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 explore the strategy of blood management in patients with massive transfusion in the frigid plateau region. [Methods] The treatment process of a patient with liver rupture in the frigid plateau region was analyzed, and the blood management strategy of the frigid plateau region was discussed in combination with the difficulties of blood transfusion and literature review. [Results] The preoperative complete blood count (CBC) test results of the patient were as follows: RBC 3.14×10¹²/L, Hb 10⁶ g/L, HCT 30.40%, PLT 115.00×10⁹/L; coagulation function: PT 18.9 s, FiB 1.31 g/L, DD > 6 μg/mL, FDP 25.86 μg/mL; ultrasound examination and imaging manifestations suggested liver contusion and laceration / intraparenchymal hematoma, splenic contusion and laceration, and massive blood accumulation in the abdominal cavity; it was estimated that the patient's blood loss was ≥ 2 000 mL, and massive blood transfusion was required during the operation; red blood cell components were timely transfused during the operation, and the blood component transfusion was guided according to the patient's CBC and coagulation function test results, providing strong support and guarantee for the successful treatment of the patient. The patient recovered well after the operation, and the CBC test results were as follows: RBC 4.32×10¹²/L, Hb 144 g/L, HCT 39.50%, PLT 329.00×10⁹/L; coagulation function: APTT 29.3 s, PT 12.1 s, FiB 2.728 g/L, DD>6 μg/mL, FDP 25.86 μg/mL. The patient was discharged after 20 days, and regular follow-up reexamination showed no abnormal results. [Conclusion] Individualized blood management strategy should comprehensively consider the patient’s clinical symptoms, the degree of hemoglobin decline, dynamic coagulation test results and existing treatment conditions. Efficient and reasonable patient blood management strategies can effectively improve the clinical outcomes of massive transfusion patients in the frigid plateau region.

OBJECTIVE: To propose a machine learning-based method for predicting the trajectories during manual acupuncture manipulation (MAM), aiming to improve the precision and consistency of acupuncture practitioner' operation and provide the real-time suggestions on MAM error correction. METHODS: Computer vision technology was used to analyze the hand micromotion when holding needle during acupuncture, and provide a three-dimensional coordinate description method of the index finger joints of the holding hand. Focusing on the 4 typical motions of MAM, a machine learning-based MAM trajectory predictive model was designed. By integrating the changes of phalangeal joint angle and hand skeletal information of acupuncture practitioner, the motion trajectory of the index finger joint was predicted accurately. Besides, the roles of machine learning-based MAM trajectory predictive model in the skill transmission of acupuncture manipulation were verified by stratified randomized controlled trial. RESULTS: The performance of MAM trajectory predictive model, based on the long short-term memory network (LSTM), obtained the highest stability and precision, up to 98%. The learning effect was improved when the model applied to the skill transmission of acupuncture manipulation. CONCLUSION The machine learning-based MAM predictive model provides acupuncture practitioner with precise action prediction and feedback. It is valuable and significant for the inheritance and error correction of manual operation of acupuncture.
Humans ; Acupuncture Therapy/instrumentation* ; Machine Learning ; Adult ; Male ; Female

OBJECTIVE: To investigate the effect of manual acupuncture manipulations (MAMs) on subcutaneous muscle tissue, by developing quantitative models of "lifting and thrusting" and "twisting and rotating", based on machine learning techniques. METHODS: A depth camera was used to capture the acupuncture operator's hand movements during "lifting and thrusting" and "twisting and rotating" of needle. Simultaneously, the ultrasound imaging was employed to record the muscle tissue responses of the participants. Amplitude and angular features were extracted from the movement data of operators, and muscle fascicle slope features were derived from the data of ultrasound images. The dynamic time warping barycenter averaging algorithm was adopted to align the dual-source data. Various machine learning techniques were applied to build quantitative models, and the performance of each model was compared. The most optimal model was further analyzed for its interpretability. RESULTS: Among the quantitative models built for the two types of MAMs, the random forest model demonstrated the best performance. For the quantitative model of the "lifting and thrusting" technique, the coefficient of determination (R²) was 0.825. For the "twisting and rotating" technique, R² reached 0.872. CONCLUSION Machine learning can be used to effectively develop the models and quantify the effects of MAMs on subcutaneous muscle tissue. It provides a new perspective to understand the mechanism of acupuncture therapy and lays a foundation for optimizing acupuncture technology and designing personalized treatment regimen in the future.
Humans ; Acupuncture Therapy/methods* ; Machine Learning ; Male ; Adult ; Female ; Subcutaneous Tissue/diagnostic imaging* ; Young Adult

The phase changes and quantity-quality transfer of 17 batches of Ostreae Concha(Ostrea rivularis) during the raw material-calcined decoction pieces-standard decoction process were analyzed. The content of calcium carbonate(CaCO_3), the main component, was determined by chemical titration, and the extract yield and transfer rate were calculated. The CaCO_3 content in the raw material, calcined decoction pieces, and standard decoction was 94.39%-98.80%, 95.03%-99.22%, and 84.58%-90.47%, respectively. The process of raw material to calcined decoction pieces showed the yield range of 96.85% to 98.55% and the CaCO_3 transfer rate range of 96.92% to 99.27%. The process of calcined decoction pieces to standard decoction showed the extract yield range of 2.86% to 5.48% and the CaCO_3 transfer rate range of 2.59% to 5.13%. The results of X-ray fluorescence(XRF) assay showed that the raw material, calcined decoction pieces, and standard decoction mainly contained Ca, Na, Mg, Si, Br, Cl, Al, Fe, Cr, Mn, and K. The chemometric results showed an increase in the relative content of Cr, Fe, and Si from raw material to calcined decoction pieces and an increase in the relative content of Mg, Al, Br, K, Cl, and Na from calcined decoction pieces to standard decoction. X-ray diffraction(XRD) was employed to establish XRD characteristic patterns of the raw material, calcined decoction pieces, and standard decoction. The XRD results showed that the main phase of all three was calcite, and no transformation of crystalline form or generation of new phase was observed. Fourier transform infrared spectroscopy(FTIR) was employed to establish the FTIR characteristic spectra of the raw material, calcined decoction pieces, and standard decoction. The FTIR results showed that the raw material had internal vibrations of O-H, C-H, C=O, C-O, and CO■ groups. Due to the loss of organic matter components after calcination, no information about the vibrations of C-H, C=O, and C-O groups was observed in the spectra of calcined decoction pieces and standard decoction. In summary, this study elucidated the quantity-quality transfer and phase changes in the raw material-calcined decoction pieces-standard decoction process by determining the CaCO_3 content, calculating the extract yield and transfer rate, and comparing the element changes, FTIR characteristic spectra, and XRD characteristic pattern. The results were reasonable and reliable, laying a foundation for the subsequent process research and quality control of the formula granules of calcined Ostreae Concha(O. rivularis Gould), and providing ideas and methods for the quality control of the whole process of raw material-decoction pieces-standard decoction-formula granules of Ostreae Concha and other testacean traditional Chinese medicine.
Drugs, Chinese Herbal/isolation & purification* ; Calcium Carbonate/analysis* ; Quality Control

Objective:To explore the effect of bumetanide(BUM)on learning and memory in a rat model of hydro-cephalus induced by kaolin.Methods:A rat model of hydrocephalus was established by injecting kaolin into the lateral ventricle.Bumetanide was administered intranasally.The volume of the lateral ventricle was measured by magnetic resonance imaging,and changes in intracranial pressure were detected.The learning and memory ability of rats was evaluated by the novel object recognition test,and neuronal damage in the hippocampus was detected by Nissl staining.Results:Compared with the control group,the lateral ventricle expansion was more obvious,intracranial pressure was elevated,and the exploring novel object time was significantly decreased in the bumetanide treatment group.Nissl stai-ning showed neuronal damage.After bumetanide treatment,the volume of the lateral ventricle decreased,intracranial pressure decreased,the exploring novel object time increased,and neuronal damage was alleviated.Conclusion:Bu-metanide can alleviate kaolin-induced hydrocephalus in rats and improve learning and memory ability.

Objective:To investigate the role and molecular mechanisms of transcription factor EB (TFEB) and its target genes in the treatment of multiple myeloma (MM) with bortezomib.Methods:TFEB target genes were predicted using the GTRD database (http://gtrd.biouml.org/), identifying Ptch1 gene for further study. Expression changes of Ptch1 in RPMI8226 and U266 MM cell lines after bortezomib treatment were assessed by real time fluorogenic quantitative PCR (RT-qPCR) and Western blot. RPMI8226 and U266 cell lines were transfected with siRNA-TFEB, and mRNA and protein levels of key factors (Ptch1, Gli1) in the Ptch1/Hedgehog signaling pathway were measured by RT-qPCR and Western blot. Furthermore, Ptch1 was overexpressed in MM cell lines via lentiviral transduction. Autophagy was evaluated by acridine orange staining, and protein levels of LC3B, Beclin-1, and Lamp-1 were measured by Western blot. Lysosomal quantity changes were assessed by lysosomal fluorescent probes.Results:Bortezomib (6.0×10 -6 mmol/L, 24 h) significantly reduced Ptch1 mRNA and protein levels in both cell lines compared with blank control group (all P<0.05). siRNA-TFEB transfection reversed bortezomib’s inhibition of Hedgehog pathway key factors Ptch1 and Gli. Ptch1 overexpression in bortezomib-treated RPMI8226 and U266 cells significantly reduced the relative expression of autophagy-related proteins LC3B, Beclin-1, and Lamp-1 (all P=0.001). Acridine orange staining showed fewer acidic vesicular organelles within two cell lines (all P=0.001). The relative fluorescence expressions of lysosomal probes reflecting the number of lysosomes were also decreased ( P values of RPMI8226 and U266 cell lines were 0.001 and 0.007, respectively) . Conclusion:The knockdown of TFEB can specifically promote the expression of the Ptch1/Hedgehog signaling pathway, thereby reducing bortezomib-induced autophagy in MM cells and reversing the inhibitory effect of bortezomib on the proliferation of MM cell lines.