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 explore the effects of acteoside on hypoxia/reoxygena tion（H/R）-induced cardiomyocyte damage by regulating Rho family GTPase 3（Rnd3）/nuclear factor κB（NF-κB）pathway. METHODS The H 9c2 cardiomyocyte were divided into control group （no administration ，no modeling ），H/R group （only modeling ），H/R+AS-L group ，H/R+AS-M group ， H/R+AS-H group （10，30，90 μmol/L acteoside for above 3 groups firstly ，and then modeling ），H/R+pcDNA group [transfecting pcDNA （empty vector ） firstly，and then modeling] ，H/R + pcDNA-Rnd 3 group [overexpression of Rnd 3 by transfecting pcDNA-Rnd3（Rnd3 overexpression vector ）firstly，and then modeling] ，H/R+AS-H+si-NC group [transfecting si-NC （negative control）firstly，and then giving 90 μmol/L acteoside and modeling]，H/R+AS-H+si-Rnd3 group [inhibiting overexpression of Rnd 3 by transfecting si-Rnd 3 （Rnd3 small interfering RNA ） firstly，and then giving 90 μ mol/L acteoside and modeling]. After corresponding treatment ，the apoptotic rate ，release of lactate dehydrogenase （LDH），malondialdehyde（MDA）level，the activity of superoxide dismutase （SOD），the level of tumor necrosis factor α（TNF-α），interleukin 1β（IL-1β）and interleukin- 6（IL-6）， mRNA and protein expression of Rnd 3 and NF-κB subunit p65（NF-κB p65），the expression of aspartate proteolytic enzyme 3 （Cleaved Caspase- 3）protein and Cleaved Caspase- 9 protein were detected. RESULTS Different concentrations of acteoside could reduce the apoptotic rate of H/R-induced H 9c2 cardiomyocyte，the protein expressions of Cleaved Caspase- 3 and Cleaved Caspase-9，mRNA and protein expressions of NF-κB p65，the levels of LDH release and MDA ，TNF-α，IL-1β and IL-6，while increase the activity of SOD and mRNA and protein expressions of Rnd 3（P＜0.05），in a dose-dependent manner. Overexpression of Rnd 3 could decrease the apoptotic rate of H 9c2 cardiomyocyte，protein expressions of NF-κB p65，Cleaved Caspase- 3 and Cleaved Caspase- 9， the levels of LDH release ， MDA， TNF-α，IL-1β and IL-6，while increase the protein expression of Rnd 3 and the activity of SOD （P＜0.05）. The inhibition overexpression of Rnd 3 could weaken the inhibitory effects of acteoside on H/R-induced apoptosis of H 9c2 cardiomyocyte， oxidative stress and inflammatory reaction （P＜0.05）. CONCLUSIONS Acteoside could regulate Rnd 3/NF-κ B pathway by promoting the expression of Rnd 3 and inhibiting the expression of NF-κB p65，inhibit cardiomyocyte apoptosis ，oxidative stress and inflammation reaction so as to relieve the H/R-induced cardiomyocyte damage.

BACKGROUND: In valerian-ligusticum extract (VLE), valeriana offici nalis extract (VOE) is γ aminobutyric acid (GABA) receptor kinetin, which can relax cerebral vascular spasm; ligusticum wallichii Fr. Extraxt (LWE)can pass through blood-brain barrier, enhance microcirculation of tissue and inhibit blood platelet aggregation and 5-Hydroxytryptamine (5-HT) release.OBJECTIVE: To probe into the effects of VLE prepared with effective components on prevention and treatment of cerebral ischemic injury.DESIGN:Complete randomized, negative and positive control experiment.SETTING: Institute of Senile Medicine and Pharmacology of Tongji Medical College of Huazhong University of Science and Technology.MATERIALS: The experiment was performed in Institute of Senile Medicine Pharmacology of Tongji Medical College of Huazhong University of ent blood perfusion in brain tissue: Fifty Kunming mice were employed,which was randomized into normal group, solvent control (model) group,ligustrazine 50 mg/kg group, VLE 170 mg/kg group and VLE 85 mg/kg Fifty Wistar rats were employed, which was randomized into solvent control (model) group, compound danshen (Radix Salviae Miltiorrhizae) 5 g/kg group,VLE 156 mg/kg group, VLE 94 mg/kg group and VLE 31.3 mg/kg group,Sixty Wistar rats were employed, which was randomized into sham-operation group, solvent control (model) group, ligustrazine 10 mg/kg group, VLE 156 mg/kg group, VLE 95 mg/kg group and VLE 31.3 mg/kg, 10 mice in each were employed, which was randomized into normal group, solvent control (model) group, ligustrazine 10 mg/kg group, VLE 200 mg/kg group and VLE 40 mg/kg, 10 mice in each one.sue, in advance, VLE (85, 170 mg/kg), ligustrazine (50 mg/kg) or solvent enhancer of equal volume (0.2 mL) were injected abdominally in each group. Twenty minutes later, pituitrin (2.5 u/kg) was injected intravenously; and 10 minutes later, isotope 99Tcm+ L, L-EthylCysteinate Dimer and Stannous Chloride (ECD) 3.7×1010Bq/ L(0.1 mL/per mouse) was injected in coccygeal nerve. Fifteen minutes later, radio-immunity counter was used periment of arteral-ovenous bypass method for thrombosis, before the opercal saline successively, continuously for 7 days, once per day. After 24 hours of medication pause, with abdominal anesthesia with pentobarbitol sodium, a catheter (with surgical thread inside) was used in vitro to connect common cervical vein and carotid artery. Thrombus mass was scaled 15 dominal anesthesia of chloral hydrate, intraluminal thread approach (ITA)was used to block unilateral MCA. Except that ITA was not used, the other management in sham-operation group was same as experimental groups.Gastric perfusion was done with VLE(156, 94, 31.3 mg/kg), ligustrazine operation and 3 hours and 12 hours after operation. 24 hours after modeling, the assessment was done for behavioral neurological damage and brain sive cerebral ischemia experiment, the model was prepared by coccygeal injection of collagen + adrenalin (AD). Respectively, 30 minutes before modeling injection and 1 hour after injection, gastric perfusion was done with VLE (200, 40 mg/kg), ligustrazine (10 mg/kg) or solvent enhancer of equal volume successively to observe the numbers of dead mice in 5 minutes after modeling and the numbers of hemiplegia mice in 15 minutes;and to determine brain mass index 8 hours later after sacrificed and lactic acid level of brain tissue homogenate with ultraviolet spectrophotometry.group.RESULTS: In the experiment of acute extensive brain ischemia in mice, in solvent control, during modeling, 3 mice were died and the rest 207 mice brain tissue in mice, the ratios of brain with and blood γ ray pulsating intensity in VLE 85 mg/kg group and VLE 170 mg/kg were higher than model group (0.53±0.09, 0.55±0.08, 0.45±0.08, t=2.234 6, 2.793 3, P method in rats, the thrombus masses in VLE 156 mg/kg group, 94 mg/kg group and 31.3 g/kg group were lower remarkably than the model group [(12.66±4.79), (13.31 ±3.97), (13.49±4.09), (19.21±5.76) g, (t=2.667 0,31.3 mg/kg group, 94 mg/kg group and 156 mg/kg group was lower remarkably than model group successively [(5.9±1.9), (6.0±2.0), (5.8±2.2),(8.7±0.9) score], and cerebral infarction index was lower than model group [(16.52±5.78)%,(16.54±3.00)%, (14.18±6.13)%, (24.03±4.85)%, (t=3.118 9-chemia in mice, brain mass indexes of VLE 40 mg/kg and 200 mg/kg groups were lower remarkably than model group [(0.91 ±0.20) and (0.82±0.24)%, (1.40±0.32)%], and lactic acid in brain tissue was lower than model group [(17.44±6.71),(14.43±2.81), (29.07±7.33) μmol/g (t=3.388 5-5.800 5, P＜ 0.01)].CONCLUSION: Valerian-liqusticum extract improves significantly cerebral ischemia in mice induced by pituitrin and the damage by medium cerebral artery embolism in rats, and it inhibits significantly blood platelet aggregation and thrombosis induced by AD+ collagen mixture or foreign objects. It is suggested that valerian-ligustrazine extract prevents and treats significantly the perfusion disturbance of cerebral microcirculation.