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 investigate the correlation between serum levels of long non-coding RNA(Ln-cRNA)-prostate androgen regulated transcript 1(PART1),LncRNA-nucleolar ribonucleic acid host gene 14(SNHG14)and disease stage,cognitive impairment and motor function in patients with Parkinson's disease(PD).Methods A total of 100 PD patients(PD group)who admitted to the Department of Neurology in the hospital from January 2021 to December 2023 and 100 healthy subjects(control group)who underwent the physical examination during the same period of time were selected.According to Hoehn-Yahr staging,PD pa-tients were divided into early stage group(grade 1.0-2.5,20 cases),middle stage group(grade 3.0,48 ca-ses)and late stage group(grade 4.0-5.0,32 cases).According to the Montreal Cognitive Assessment(Mo-CA)score,the patients were divided into normal cognitive group(MoCA score≥26 points,33 cases),PD-mild cognitive impairment group(MoCA score 21-＜26 points,46 cases)and PD dementia group(MoCA score＜21 points,21 cases).According to the Unified Parkinson's Disease Rating Scale(UPDRS)-Ⅲ score,the pa-tients were divided into mild dyskinesia group(0-15 points,29 cases),moderate dyskinesia group(＞15-40 points,46 cases)and severe dyskinesia group(＞40-56 points,25 cases).Real-time fluorescence quantitative PCR was used to detect serum LncRNA-PART1 and LncRNA-SNHG14 levels.Spearman method was used to analyze the correlation between serum LncRNA-PART1,LncRNA-SNHG14 levels and Hoehn-Yahr staging,MoCA score and UPDRS-Ⅲ score in PD patients.Results The level of serum LncRNA-PART1 in PD group was lower than that in control group(P＜0.05),and the level of LncRNA-SNHG14 was higher than that in control group(P＜0.05).The serum levels of LncRNA-PART1 in the middle stage group and late stage groups were lower than those in the early stage group(P＜0.05),and the levels of LncRNA-SNHG14 were higher than those in the early stage group(P＜0.05).In addition,the serum level of LncRNA-PART1 in the late stage group was lower than that in the middle stage group(P＜0.05),and the level of LncRNA-SNHG14 was higher than that in the middle stage group(P＜0.05).The serum LncRNA-PART1 levels in the PD-mild cognitive impairment group and PD dementia group were lower than those in the normal cognitive group(P＜0.05),while the LncRNA-SNHG14 levels were higher than those in the normal cognitive group(P＜0.05).Additionally,the serum LncRNA-PART1 level in the PD dementia group was lower than that in the PD-mild cognitive impairment(P＜0.05),while the LncRNA-SNHG14 level was higher than that in the PD-mild cog-nitive impairment group(P＜0.05).The serum levels of LncRNA-PART1 in the moderate dyskinesia group and severe dyskinesia group were lower than those in the mild dyskinesia group(P＜0.05),and the levels ofLncRNA-SNHG14 were higher than that in the mild dyskinesia group(P＜0.05).In addition,the serum level of LncRNA-PART1 in the severe dyskinesia group was lower than that in the moderate dyskinesia group(P＜0.05),and the level of LncRNA-SNHG14 was higher than that in the moderate dyskinesia group(P＜0.05).Spearman method results showed that serum LncRNA-PART1 level was negatively correlated with Hoehn-Yahr staging and UPDRS-Ⅲ score in PD patients,and positively correlated with MoCA score(P＜0.05).The level of serum LncRNA-SNHG14 was positively correlated with Hoehn-Yahr staging and UPDRS-Ⅲ score in PD patients,and negatively correlated with MoCA score(P＜0.05).Conclusion The level of ser-um LncRNA-PART1 in PD patients is decreased,and the level of LncRNA-SNHG14 is increased,both of them are related to the disease stage,cognitive impairment and motor function of PD patients,which may be-come evaluation indicators for PD progression.

Objective:To investigate the effect of tofacitinib on early atherosclerosis of patients with systemic lupus erythematosus and explore the possible relationship between lupus nephritis and early atherosclerosis of systemic lupus erythematosus.Methods:Sixteen 8-week-old female MRL/lpr mice with a body weight of 20~25 g were selected and randomly divided into the treatment group and placebo group, with 8 mice in each group. The treatment group diluted tofacitinib by normal saline, and given at a dose of 10 mg·kg -1·d -1, and the placebo group (starch tablets) administered the medication in the same way as the treatment group for a total of 8 weeks. The ELISA method was applied to detect serum anti-dsDNA antibody levels in the two groups of mice. Bradford method protein concentration was used to determine the level of urine protein in mice. Automatic biochemical analyzer was used to detect blood lipids, urea nitrogen, serum creatinine, complement C3, complement C4 levels. Western blotting was used to determine the protein expression levels of monocyte chemoattractant protein-1 (MCP-1), non-receptor protein tyrosine kinase family 1 (JAK1), signal transducer and activator of transcription 1 (STAT1) and signal transducer and activator of transcription 2 (STAT2) in aortic and kidney tissues. After the aortic arch section were prepared, oil red O was used to stain the sections, and the vascular plaque area and intimal thickness were evaluated by ImageJ software. The kidneys were dissected and stained with HE, and the active lesions of lupus nephritis were evaluated using the glomerular activity scoring system. SPSS 23.0 software was used for statistical analysis, in which the between-group comparison was performed using two independent samples t-test, and the correlation analysis was performed using the Spearman method. Results:①The serum anti-dsDNA antibody expression level in the treatment group [(5.2±1.0) U/ml] was lower than that in the placebo group [(6.9±1.2) U/ml], ( Z=-3.07, P=0.008), and the levels of complement C3 and complement C4 were higher than those in the placebo group [(293±10) mg/L vs. (260±19) mg/L, Z=2.72, P=0.017]; (16±6) mg/L vs. (8±9) mg/L, Z=3.78, P=0.006]. There was no significant difference in serum BUN and Scr between the treatment group and the placebo group [(10.6±0.7) mmol/L vs. (11.5±1.1) mmol/L, Z=-1.96, P=0.071; (17±5) μmol/L vs. (22±6) μmol/L, Z=-1.79, P=0.095]. ② Compared with the placebo group, the levels of LDL, TC and TG in the treatment group decreased [(0.83±0.15) mmol/L vs. (1.08±1.05) mmol/L, Z=-3.95, P=0.001; (2.90±0.08) mmol/L vs. (1.81±0.97) mmol/L, Z=-5.17, P=0.001; (1.10±0.08) mmol/L vs. (1.60±0.42) mmol/L, Z=-3.23, P=0.013], and HDL level increased [(2.02±0.99) mmol/L vs. (1.81±0.97) mmol/L, Z=4.42, P=0.001]. ③ Compared with the placebo group, the levels of aortic MCP-1, JAK1, STAT1 and STAT2 in the treatment group were reduced [(0.17±0.30) vs. (0.23±0.05), Z=-3.06, P=0.009; (0.83±0.09) vs. (1.05±0.19), Z=-3.07, P=0.008; (0.77±0.07) vs. (0.94±0.13), Z=-2.83, P=0.014; (0.70±0.07) vs. (0.82±0.09), Z=-2.83, P=0.013], the aortic plaque area and aortic intimal thickness were lower than those in the placebo group [(12±31) μm 2vs. (1 242±1 101) μm 2, Z=-3.12, P=0.016; (63±7) μm vs. (82.10±8.06) μm, Z=-5.13, P<0.001]. ④ Compared with the placebo group, the urine protein level and glomerulonephritis activity score in the treatment group were decreased [(0.08±0.03) mg/mL vs. (0.20±0.11) mg/mL, Z=-3.08, P=0.015; (1.79±0.38) vs. (2.79±0.14) points, Z=-7.08, P<0.001)], and renal tissue MCP-1, JAK1, STAT1.Compared with the placebo group, STAT2 levels were reduced [(0.364±0.040) vs. (0.425±0.021), Z=-3.85, P=0.003; (0.689±0.074) vs. (0.838±0.068), Z=-4.19, P=0.001; (0.508±0.070) vs. (0.646±0.019), Z=-2.85, P=0.015; (0.618±0.062) vs. (0.740±0.101), Z=-2.94, P=0.013. ⑤ The glomerular mobility scores of the two groups were positively correlated with LDL, TCHO, TG, aortic plaque area and aortic intimal thickness ( r=0.51, P=0.043; r=0.79, P<0.001; r=0.64, P=0.008; r=0.82, P<0.001; r=0.74, P=0.001), and negatively correlated with HDL ( r=-0.53, P=0.036). The urine protein levels in the two groups were positively correlated with LDL, TC, TG, aortic plaque area and aortic intimal thickness ( r=0.67, P=0.004; r=0.68, P=0.004; r=0.53, P=0.033; r=0.80, P<0.001; r=0.74, P=0.001), and negatively correlated with HDL ( r=-0.57, P=0.021). Conclusion:The severity of lupus nephritis is correlated with atherosclerosis and dyslipidemia in the early stage of systemic lupus erythematosus. Tofacitinib may reduce the degree of early arteriosclerosis and lupus nephritis in MRL/LPR mice, and reduce blood lipid levels, which may be effective in improving the prognosis of SLE and improving the survival rate of patients.

Objectives:To evaluate the safety and efficacy of implantation of subcutaneous implantable cardioverter defibrillator(S-ICD)after transvenous lead extraction(TLE)in ICD patients without pacing indications. Methods:All patients who underwent TLE at Peking University People's Hospital from June 2018 to October 2023 were consecutively included.TLE indication,S-ICD implantation indication,defibrillation threshold(DFT)test,complications and postoperative follow-up were collected and analyzed. Results:A total of 11 patients with TV-ICD underwent S-ICD implantation after TLE,eight patients were males and median age were 56(44,65)years.The indications for TLE were infection and lead dysfunction.Nine patients were implanted with S-ICD for secondary prevention,and the most common cause of implantation was ion channelopathies(5 cases).The operative time for S-ICD was 51(48,58)minutes and no perioperative complications were noted.Eight patients underwent DFT testing,and 100%were successful.During a median follow up of 30(9,39)months,a total of six appropriate treatments occurred in two patients,and no complications occurred,including inappropriate treatment,ineffective treatment,infection,lead malfunction and death. Conclusions:Our study provides evidence for S-ICD implantation as a replacement after TV-ICD removal.The S-ICD implantation after TLE is safe and effective.

Objectives:To assess the clinical characteristics and lead extraction in patients with venous occlusion related to infection of cardiovascular implantable electronic devices. Methods:Clinical data of 405 patients(147 men,mean age[62.4±13.2]years)who underwent lead extraction from January 2020 to January 2024 in Peking University People's Hospital were reviewed.Contrast venography of the access vein was retrospectively analyzed.The patients were divided into venous occlusion group(n=119)and non-venous occlusion group(n=286)according to the presence or absence of venous occlusion.The clinical characteristics and lead extraction of patients in two groups were analyzed. Results:Occlusion of the access vein occurred in 119 patients(29.4%).The subclavian vein was occluded in 48 cases(40.3%),brachiocephalic vein was occluded in 37 cases(31.1%),axillary vein was occluded in 30 cases(25.2%),superior vena cava was occluded in 4 cases(3.4%).There were no significant differences between venous occlusion group and non-venous occlusion group in terms of age,sex,device type,number of leads,or anticoagulation therapy(all P>0.05).Time from implant of the initial leads was significantly longer in the venous occlusion group than in the non-venous occlusion group([10.4±3.8]years vs.[5.9±4.1]years,P=0.042).Clinical extraction success rate and complications were similar between the venous occlusion group and the non-venous occlusion group(both P>0.05).Procedural duration and fluoroscopy exposure time were significantly lower in non-venous occlusion group than in the venous occlusion group(both P<0.05).Patients in the venous occlusion group required more advanced tools(such as laser sheaths,evolution sheaths,and needle's eye snares)for lead extraction compared to patients in the non-venous occlusion group(84.0%vs.67.1%,P=0.001). Conclusions:The incidence of venous occlusion related to infection of cardiovascular implantable electronic devices is 29.4%.Time from implant of the initial leads is significantly longer and lead extraction is more difficult in patients with venous occlusion,and requires more advanced tools and more time to achieve the successful lead extraction.

Objectives:To evaluate the strategy and clinical outcomes of reimplanting biventricular cardiac resynchronization therapy (Biv-CRT) devices after transvenous removal of coronary sinus left ventricular leads due to device-related infections. Methods:A retrospective analysis was conducted on the clinical data and surgical outcomes of all patients who underwent transvenous removal of infectious coronary sinus left ventricular leads and subsequent reimplantation of Biv-CRT devices at Peking University People's Hospital from January 2013 to December 2022.Follow-up was performed to assess the incidence of complications and all-cause mortality. Results:A total of 167 patients underwent coronary sinus left ventricular lead removal due to infection,removal was successful in 161 cases (96.4％) and failed in 6 cases (3.6％).Among the patients with successful removal,109 cases (67.7％) were scheduled for Biv-CRT device reimplantation.After a median time of 7 (5,7) days,6 cases (5.5％) of reimplantation failed,while 103 reimplantations (94.5％) were successful.Among these successful reimplantation cases,102 patients (99.0％) were through the right-side approach,and 1 case (1.0％) through the left-side approach due to bilateral pocket infections.Of the 161 patients with successful removal,58 cases (36.0％) did not undergo left ventricular lead reimplantation,including 39 cases (24.2％) where the initial indications for Biv-CRT were questionable or had resolved.During the one-year postoperative follow-up,among the 103 patients who had undergone Biv-CRT device reimplantation,7 patients (6.8％) died,1 patient (1.0％) experienced pocket infection,and 1 patient (1.0％) had right atrial lead dislodgment. Conclusions:Reimplantation of Biv-CRT devices after removal of coronary sinus left ventricular leads due to infections is feasible,with a high success rate,low complication rate,and low mortality rate for right-side approach implantation.Therefore,for patients re-evaluated to have indications for repeated Biv-CRT after lead removal,right-side reimplantation of the coronary sinus left ventricular lead should be recommended.

Objective:To evaluate the safety of leadless pacemaker implantation in super-elderly patients.Methods:Eleven patients with average age of 90 (86, 92) years who underwent leadless pacemaker implantation in the Department of Cardiology, Peking University People′s Hospital from March 2021 to May 2022 were included. The clinical data and implantation information were collected. The complications (cardiac tamponade, myocardial infarction, cerebral infarction, pulmonary embolism, pacemaker reinfection, femoral vein hematoma) and death of patients were documented at 24 h, 3 d, and 1, 3, 6 months after pacemaker implantation.Results:There were 9 males and 2 females with the body mass index of 21(19, 23)kg/m 2. The underlying diseases were hypertension, diabetes, coronary heart disease, chronic kidney disease, chronic obstructive pulmonary disease, previous cerebral infarction and moderate to severe tricuspid regurgitation in 9, 9, 9, 6, 4, 4, 4 patients, respectively. The left ventricular ejection fraction was 49% (45%, 52%), the hemoglobin concentration was 118 (114, 122)g/L, 4 patients were taking anticoagulant drugs, and 6 patients were taking antiplatelet drugs. Eight patients were newly implanted with a leadless pacemaker, 2 patients were implanted after removal the old ones, and 1 case was implanted at the same time as removal. The implantation time was 45(40, 47) minutes, the X-ray exposure time was 14 (13, 15) minutes, the release time was 1 (1, 2), the threshold value was 0.50(0.38, 0.75)V/0.24 ms, the impedance was 730 (700, 770) Ω, and the perceived R-wave amplitude 8.2(6.7, 12.8) mV. During the follow-up period of 8 (6, 10) months, no patient had pacemaker dysfunction; and the threshold, R wave sensing, and impedance were stable and maintained within the normal range. No cardiac tamponade, myocardial infarction, cerebral infarction, pulmonary embolism, pacemaker reinfection or death occurred perioperatively and during the follow-up period; 1 patient had hematoma after femoral vein puncture, which improved after compression treatment. Conclusion:This single-center and small-sample study shows that leadless pacemaker implantation is safe for super-elderly patients.

Objective:To analyze the serum levels of integrin-associated proteins (CD47) in patients with rheumatoid arthritis (RA), and to explore its association with disease activity and bone destruction in RA.Methods:Serum and clinical data were collected from 65 RA patients and 25 healthy subjects. RA patients were grouped into low, moderate, and high bone erosion groups according to 7-joint ultrasonography score (US7). The levels of serum CD47, thrombospondin-1 (TSP-1) and receptor activator of nuclear factor-κB ligand (RANKL) were measured by enzyme-linked immunosorbnent assay (ELISA) in patients with RA and healthy subjects. The statistical analysis was carried out with independent t-test, analysis of variance, nonparametric rank sum test, pearson or Spearman correlation and logistic regression. Results:① The Serum levels of CD47, TSP-1, and RANKL were higher in the RA group than in the healthy controls ( P<0.01). ② In RA patients, serum CD47 level was positively correlated with disease course ( r=0.301, P<0.05), C-reactionprotein (CRP)( r=0.316, P<0.05), number of tender joints (TJC) ( r=0.254, P<0.05), number of swollen joints (SJC) ( r=0.316, P<0.05), disease activity score in 28 joints (DAS28) ( r=0.255, P<0.05), RANKL ( r=0.252, P<0.05) and TSP-1 ( r=0.260, P<0.05). Serum TSP-1 level was positively correlated with CRP ( r=0.299, P<0.05), TJC ( r=0.335, P<0.01), DAS28 ( r=0.315, P<0.05), RANKL ( r=0.305, P<0.05). ③ The disease course [ OR(95% CI)=1.048(1.033, 1.017)] and TSP-1 [ OR(95% CI)=1.013(1.000, 1.026)] were independently relevant factors affecting bone destruction. Conclusion:CD47 levels is significantly higher in RA patients than in healthy controls, and is associated with disease activity and bone destruction. CD47 may be involved in the bone destruction process of RA by acting on TSP-1.

Objective To investigate the effect of Kunxian capsule combined with candesartan axetil on blood pressure, renal function, blood lipids and inflammatory factors in patients with chronic glomerulonephritis complicated with hypertension. Methods 101 patients with chronic glomerulonephritis who were admitted to our hospital from November 2017 to December 2019 were selected and randomly divided into observation group (51 cases) and control group (50 cases). The control group was treated with candesartan cilexetil on the basis of conventional treatment, and the observation group was treated with Kunxian capsule on the basis of the control group. The clinical efficacy and adverse reactions of the two groups were compared. Results The total effective rate of the observation group was significantly higher than that of the control group (P<0.05). After treatment, the blood pressure, renal function indexes (24 h Upro, BUN, Scr), blood lipid indexes (TG, TC, LDL-C), and inflammatory factors of the two groups (IL-6, hs-CRP) significantly decreased, and the observation group was significantly lower than the control group (P<0.05). After treatment, the blood lipid index HDL-C of the two groups increased significantly, and the blood lipid index HDL-C of the observation group was significantly higher than that of the control group (P<0.05). Conclusion Kunxian capsule combined with candesartan axetil can enhance the clinical efficacy, improve renal function, regulate blood lipids and reduce inflammation in patients with chronic glomerulonephritis complicated with hypertension.

The massive loss of oligodendrocytes caused by various pathological factors is a basic feature of many demyelinating diseases of the central nervous system (CNS). Based on a variety of studies, it is now well established that impairment of oligodendrocyte precursor cells (OPCs) to differentiate and remyelinate axons is a vital event in the failed treatment of demyelinating diseases. Recent evidence suggests that Foxg1 is essential for the proliferation of certain precursors and inhibits premature neurogenesis during brain development. To date, very little attention has been paid to the role of Foxg1 in the proliferation and differentiation of OPCs in demyelinating diseases of the CNS. Here, for the first time, we examined the effects of Foxg1 on demyelination and remyelination in the brain using a cuprizone (CPZ)-induced mouse model. In this work, 7-week-old Foxg1 conditional knockout and wild-type (WT) mice were fed a diet containing 0.2% CPZ w/w for 5 weeks, after which CPZ was withdrawn to enable remyelination. Our results demonstrated that, compared with WT mice, Foxg1-knockout mice exhibited not only alleviated demyelination but also accelerated remyelination of the demyelinated corpus callosum. Furthermore, we found that Foxg1 knockout decreased the proliferation of OPCs and accelerated their differentiation into mature oligodendrocytes both in vivo and in vitro. Wnt signaling plays a critical role in development and in a variety of diseases. GSK-3β, a key regulatory kinase in the Wnt pathway, regulates the ability of β-catenin to enter nuclei, where it activates the expression of Wnt target genes. We then used SB216763, a selective inhibitor of GSK-3β activity, to further demonstrate the regulatory mechanism by which Foxg1 affects OPCs in vitro. The results showed that SB216763 clearly inhibited the expression of GSK-3β, which abolished the effect of the proliferation and differentiation of OPCs caused by the knockdown of Foxg1. These results suggest that Foxg1 is involved in the proliferation and differentiation of OPCs through the Wnt signaling pathway. The present experimental results are some of the first to suggest that Foxg1 is a new therapeutic target for the treatment of demyelinating diseases of the CNS.