino acid metabolism reprogramming, inhibit mTOR pathway activation, alleviate insulin resistance in the liver, and mitigate pathological damage of liver tissue in T2DM rats by downregulating hepatic BCAAs and SAM and regulating key enzymes involved in amino acid metabolism, such as BCKDHA, ARG, and MAT, as well as amino acid metabolites and transporters.
Animals ; Drugs, Chinese Herbal/administration & dosage* ; Rats ; Insulin Resistance ; Diabetes Mellitus, Type 2/genetics* ; Male ; Liver/drug effects* ; Amino Acids/metabolism* ; Rats, Sprague-Dawley ; Humans ; Metabolic Reprogramming

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.