BACKGROUND:Rapid and accurate segmentation of brain tissue in medical images is of great significance for three-dimensional biomechanical modeling and diagnosis of craniocerebral injuries.Currently,artificial intelligence(AI)-based baseline models exhibit excellent generalization capabilities on large-scale datasets.However,due to the specificity and complexity of craniocerebral tissues,these models have certain limitations in their application to craniocerebral tissue segmentation.Additionally,the scarcity of craniocerebral tissue samples makes it difficult for baseline models to achieve precise segmentation results through fine-tuning.OBJECTIVE:To construct a craniocerebral tissue segmentation model based on texture feature retrieval enhancement to improve segmentation accuracy under a small number of samples.METHODS:Segment Anything in Medical Images(MedSAM)model was selected as the basic framework,and texture features were combined with deep learning to build a brain tissue segmentation model based on texture feature retrieval enhancement(DP-MedSAM).Dice Coefficient and mean intersection over union(MIoU)were selected to evaluate the efficiency of image segmentation results.In comparison with the original MedSAM model,the ablation experiment systematically evaluated the influence of key components on the model performance.The sensitivities of MedSAM,the Segment Anything Model(SAM)for medical image segmentation(SAM-Med2D)and DP-MedSAM in the mandible,left optic nerve,and left parotid gland were compared.RESULTS AND CONCLUSION:(1)By verifying the impact of the number of point prompts on segmentation results on the HaN-Seg dataset,the experimental results indicated that the optimal Dice score was achieved with the addition of three points.(2)DP-MedSAM demonstrated performance improvements compared with MedSAM and SAM-Med2D on two datasets(HaN and Public Domain Database for Computational Anatomy).Especially on the Public Domain Database for Computational Anatomy dataset,in terms of the MIoU metric,DP-MedSAM outperformed MedSAM by 6.59％and SAM-Med2D by 37.35％;in terms of the Dice metric,DP-MedSAM outperformed MedSAM and SAM-Med2D by 4.34％and 25.32％,respectively.(3)The ablation experiment results showed that removing the texture feature extraction module in the DP-MedSAM model,relying solely on original image features,led to a significant decrease in results on the test set.Furthermore,removing the vector cache database and its retrieval enhancement function from the model,which deprived the ability of the model to perform similarity retrieval using an external knowledge base,further reduced model performance.(4)Under conditions of limited data resources,the DP-MedSAM model outperformed the other two models in all evaluation metrics.The DP-MedSAM model performed excellently when processing simple and moderately difficult samples,demonstrating a clear advantage over the other two models and indicating good generalization ability.Processing the fine structures of difficult samples placed higher demands on the model's segmentation capabilities.Although the performance of the DP-MedSAM model declined slightly,it still outperformed the other two models.(5)This study proposes an innovative craniocerebral tissue segmentation model,DP-MedSAM,which improves the baseline model's performance in capturing local details and global structural information in medical images by introducing target region texture feature extraction.Through vector similarity retrieval technology,DP-MedSAM can retrieve the feature vector most similar to the current target region from a pre-constructed vector database,providing more precise guiding information for the segmentation process.

BACKGROUND:Rapid and accurate segmentation of brain tissue in medical images is of great significance for three-dimensional biomechanical modeling and diagnosis of craniocerebral injuries.Currently,artificial intelligence(AI)-based baseline models exhibit excellent generalization capabilities on large-scale datasets.However,due to the specificity and complexity of craniocerebral tissues,these models have certain limitations in their application to craniocerebral tissue segmentation.Additionally,the scarcity of craniocerebral tissue samples makes it difficult for baseline models to achieve precise segmentation results through fine-tuning.OBJECTIVE:To construct a craniocerebral tissue segmentation model based on texture feature retrieval enhancement to improve segmentation accuracy under a small number of samples.METHODS:Segment Anything in Medical Images(MedSAM)model was selected as the basic framework,and texture features were combined with deep learning to build a brain tissue segmentation model based on texture feature retrieval enhancement(DP-MedSAM).Dice Coefficient and mean intersection over union(MIoU)were selected to evaluate the efficiency of image segmentation results.In comparison with the original MedSAM model,the ablation experiment systematically evaluated the influence of key components on the model performance.The sensitivities of MedSAM,the Segment Anything Model(SAM)for medical image segmentation(SAM-Med2D)and DP-MedSAM in the mandible,left optic nerve,and left parotid gland were compared.RESULTS AND CONCLUSION:(1)By verifying the impact of the number of point prompts on segmentation results on the HaN-Seg dataset,the experimental results indicated that the optimal Dice score was achieved with the addition of three points.(2)DP-MedSAM demonstrated performance improvements compared with MedSAM and SAM-Med2D on two datasets(HaN and Public Domain Database for Computational Anatomy).Especially on the Public Domain Database for Computational Anatomy dataset,in terms of the MIoU metric,DP-MedSAM outperformed MedSAM by 6.59％and SAM-Med2D by 37.35％;in terms of the Dice metric,DP-MedSAM outperformed MedSAM and SAM-Med2D by 4.34％and 25.32％,respectively.(3)The ablation experiment results showed that removing the texture feature extraction module in the DP-MedSAM model,relying solely on original image features,led to a significant decrease in results on the test set.Furthermore,removing the vector cache database and its retrieval enhancement function from the model,which deprived the ability of the model to perform similarity retrieval using an external knowledge base,further reduced model performance.(4)Under conditions of limited data resources,the DP-MedSAM model outperformed the other two models in all evaluation metrics.The DP-MedSAM model performed excellently when processing simple and moderately difficult samples,demonstrating a clear advantage over the other two models and indicating good generalization ability.Processing the fine structures of difficult samples placed higher demands on the model's segmentation capabilities.Although the performance of the DP-MedSAM model declined slightly,it still outperformed the other two models.(5)This study proposes an innovative craniocerebral tissue segmentation model,DP-MedSAM,which improves the baseline model's performance in capturing local details and global structural information in medical images by introducing target region texture feature extraction.Through vector similarity retrieval technology,DP-MedSAM can retrieve the feature vector most similar to the current target region from a pre-constructed vector database,providing more precise guiding information for the segmentation process.

Objective: To analyze characteristics of hepatitis B infection in HBsAg-/HBV DNA+ blood donors in Taiyuan, so as to provide evidence for adjusting blood screening strategies. Methods: Blood samples of HBsAg-/HBV DNA+ donors were tested using enzyme-linked immunosorbent assay (ELISA), chemiluminescence assay, nucleic acid qualitative test, and nucleic acid quantitative test. Data on HBsAg-/HBV DNA+ donors in Taiyuan region from January 1, 2016 to December 31, 2024 were statistically analyzed to evaluate the detection rate, demographic characteristics, influencing factors of detection rate, nucleic acid quantitative results, and serological patterns of HBsAg-/HBV DNA+ donors. Results: From January 1, 2016 to December 31, 2024, 991 565 donor samples underwent nucleic acid testing in Taiyuan. A total of 309 HBsAg-/HBV DNA+ samples were detected, resulting in an HBsAg-/HBV DNA+ detection rate of 3.12 per 10 000. The detection rate varied significantly across different years (P<0.05). Males had a significantly higher HBsAg-/HBV DNA+ detection rate than females, first-time donors had a higher rate than repeat donors, and whole blood donors had a higher rate than apheresis donors (P<0.05). The detection rate also differed significantly among age groups (P<0.05). Logistic regression analysis showed that gender, age, donation frequency, and donation type were all influencing factors for HBsAg-/HBV DNA+ detection (all P<0.05). The predominant serological patterns among HBsAg-/HBV DNA+ donors were HBsAb+/HBcAb+ (43.69%, 135/309) or HBcAb+ alone (24.27%, 75/309). Viral load was detectable in 53.40% (165/309) of the HBsAg-/HBV DNA+ donors. Among these, 61.21% (101/165) donors had a viral load<20 IU/mL, and 94.55% (156/165) had a viral load<200 IU/mL. Donors with viral load<200 IU/mL primarily exhibited HBsAb+/HBcAb+ (41.67%, 65/156) or HBcAb+alone (36.54%, 57/156) serological patterns. Conclusion: The prevalence of HBsAg-/HBV DNA+ is low among blood donors in Taiyuan. Higher detection rates were observed in the 46-55 years age group, males, first-time donors, and whole blood donors. HBsAg-/HBV DNA+ donors exhibit specific serological patterns and generally have low viral loads, indicating a potential residual transfusion risk. It is recommended to add HBcAb testing, together with high-sensitivity nucleic acid testing technologies and donor follow-up, to ensure blood safety and guide donor reentry.

OBJECTIVE To prepare the β -cyclodextrin （ β -CD） inclusion complex with volatile oil from Qianghuo qushi qingwen granules， and to characterize and quantitatively analyze the inclusion complex. METHODS The comprehensive scores calculated by inclusion rate and inclusion compound yield were used as indicators for screening the inclusion method. The single-factor experiments and Box-Behnken response surface experiments were used to op timize the inclusion conditions， with the above comprehensive score as response value， and taking the ratio of β -CD to volatile oil， inclusion temperature and inclusion time as indexes. The volatile oil inclusion complex of Qianghuo qushi qingwen granules was prepared according to the determined optimal process， followed by validation. Ultraviolet （UV）-visible spectroscopy， thin-layer chromatography （TLC）， and microscopic imaging were also performed. Ultra-high performance liquid chromatography was used to determine the contents of perillaldehyde， pogostone and atractylodin. RESULTS The saturation aqueous solution method was adopted. The optimal inclusion process conditions were as follows： the ratio of β -CD to volatile oil was 7.5∶1， the inclusion temperature was 40 ℃， and the inclusion time was 2.2 h. In three verification experiments， the average inclusion rate was 72.32%， the average yield of inclusion compound was 74.45%， the average comprehensive score was 72.96 points， and the relative error with the predicted value （74.15 points） of the model was 1.61%. UV-visible spectroscopy， TLC and microscopic imaging showed that β -CD and volatile oil successfully formed a new inclusion complex. The average contents of perillaldehyde， pogostone and atractylodin were 4.498 2， 0.814 9， 0.905 7 mg/g， respectively， with RSDs of 0.31%， 0.56% and 0.63% （ n ＝3）. CONCLUSIONS A stable and feasible preparation process of the volatile oil inclusion complex of Qianghuo qushi qingwen granules is successfully established.

With the integration of 5G communication technology and robotic surgical systems, remote robot-assisted thoracic surgery is overcoming geographical barriers, offering an innovative approach to addressing the uneven distribution of medical resources. This study conducted a systematic literature review—using databases such as PubMed and CNKI, with the search period extending up to 2025—incorporating clinical studies, case reports, and review articles to comprehensively evaluate the clinical efficacy and safety of 5G-enabled remote robot-assisted thoracic surgery (5G-RRATS). The analysis also examined current technological limitations and potential future development trajectories. Existing evidence indicates that, given adequate technical support, 5G-RRATS can achieve perioperative outcomes comparable to those of conventional local robotic surgeries across procedures including pulmonary wedge resection, lobectomy, and esophagectomy. Furthermore, it demonstrates potential advantages in minimizing surgical incisions and reducing intraoperative blood loss. Nevertheless, challenges related to network stability, latency control, interdisciplinary collaboration between medical and engineering teams, and legal, regulatory, and ethical considerations continue to hinder widespread clinical adoption. Looking ahead, the emergence of a "one-to-many" remote surgical model, combined with the integration of artificial intelligence and augmented reality technologies, as well as advancements in low-orbit satellite communications, may enable 5G-RRATS to further advance precision and efficiency in thoracic surgery, thereby facilitating equitable access to high-quality care for a broader patient population.

By reviewing the physical biological research on the activation of acupoint effects by acupuncture， this paper explains the activation mechanism from the perspective of the generation and transmission of mechanical signals caused by acupuncture， and reveals the physical-chemical coupling processes in the acupoint microenvironment. Future research should focus on locally mechanosensitive cells， further exploring how acupuncture mechanical signals trigger dynamic changes in cells and molecules in the acupoints， and the physical-chemical information transduction mechanism， which will provide scientific evidence for the acupoint activation during acupuncture. Related studies will contribute to a deeper understanding of the scientific principles behind acupuncture and promote its clinical application and development.

AIM: To investigate the expression of Ficolin-3 and secreted frizzled-related protein 5(SFRP5)in the serum of patients with type 2 diabetes mellitus(T2DM)combined with diabetic retinopathy(DR)and their diagnostic value.METHODS: Prospectively selected patients with T2DM combined with DR admitted to the hospital from May 2023 to May 2025 were divided into non-proliferative and proliferative groups according to the severity of DR. Another patients with T2DM alone during the same period were selected as the T2DM group. ELISA was used to detect Ficolin-3 and SFRP5 levels; Correlation of serum Ficolin-3, SFRP5 levels, and inflammatory markers in T2DM patients with DR were analyzed using Pearson method; Logistic regression was used to analyze related influencing factors; ROC curve analysis was used to evaluate the diagnostic value of serum Ficolin-3 and SFRP5 for DR in T2DM patients.RESULTS: This study included a total of 108 patients with T2DM combined with DR(57 cases in the non-proliferative group, 51 cases in the proliferative group)and 108 cases in the T2DM group. The non-proliferative group had an average age of 59.01±6.28 y, with 34 males and 23 females. The proliferative group had an average age of 59.09±6.35 y, with 30 males and 21 females. The T2DM group had an average age of 58.96±6.18 y, with 62 males and 46 females.The serum levels of Ficolin-3, TNF-α, and IL-6 in the non-proliferative and proliferative groups were higher than those in the T2DM group(all P<0.05), while the level of SFRP5 was lower than that in the T2DM group(all P<0.05). The serum levels of Ficolin-3, TNF-α, and IL-6 in the proliferative group were higher than those in the non-proliferative group(all P<0.05), and the level of SFRP5 was lower than that in the non-proliferative group(P<0.05).Complying with Pearson correlation analysis showed that serum Ficolin-3 was negatively correlated with SFRP5(P<0.05), and both were related to TNF-α and IL-6(all P<0.001). Logistic analysis showed that the course of diabetes, SUA, HbA1c, Ficolin-3, TNF-α, and IL-6 were the risk factors for T2DM patients with DR(all P<0.05), and SFRP5 was a protective factor(P<0.05). Complying with the ROC curve, the AUC values of serum Ficolin-3 and SFRP5 alone and their combination for diagnosing T2DM patients with DR were 0.774, 0.793, and 0.864, respectively. The AUC of combined diagnosis was better than that of single diagnosis(Z=2.694, Z=2.708, both P<0.05).CONCLUSION: In patients with T2DM complicated by DR, serum levels of Ficolin-3 and SFRP5 are abnormally expressed. Both are influencing factors for T2DM with DR, and the combined detection can improve the diagnostic value in these patients.

ObjectiveFunctional near-infrared spectroscopy (fNIRS), a novel non-invasive technique for monitoring cerebral activity, can be integrated with upper limb rehabilitation robots to facilitate the real-time assessment of neurological rehabilitation outcomes. The rehabilitation robot is designed with 3 training modes: passive, active, and resistance. Among these, the resistance mode has been demonstrated to yield superior rehabilitative outcomes for patients with a certain level of muscle strength. The control modes in the resistance mode can be categorized into dynamic and static control. However, the effects of different control modes in the resistance mode on the motor function of patients with upper limb hemiplegia in stroke remain unclear. Furthermore, the effects of force, an important parameter of different control modes, on the activation of brain regions have rarely been reported. This study investigates the effects of dynamic and static resistance modes under varying resistance levels on cerebral functional alterations during motor rehabilitation in post-stroke patients. MethodsA cohort of 20 stroke patients with upper limb dysfunction was enrolled in the study, completing preparatory adaptive training followed by 3 intensity-level tasks across 2 motor paradigms. The bilateral prefrontal cortices (PFC), bilateral primary motor cortices (M1), bilateral primary somatosensory cortices (S1), and bilateral premotor and supplementary motor cortices (PM) were examined in both the resting and motor training states. The lateralization index (LI), phase locking value (PLV), network metrics were employed to examine cortical activation patterns and topological properties of brain connectivity. ResultsThe data indicated that both dynamic and static modes resulted in significantly greater activation of the contralateral M1 area and the ipsilateral PM area when compared to the resting state. The static patterns demonstrated a more pronounced activation in the contralateral M1 in comparison to the dynamic patterns. The results of brain network analysis revealed significant differences between the dynamic and resting states in the contralateral PFC area and contralateral M1 area (F=4.709, P=0.038), as well as in the contralateral PM area and ipsilateral M1 area (F=4.218, P=0.049). Moreover, the findings indicated a positive correlation between the activation of the M1 region and the increase in force in the dynamic mode, which was reversed in the static mode. ConclusionBoth dynamic and static resistance training modes have been demonstrated to activate the corresponding brain functional regions. Dynamic resistance modes elicit greater oxygen changes and connectivity to the region of interest (ROI) than static resistance modes. Furthermore, the effects of increasing force differ between the two modes. In patients who have suffered a stroke, dynamic modes may have a more pronounced effect on the activation of exercise-related functional brain regions.

Age-related macular degeneration(AMD)is a prevalent retinal degenerative disease closely linked to age and stands as a leading cause of central vision loss among the elderly. Under physiological condition, microglia in the retina plays crucial roles in tissue homeostasis, immune surveillance, and tissue repair. However, in pathological state, microglia can be abnormally activated and migrate to AMD lesion sites, which results in exacerbating damage to retinal pigment epithelial cells and photoreceptor cells, thus promoting the progression of AMD. This review focuses on the origins, distribution, and functional changes of microglia under physiological and pathological conditions. Recent advances in microglia-targeted therapies for AMD are also summarized, which provides a theoretical basis for the development of novel treatment strategies.

Objective To investigate the efficacy of leaflet augmentation technique to repair the recurrent mitral valve (MV) regurgitation after mitral repair in children. Methods A retrospective analysis was conducted on the clinical data of children who underwent redo MV repair for recurrent regurgitation after initial MV repair, using a leaflet augmentation technique combined with a standardized repair strategy at Fuwai Hospital, Chinese Academy of Medical Sciences, from 2018 to 2022. The pathological features of the MV, key intraoperative procedures, and short- to mid-term follow-up outcomes were analyzed. Results A total of 24 patients (12 male, 12 female) were included, with a median age of 37.6 (range, 16.5–120.0) months. The mean interval from the initial surgery was (24.9±17.0) months. All children had severe mitral regurgitation preoperatively. The cardiopulmonary bypass time was (150.1±49.5) min, and the aortic cross-clamp time was (94.0±24.2) min. There were no early postoperative deaths. During a mean follow-up of (20.3±9.1) months, 3 (12.5%) patients developed moderate or severe mitral regurgitation (2 severe, 1 moderate). One (4.2%) patient died during follow-up, and one (4.2%) patient underwent a second MV reoperation. The left ventricular end-diastolic diameter was significantly reduced postoperatively compared to preoperatively [ (43.5±8.6) mm vs. (35.8±7.8)mm, P<0.001]. Conclusion The leaflet augmentation technique combined with a standardized repair strategy can achieve satisfactory short- to mid-term outcomes for the redo mitral repair after previous MV repair. It can be considered a safe and feasible technical option for cases with complex valvular lesions and severe pathological changes.