Recent advances in large models demonstrate significant prospects for transforming the field of medical imaging. These models, including large language models, large visual models, and multimodal large models, offer unprecedented capabilities in processing and interpreting complex medical data across various imaging modalities. By leveraging self-supervised pretraining on vast unlabeled datasets, cross-modal representation learning, and domain-specific medical knowledge adaptation through fine-tuning, large models can achieve higher diagnostic accuracy and more efficient workflows for key clinical tasks. This review summarizes the concepts, methods, and progress of large models in medical imaging, highlighting their potential in precision medicine. The article first outlines the integration of multimodal data under large model technologies, approaches for training large models with medical datasets, and the need for robust evaluation metrics. It then explores how large models can revolutionize applications in critical tasks such as image segmentation, disease diagnosis, personalized treatment strategies, and real-time interactive systems, thus pushing the boundaries of traditional imaging analysis. Despite their potential, the practical implementation of large models in medical imaging faces notable challenges, including the scarcity of high-quality medical data, the need for optimized perception of imaging phenotypes, safety considerations, and seamless integration with existing clinical workflows and equipment. As research progresses, the development of more efficient, interpretable, and generalizable models will be critical to ensuring their reliable deployment across diverse clinical environments. This review aims to provide insights into the current state of the field and provide directions for future research to facilitate the broader adoption of large models in clinical practice.
Humans ; Diagnostic Imaging/methods* ; Precision Medicine/methods* ; Image Processing, Computer-Assisted/methods*

Objective: To analyze the results of national external quality assessment (EQA) for transfusion compatibility test from 2018 to 2023, with the aim of providing references for improving laboratory testing quality and ensuring the safety of clinical blood transfusion. Methods: Three EQA programs were conducted annually, each distributing 22 quality assessment samples. Participating transfusion laboratories were required to complete testing within specified deadlines and to submit results along with documentation of testing methodologies, reagents, and equipment used. National Center for Clinical Laboratories (NCCL) conducted statistical analysis of laboratory results, evaluated testing outcomes and related circumstances, and provided feedback to participating laboratories. EQA data from transfusion laboratories across China from 2018 to 2023 were collected and systematically analyzed. Results: From 2018 to 2023, the qualification rates for all five items (ABO forward typing, ABO reverse typing, Rh blood group typing, antibody screening, and cross-matching) were 67.59%, 77.11%, 77.38%, 72.78%, 79.96%, and 85.16%, respectively. The mean qualification rates for ABO forward typing, ABO reverse typing, RhD blood group typing, antibody screening, and cross-matching over the past six years were 96.25%±0.59%, 90.45%±4.52%, 96.05%±0.71%, 90.88%±2.86%, and 88.34%±3.48%, respectively. The qualification rates in 2019, 2020, 2022, and 2023 all showed a stable trend of "blood stations>tertiary hospitals>secondary hospitals". The mean qualification rate of laboratories in secondary hospitals from 2018 to 2023 was significantly lower than those of laboratories in tertiary hospitals and blood stations (P<0.05), while no significant difference was observed between laboratories in tertiary hospitals and blood stations (P>0.05). The micro column agglutination method was the most widely used in all five tests. In the four test items, namely ABO forward typing, ABO reverse typing, antibody screening, and cross-matching, there was a statistically significant difference in the qualification rate of micro column agglutination method compared to other methods (P<0.05). There was a statistical difference in the qualification rate between manual and automated detection using micro column agglutination method in the cross-matching tests (P<0.05), whereas no significant difference was noted for the other test items (P>0.05). Conclusion: From 2018 to 2023, the number of laboratories participating in EQA activities has been increasing year by year, and the qualification rate has shown an overall upward trend. The type of laboratory is a key factor affecting the qualification rate, and the testing capabilities of some laboratories still need to be improved. The micro column agglutination method is widely used in transfusion compatibility tests. The established EQA program effectively monitors quality issues in laboratories, drives continuous improvement, and ensures sustained enhancement of testing standards to safeguard clinical blood safety.

Objectives:To evaluate the performance of AVEIR VR single chamber leadless pacemaker(AVEIR VR)in the real world in China and the predictive factors of pacing threshold to assess the feasibility of AVEIRTM VR implantation in the domestic population.Methods:All patients who underwent AVEIR VR implantation by experienced operators from Peking University People's Hospital in multiple domestic hospitals from June 2024 to October 2024 were consecutively included,and their baseline characteristics,procedural data(including pacemaker electrical measurements at various stages),and follow-up results were observed,recorded,and statistically analyzed.Results:This study included 20 patients who met the indication for pacemaker implantation and underwent AVEIR VR implantation.Their mean age was(71.40±13.37)years,body mass index(BMI)was(23.05±3.71)kg/m2,body surface area(BSA)was(1.70±0.16)m2,all 20 patients were successfully implanted with pacemakers,there were no major complications(newly occurred pericardial effusion during the perioperative period,cardiac tamponade,poor pacemaker function,pacemaker dislocation/perforation,tricuspid valve injury,myocardial infarction,stroke,pulmonary embolism,and clinical death).One patient developed new frequent ventricular premature contractions after pacemaker release.The pacing threshold(PCT)at one week after implantation was correlated with PCT in tether mode(r=0.650,P=0.009)and PCT in release phase(r=0.596,P=0.019),but not with the sensing and impedance in each phase.Conclusions:This preliminary exploration study show that AVEIRTM VR implantation demonstrates satisfactory performance among real world patients in China,especially in elderly and low weight populations,confirming its safety.

Objective:To evaluate the clinical feasibility of subcutaneous implantable cardioverter-defibrillator (S-ICD) implantation following transvenous lead extraction (TLE).Methods:This was a retrospective study. Consecutive patients who underwent S-ICD implantation at Peking University People′s Hospital between June 2015 and October 2023 were enrolled. Patients were divided into the TLE group and the newly implanted group based on whether they received TLE prior to S-ICD implantation. Baseline characteristics, S-ICD indication, defibrillation threshold test results, complications, and postoperative follow-up data were collected and compared between the two groups.Results:A total of 27 patients were included, aged (49.2±14.2) years, including 19 males. There were 12 patients in the TLE group and 15 in the newly implanted group. Compared with the TLE group, patients in the newly implanted group were younger ((43.3±13.7) years vs. (55.6±12.0) years, P=0.013). The main S-ICD indication in the TLE group was high infection risk (9/12), whereas in the newly implanted group it was younger age (11/15). All patients underwent successful S-ICD implantation, with 18 patients completing defibrillation threshold testing (all successful). Additionally, the TLE group had longer follow-up duration than the newly implanted group (42 (19, 60) months vs. 12 (3, 28) months, P=0.001). No complications or deaths occurred during follow-up, with normal device function in both groups. A total of 17 ventricular tachycardia or fibrillation events were recorded, of which 7 met defibrillation criteria and all received effective therapy. Conclusion:S-ICD demonstrates safety and efficacy as a therapeutic option for patients after TLE, with comparable device functionality and follow-up outcomes to patients with newly implanted S-ICD.

Objective:To evaluate the incidence and frequency of inappropriate shocks caused by defibrillation lead failure in patients with implantable cardioverter-defibrillators (ICD), and to explore methods for reducing the incidence and frequency of such inappropriate shocks.Methods:This was a single-center retrospective study involving patients treated for defibrillation lead failures at Peking University People′s Hospital between March 2015 and May 2024. Patients were divided into an alarm function group and a non-alarm function group based on whether their ICDs were equipped with lead alarm functions. Clinical data, lead data, and the incidence and frequency of inappropriate shocks were collected and compared between the two groups. A multivariate logistic regression model was used to analyze factors influencing the incidence and frequency of inappropriate ICD shocks. Kaplan-Meier survival curves were plotted to compare the trends in the incidence and frequency of inappropriate shocks over time since ICD implantation between the two groups.Results:A total of 59 patients were enrolled, with a age of (56.7±15.2) years, including 42 males (71%). The lifespan of the failed leads in the entire cohort was 64.0 (36.0, 96.0) months. There were 26 patients in the alarm function group and 33 in the non-alarm function group. The most common manifestations of lead failure were oversensing (85%, 50/59) and abnormal pacing impedance (42%, 25/59). A total of 33 patients (56%, 33/59) experienced inappropriate shock therapy, with an average of 27.3 shocks per patient. The frequency of inappropriate ICD shocks in the non-alarm function group was higher than that in the alarm function group (25.0 (10.0, 60.0) times/year vs. 5.0 (2.8, 7.8) times/year, P=0.001). Multivariate logistic regression analysis showed that oversensing ( OR=2.057, 95% CI 1.125-6.763, P=0.019) was an influencing factor for incidence of inappropriate shocks, while the lead alert function ( OR=0.062, 95% CI 0.005-0.719, P=0.001) was a factor influencing the frequency of inappropriate shocks. Kaplan-Meier survival analysis revealed that the incidence and frequency of inappropriate shocks increased with the duration of ICD implantation in both groups, but the differences were not statistically significant (incidence: log-rank P=0.908; frequency: log-rank P=0.767). Conclusion:The lead alert function can reduce the frequency of inappropriate shocks caused by lead failure.

Objectives:To analyze the clinical characteristics,strategies,success and complication rates of lead extraction and re-implantation during the upgrade of cardiovascular implantable electronic devices(CIED)in non-infectious patients.Methods:This retrospective study collected and analyzed the baseline clinical data and surgical data of 66 non-infected patients who had their existing CIEDs(including cardiac pacemaker,implantable cardioverter defibrillator[ICD],cardiac resynchronization therapy pacemaker[CRT-P])upgraded to ICD or CRT-P or cardiac resynchronization therapy defibrillator(CRT-D)or subcutaneous implantable cardioverter defibrillator(S-ICD)in Peking University People's Hospital from March 2018 to March 2024.We analyzed the strategies of lead extraction and reimplantation as well as the operation success rate and complication rate.Results:Among the 66 patients,preoperative imaging revealed that 12 patients(18.2%)had severe stenosis/occlusion of the venous access route,with lead wear/perforation in 26 patients(39.4%).32 patients(48.5%)underwent transvenous lead extraction(TLE),of which all leads were removed in 27 patients(84.4%),and only non-functional leads were removed in 5 patients(15.6%).The success rate of the TLE procedure was 100%and no complication occurred.Among the 66 patients,functional leads retained and new leads were implanted on the same side in 28 patients(42.4%),all leads were removed and new leads were reimplanted on the opposite side in 22 patients(33.3%),only non-functional leads were removed and new leads were reimplanted on the same side in 5 patients(7.6%),all leads were removed and new leads were reimplanted on the same side in 5 patients(7.6%),and 6 patients(9.1%)had the leads abandoned and then were re-implanled.The success rate of the upgrade surgery was 100%,no complications were reported.Conclusions:When the existing CIEDs(including cardiac pacemaker,ICD,CRT-P)of non-infected patients are upgraded to ICD,CRT-P,CRT-D or S-ICD,lead extraction and reimplantation are safe and feasible,and reimplantation can be performed on the ipsilateral or contralateral side.

Objective:To analyze the results of national external quality assessment (EQA) for transfusion compatibility test in 2023, and provide reference for quality management of clinical transfusion compatibility testing.Methods:The EQA of clinical transfusion compatibility testing by NCCL was performed 3 times in 2023 among included laboratories. The panel consisting of 22 samples was distributed to 4 186 laboratories across 31 provinces (Including 2 961 tertiary hospital laboratories, 1 085 secondary hospital laboratories, 23 primary hospital laboratories, 106 blood station laboratories and 11 independent clinical laboratories). Each panel contains 11 red blood cell and 11 plasma samples per 1.5 ml/tube. Each participant laboratory of the EQA program was required to carry out the detection and return results in expected time. Statistical analysis and evaluation on the reported results were conducted by NCCL from the aspects of regional distribution, laboratory grading, testing methodology, reagent and testing system usage.Results:The qualification rates of EQA for five items including ABO positive typing, ABO reverse typing, RhD blood type, antibody screening, and cross matching were 96.68%, 95.10%, 96.46%, 95.32%, and 91.04%, respectively. The EQA qualification rate of tertiary hospital laboratories was 87.77% (2 599/2 961), which was significantly higher than the 77.79% (844/1 085) of secondary hospital laboratories. There were significant differences in the qualification rate of participating laboratories among different regions. The utilization rates of micro column agglutination method in ABO positive typing, ABO reverse typing, RhD blood type, antibody screening, and cross matching were 80.81% (10 080/12 474), 75.06% (9 337/12 440), 81.38% (10 118/12 433), 89.59% (11 104/12 394) and 76.25% (9 495/12 453), respectively. The qualification rate of micro column agglutination method was significantly higher than that of saline slide method in ABO positive typing detection ( P<0.05). The qualification rate of micro column agglutination method was significantly higher than that of the polyamine method and anti-human globulin test tube method in antibody screening ( P<0.05). There were statistically significant differences in qualification rate of 7 reagents in ABO reverse typing, antibody screening and cross matching ( P<0.05). There was no statistically significant difference in the qualification rate between the two detection systems for other reagents, except for the ABO reverse typing where the qualification rate of reagent 1 in a single system was higher than that in a mixed system ( P<0.05). Conclusion:The testing capabilities of clinical laboratories in different regions and different type varied significantly in China. Micro column agglutination method was the most popular selection in transfusion compatibility testing. The regents used in these laboratories showed good performance. However, the detection efficiency of some reagents still need to be improved. EQA could be used to evaluate, monitor, and improve the quality of testing.

Objective:To screen the risk factors for non-complete procedural success of transvenous lead extraction(TLE),and to establish a prediction model based on the results and evaluate its predictive efficacy.Methods:A total of 1 029 patients who underwent TLE in Peking University People's Hospital from January 2014 to December 2020 were enrolled and divided into training set(n=720)and validation set(n=309)using the random number method.There were no statistically significant differences among the variables in the training set and the validation set.The training set was divided into the complete procedural success(CPS)group(n=664)and the non-CPS group(n=56).Univariate analysis was employed to screen the relevant indicators of non-CPS,followed by binary logistic regression analysis to identify the independent risk factors of non-CPS.Subsequently,a predictive model and nomogram were constructed.The receiver operating characteristic(ROC)curve analysis was applied to evaluate the ability of the model to distinguish non-CPS from TLE patients in the training set and validation set.The Hosmer-Lemeshow goodness-of-fit test was used to assess the consistency between the predicted risk and the actual risk of the model.Results:Univariate analysis showed that the relevant variables with P<0.1 including the age at the first implantation of the lead,the number of leads extracted,the oldest dwell time of lead extracted,the presence of abandoned leads,non-manual traction for lead extracted,the number of extracted leads>3,bilateral lead implantation,and the indications for TLE.The binary logistic regression analysis revealed that the presence of abandoned leads(OR=2.252,95%CI:1.111-4.564,P=0.024),the oldest dwell time of the extracted leads(OR=1.009,95%CI:1.005-1.012,P<0.001),and the number of extracted leads>3(OR=3.177,95%CI:1.306-7.733,P=0.011)were independent risk factors for non-CPS of TLE.ROC curve analysis revealed that the area under the ROC curve(AUC)of the training set was 0.80(95%CI:0.75-0.85,P<0.001).The AUC of the validation set was 0.81(95%CI:0.72-0.90,P<0.001).The Hosmer-Lemeshow goodness-of-fit test indicated that the P values of both the training set(P=0.089)and the validation set(P=0.136)were greater than 0.05.Conclusions:The presence of abandoned leads,the oldest dwell time of lead extracted,and the number of extracted leads>3 are independent risk factors for non-CPS in patients undergoing TLE.The nomogram model based on the above factors has satisfactory predictive ability.

Objective:To screen the risk factors for non-complete procedural success of transvenous lead extraction(TLE),and to establish a prediction model based on the results and evaluate its predictive efficacy.Methods:A total of 1 029 patients who underwent TLE in Peking University People's Hospital from January 2014 to December 2020 were enrolled and divided into training set(n=720)and validation set(n=309)using the random number method.There were no statistically significant differences among the variables in the training set and the validation set.The training set was divided into the complete procedural success(CPS)group(n=664)and the non-CPS group(n=56).Univariate analysis was employed to screen the relevant indicators of non-CPS,followed by binary logistic regression analysis to identify the independent risk factors of non-CPS.Subsequently,a predictive model and nomogram were constructed.The receiver operating characteristic(ROC)curve analysis was applied to evaluate the ability of the model to distinguish non-CPS from TLE patients in the training set and validation set.The Hosmer-Lemeshow goodness-of-fit test was used to assess the consistency between the predicted risk and the actual risk of the model.Results:Univariate analysis showed that the relevant variables with P<0.1 including the age at the first implantation of the lead,the number of leads extracted,the oldest dwell time of lead extracted,the presence of abandoned leads,non-manual traction for lead extracted,the number of extracted leads>3,bilateral lead implantation,and the indications for TLE.The binary logistic regression analysis revealed that the presence of abandoned leads(OR=2.252,95%CI:1.111-4.564,P=0.024),the oldest dwell time of the extracted leads(OR=1.009,95%CI:1.005-1.012,P<0.001),and the number of extracted leads>3(OR=3.177,95%CI:1.306-7.733,P=0.011)were independent risk factors for non-CPS of TLE.ROC curve analysis revealed that the area under the ROC curve(AUC)of the training set was 0.80(95%CI:0.75-0.85,P<0.001).The AUC of the validation set was 0.81(95%CI:0.72-0.90,P<0.001).The Hosmer-Lemeshow goodness-of-fit test indicated that the P values of both the training set(P=0.089)and the validation set(P=0.136)were greater than 0.05.Conclusions:The presence of abandoned leads,the oldest dwell time of lead extracted,and the number of extracted leads>3 are independent risk factors for non-CPS in patients undergoing TLE.The nomogram model based on the above factors has satisfactory predictive ability.

Objectives:To evaluate the performance of AVEIR VR single chamber leadless pacemaker(AVEIR VR)in the real world in China and the predictive factors of pacing threshold to assess the feasibility of AVEIRTM VR implantation in the domestic population.Methods:All patients who underwent AVEIR VR implantation by experienced operators from Peking University People's Hospital in multiple domestic hospitals from June 2024 to October 2024 were consecutively included,and their baseline characteristics,procedural data(including pacemaker electrical measurements at various stages),and follow-up results were observed,recorded,and statistically analyzed.Results:This study included 20 patients who met the indication for pacemaker implantation and underwent AVEIR VR implantation.Their mean age was(71.40±13.37)years,body mass index(BMI)was(23.05±3.71)kg/m2,body surface area(BSA)was(1.70±0.16)m2,all 20 patients were successfully implanted with pacemakers,there were no major complications(newly occurred pericardial effusion during the perioperative period,cardiac tamponade,poor pacemaker function,pacemaker dislocation/perforation,tricuspid valve injury,myocardial infarction,stroke,pulmonary embolism,and clinical death).One patient developed new frequent ventricular premature contractions after pacemaker release.The pacing threshold(PCT)at one week after implantation was correlated with PCT in tether mode(r=0.650,P=0.009)and PCT in release phase(r=0.596,P=0.019),but not with the sensing and impedance in each phase.Conclusions:This preliminary exploration study show that AVEIRTM VR implantation demonstrates satisfactory performance among real world patients in China,especially in elderly and low weight populations,confirming its safety.