Objective This study aimed to explore the ability of ultrasonographic radiomics in predicting the occurrence of in-stent restenosis(ISR)after carotid artery stenting(CAS)by analyzing the correlation between radiomic features of responsible plaques in carotid artery stenosis and the incidence of ISR.Methods A retrospective collection was conducted on 206 cases that underwent CAS treatment at our hospital.The enrolled patients were randomly split into a training set(144 cases)and a test set(62 cases)at a 7∶3 ratio.We utilized the Darwin Intelligent Research Platform to extract radiomic features from each region of interest,and then screened 1125 ultrasonographic radiomic features.Different machine learning algorithms were employed to construct diagnostic models,and the best-performing classifier was selected.Various prediction models were established,including a clinical-ultrasonographic feature model,a radiomic model,and a combined clinical-ultrasonographic-radiomic model.Results Multivariate logistic regression analysis in the training set revealed that hypertension,hyperuricemia,triglycerides,and plaque location were independent risk factors for ISR after CAS.For the clinical-ultrasonographic model,the area under the curve(AUC)values for the training and validation sets were 0.896 and 0.644,respectively.The corresponding AUC values for the radiomic model were 0.961 and 0.715,while those for the combined model were 0.947 and 0.727.Conclusion The radiomic model demonstrates superior performance in predicting ISR compared to the traditional clinical-ultrasonographic model.The combined model exhibited an enhanced ability to predict ISR occurrence,thereby improving the diagnostic performance of traditional assessments.

Objective This study aimed to explore the ability of ultrasonographic radiomics in predicting the occurrence of in-stent restenosis(ISR)after carotid artery stenting(CAS)by analyzing the correlation between radiomic features of responsible plaques in carotid artery stenosis and the incidence of ISR.Methods A retrospective collection was conducted on 206 cases that underwent CAS treatment at our hospital.The enrolled patients were randomly split into a training set(144 cases)and a test set(62 cases)at a 7∶3 ratio.We utilized the Darwin Intelligent Research Platform to extract radiomic features from each region of interest,and then screened 1125 ultrasonographic radiomic features.Different machine learning algorithms were employed to construct diagnostic models,and the best-performing classifier was selected.Various prediction models were established,including a clinical-ultrasonographic feature model,a radiomic model,and a combined clinical-ultrasonographic-radiomic model.Results Multivariate logistic regression analysis in the training set revealed that hypertension,hyperuricemia,triglycerides,and plaque location were independent risk factors for ISR after CAS.For the clinical-ultrasonographic model,the area under the curve(AUC)values for the training and validation sets were 0.896 and 0.644,respectively.The corresponding AUC values for the radiomic model were 0.961 and 0.715,while those for the combined model were 0.947 and 0.727.Conclusion The radiomic model demonstrates superior performance in predicting ISR compared to the traditional clinical-ultrasonographic model.The combined model exhibited an enhanced ability to predict ISR occurrence,thereby improving the diagnostic performance of traditional assessments.

OBJECTIVE To establish the fingerprint of Gerbera delavayi and the methods for the content determination of 11 components in G. delavayi. METHODS High-performance liquid chromatography（HPLC）was adopted to establish the fingerprints of 13 batches of G. delavayi（No. S1-S13）， and the similarities were evaluated according to Similarity Evaluation System of Chromatographic Fingerprint of TCM （2012 edition）， while the common peaks were identified. Hierarchical clustering analysis （HCA）， principal component analysis （PCA） and orthogonal partial least square-discriminant analysis （OPLS-DA） were carried out by using SPSS 25.0 software and SIMCA 14.1 software. The contents of neochlorogenic acid， chlorogenic acid， cryptochlorogenic acid， 3，8-dihydroxy-4-methoxy-2-oxo-2H-1-benzopyran-5-carboxylic acid， caffeic acid， 3-hydroxy-4-methoxy-2- oxo-2H-1-benzopyran- 5-carboxylic acid， luteolin-7-O-β-D-glucoside， isochlorogenic acid A， apigenin-7-O-β-D-glucoside， isochlorogenic acid C and xanthotoxin were determined by HPLC. RESULTS The similarities in HPLC fingerprint of 13 batches of G. delavayi were 0.801-0.994； a total of 38 common peaks were identified and 13 common peaks were identified. The results of HCA showed that S1-S5 and S7 were clustered into one group， S6 into one category， S8 into one category， S9 and S11 into one category， S10， S12 and S13 into one category， and the results of PCA were consistent with them. The results of OPLS-DA showed that variable importance values for the projection of peak 7 （chlorogenic acid）， peak 21 （isochlorogenic acid A）， peak 26 （xanthotoxin）， peak 19 （isochlorogenic acid B）， peak 33， peak 13， peak 23 （isochlorogenic acid C）， peak 2 （new chlorogenic acid）， peak 17 （luteolin-7-O-β-D- glucoside） were greater than 1. The above 11 components had good linearity in their respective detection concentration ranges （r was greater than 0.999）. RSDs of precision， repeatability， and stability tests were not more than 2% （n＝6）. The average recovery rates were 92.54%-105.55%， and the RSDs were 0.83%-1.93% （n＝6）. The average contents of 11 components were 0.744， 5.014， 0.646， 0.431， 0.069， 0.582， 0.979， 2.754， 0.157， 1.284 and 2.943 mg/g， respectively. CONCLUSIONS The constructed HPLC fingerprint and content determination methods are simple， accurate and stable， which can provide reference for quality control of G. delavayi. Xanthotoxin， chlorogenic acid， isochlorogenic acid A， luteolin-7-O- β -D-glucoside， isochlorogenic acid C and new chlorogenic acid can be used as markers for G. delavayi.

Lung transplantation is a key therapeutic approach for patients with end-stage lung diseases. Although its clinical outcomes have significantly improved, multidimensional injuries sustained by donor lungs during procurement, preservation, and transplantation remain major challenges affecting graft survival and long-term prognosis. This article proposes the "Guangzhou Classification" for full-course management of donor lung injury, characterized by spatiotemporal dynamics. Based on the progression of disease stages, donor lung injuries are systematically divided into three types: primary injuries (including donor ICU-related lung injury, pathogen colonization, and cold ischemia injury), secondary injuries (such as ventilator-induced lung injury after transplantation, ischemia-reperfusion inflammatory storm, and early rejection), and accompanying injuries (organ toxicity caused by accumulation of postoperative sedatives, analgesics, and vasoactive drugs). Drawing on previous studies and the clinical experience of our center, this paper elaborates the temporal evolution, key risk factors, and prevention and treatment strategies of each injury category, and discusses future research directions. By targeting critical injury factors at each stage, this classification aims to optimize both short-term and long-term outcomes of lung transplantation.

In recent years, lung function assessment has attracted increasing attention in the perioperative management of thoracic surgery. However, traditional pulmonary function testing methods remain limited in clinical practice due to high equipment requirements and complex procedures. With the rapid development of artificial intelligence (AI) technology, lung function assessment based on multimodal chest imaging (such as X-rays, CT, and MRI) has become a new research focus. Through deep learning algorithms, AI models can accurately extract imaging features of patients and have made significant progress in quantitative analysis of pulmonary ventilation, evaluation of diffusion capacity, measurement of lung volumes, and prediction of lung function decline. Previous studies have demonstrated that AI models perform well in predicting key indicators such as forced expiratory volume in one second, diffusing capacity for carbon monoxide, and total lung capacity. Despite these promising prospects, challenges remain in clinical translation, including insufficient data standardization, limited model interpretability, and the lack of prediction models for postoperative complications. In the future, greater emphasis should be placed on multicenter collaboration, the construction of high-quality databases, the promotion of multimodal data integration, and clinical validation to further enhance the application value of AI technology in precision decision-making for thoracic surgery.

Objective To construct a combined model integrating ultrasonic features and radiomics derived from ultrasound images,and to evaluate its diagnostic performance in re-assessing the benign or malignant nature of Bethesda Ⅲ nodules.Methods A retrospective study was carried out on 442 patients with thyroid nodules classified as Bethesda Ⅲ after fine-needle aspiration biopsy(FNAB)between January 2019 and September 2024.All patients had undergone surgical pathology.The patients were randomly allocated into a training set and a testing set at a ratio of 7∶3.Relevant clinical characteristics were gathered,and regions of interest(ROI)were outlined on the most suspicious slice of the lesion prior to biopsy.Ultrasound radiomics features were extracted,key radiomics features were selected,and radiomics scores(Rad-score)were computed.The ultrasound model,radiomics model,and combined model were constructed.Subsequently,the diagnostic efficacy and clinical application value of each model were evaluated using the area under the receiver operating characteristic curve(AUC)and decision curve analysis(DCA).Results Univariate analysis and multivariate logistic regression analysis findings indicated that microcalci-fication,irregular margin,and Rad-score were independent risk factors for the malignant transformation of BethesdaⅢ nodules.In the testing set,the AUC values of the ultrasound model,radiomics model,and combined model were 0.76,0.71,and 0.81,respectively.The calibration curve of the combined model revealed a good consistency between the predicted values and the actual outcomes.The DCA of the testing set demonstrated that the combined model exhibited high clinical utility.Conclusion The combined model,established based on ultrasonic features and ultrasound radiomics,provides a higher predictive value for evaluating the malignancy risk of Bethesda Ⅲ nodules.

Lung transplantation is a key therapeutic approach for patients with end-stage lung diseases. Although its clinical outcomes have significantly improved, multidimensional injuries sustained by donor lungs during procurement, preservation, and transplantation remain major challenges affecting graft survival and long-term prognosis. This article proposes the "Guangzhou Classification" for full-course management of donor lung injury, characterized by spatiotemporal dynamics. Based on the progression of disease stages, donor lung injuries are systematically divided into three types: primary injuries (including donor ICU-related lung injury, pathogen colonization, and cold ischemia injury), secondary injuries (such as ventilator-induced lung injury after transplantation, ischemia-reperfusion inflammatory storm, and early rejection), and accompanying injuries (organ toxicity caused by accumulation of postoperative sedatives, analgesics, and vasoactive drugs). Drawing on previous studies and the clinical experience of our center, this paper elaborates the temporal evolution, key risk factors, and prevention and treatment strategies of each injury category, and discusses future research directions. By targeting critical injury factors at each stage, this classification aims to optimize both short-term and long-term outcomes of lung transplantation.

Objective To construct a combined model integrating ultrasonic features and radiomics derived from ultrasound images,and to evaluate its diagnostic performance in re-assessing the benign or malignant nature of Bethesda Ⅲ nodules.Methods A retrospective study was carried out on 442 patients with thyroid nodules classified as Bethesda Ⅲ after fine-needle aspiration biopsy(FNAB)between January 2019 and September 2024.All patients had undergone surgical pathology.The patients were randomly allocated into a training set and a testing set at a ratio of 7∶3.Relevant clinical characteristics were gathered,and regions of interest(ROI)were outlined on the most suspicious slice of the lesion prior to biopsy.Ultrasound radiomics features were extracted,key radiomics features were selected,and radiomics scores(Rad-score)were computed.The ultrasound model,radiomics model,and combined model were constructed.Subsequently,the diagnostic efficacy and clinical application value of each model were evaluated using the area under the receiver operating characteristic curve(AUC)and decision curve analysis(DCA).Results Univariate analysis and multivariate logistic regression analysis findings indicated that microcalci-fication,irregular margin,and Rad-score were independent risk factors for the malignant transformation of BethesdaⅢ nodules.In the testing set,the AUC values of the ultrasound model,radiomics model,and combined model were 0.76,0.71,and 0.81,respectively.The calibration curve of the combined model revealed a good consistency between the predicted values and the actual outcomes.The DCA of the testing set demonstrated that the combined model exhibited high clinical utility.Conclusion The combined model,established based on ultrasonic features and ultrasound radiomics,provides a higher predictive value for evaluating the malignancy risk of Bethesda Ⅲ nodules.

In recent years, lung function assessment has attracted increasing attention in the perioperative management of thoracic surgery. However, traditional pulmonary function testing methods remain limited in clinical practice due to high equipment requirements and complex procedures. With the rapid development of artificial intelligence (AI) technology, lung function assessment based on multimodal chest imaging (such as X-rays, CT, and MRI) has become a new research focus. Through deep learning algorithms, AI models can accurately extract imaging features of patients and have made significant progress in quantitative analysis of pulmonary ventilation, evaluation of diffusion capacity, measurement of lung volumes, and prediction of lung function decline. Previous studies have demonstrated that AI models perform well in predicting key indicators such as forced expiratory volume in one second, diffusing capacity for carbon monoxide, and total lung capacity. Despite these promising prospects, challenges remain in clinical translation, including insufficient data standardization, limited model interpretability, and the lack of prediction models for postoperative complications. In the future, greater emphasis should be placed on multicenter collaboration, the construction of high-quality databases, the promotion of multimodal data integration, and clinical validation to further enhance the application value of AI technology in precision decision-making for thoracic surgery.