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 study the correlation between iodine level in drinking water and conventional water quality indicators.Methods:From June 2017 to July 2018, a simple random sampling method was used to select administrative villages (communities) from 88 counties (cities, districts) in Guizhou Province with a sampling size greater than 5%. One drinking water sample was collected from each administrative village (community), and conventional water quality indicators (including fluorine, aluminum, mercury, selenium, sulfate, total dissolved solids and total hardness) were tested in accordance with the methods outlined in the "Standards for Drinking Water Quality" (GB 5749-2006). The cerium sulfate catalytic spectrophotometric method was employed to test drinking water iodine level. Spearman method was utilized to analyze the correlation between iodine level in drinking water and conventional water quality indicators.Results:A total of 904 drinking water samples were tested, with a median iodine level of 1.90 μg/L and a range of 0.10 - 36.70 μg/L. There were 899 administrative villages (communities) with a water iodine level of less than 10 μg/L, accounting for 99.45%. There were only 5 administrative villages (communities) with a water iodine level of greater than 10 μg/L, accounting for 0.55%. Correlation analysis revealed that in Guizhou Province, the iodine level in drinking water was positively correlated with the levels of fluorine, aluminum, sulfate, total dissolved solids, and total hardness [correlation coefficients ( r) = 0.11, 0.07, 0.07, 0.08, 0.07, P < 0.05], and was a negatively correlated with mercury and selenium levels ( r = - 0.12, - 0.12, P < 0.001). Conclusions:External environment in Guizhou Province is generally deficient in iodine. The iodine level in drinking water is positively correlated with the levels of fluorine, aluminum, sulfate, total dissolved solids, and total hardness, and negatively correlated with the levels of mercury and selenium.

Objective To analyze the modeling and evaluation method for vascular restenosis animal models in the last 10 years,to provide a reference for improving animal models of vascular restenosis.Methods Literature related to vascular restenosis was retrieved from mainstream Chinese and English databases from 2013 to 2023.Data on experimental animal strains,modeling method,modeling cycles,and detection method were extracted from the included literature,and a database was established using Excel for summary analysis.Results Among the 122 identified articles,the main experimental animals were rats,rabbits,and pigs,and most animals were male.The most common modeling method was balloon injury,and the modeling cycle was mainly within 4～8 weeks.The main detection indexes were histopathology,accounting for 37.18％,including routine hematoxylin-eosin,Masson,and Elastica van Gieson(EVG)staining.Conclusions The translatability of porcine vascular restenosis models is currently more in line with expectations,but their cost is high and they are unpopular,and rats and rabbits thus remain the main animal models.Balloon injury is the main mode of modeling.Different animal models and modeling method for vascular restenosis have advantages and disadvantages,and the model should be selected according to the experimental purpose.Animal models of vascular restenosis still have some limitations,however,and better animal models are required in the future.

Objective To analyze the modeling and evaluation method for vascular restenosis animal models in the last 10 years,to provide a reference for improving animal models of vascular restenosis.Methods Literature related to vascular restenosis was retrieved from mainstream Chinese and English databases from 2013 to 2023.Data on experimental animal strains,modeling method,modeling cycles,and detection method were extracted from the included literature,and a database was established using Excel for summary analysis.Results Among the 122 identified articles,the main experimental animals were rats,rabbits,and pigs,and most animals were male.The most common modeling method was balloon injury,and the modeling cycle was mainly within 4～8 weeks.The main detection indexes were histopathology,accounting for 37.18％,including routine hematoxylin-eosin,Masson,and Elastica van Gieson(EVG)staining.Conclusions The translatability of porcine vascular restenosis models is currently more in line with expectations,but their cost is high and they are unpopular,and rats and rabbits thus remain the main animal models.Balloon injury is the main mode of modeling.Different animal models and modeling method for vascular restenosis have advantages and disadvantages,and the model should be selected according to the experimental purpose.Animal models of vascular restenosis still have some limitations,however,and better animal models are required in the future.

Objective:To study the correlation between iodine level in drinking water and conventional water quality indicators.Methods:From June 2017 to July 2018, a simple random sampling method was used to select administrative villages (communities) from 88 counties (cities, districts) in Guizhou Province with a sampling size greater than 5%. One drinking water sample was collected from each administrative village (community), and conventional water quality indicators (including fluorine, aluminum, mercury, selenium, sulfate, total dissolved solids and total hardness) were tested in accordance with the methods outlined in the "Standards for Drinking Water Quality" (GB 5749-2006). The cerium sulfate catalytic spectrophotometric method was employed to test drinking water iodine level. Spearman method was utilized to analyze the correlation between iodine level in drinking water and conventional water quality indicators.Results:A total of 904 drinking water samples were tested, with a median iodine level of 1.90 μg/L and a range of 0.10 - 36.70 μg/L. There were 899 administrative villages (communities) with a water iodine level of less than 10 μg/L, accounting for 99.45%. There were only 5 administrative villages (communities) with a water iodine level of greater than 10 μg/L, accounting for 0.55%. Correlation analysis revealed that in Guizhou Province, the iodine level in drinking water was positively correlated with the levels of fluorine, aluminum, sulfate, total dissolved solids, and total hardness [correlation coefficients ( r) = 0.11, 0.07, 0.07, 0.08, 0.07, P < 0.05], and was a negatively correlated with mercury and selenium levels ( r = - 0.12, - 0.12, P < 0.001). Conclusions:External environment in Guizhou Province is generally deficient in iodine. The iodine level in drinking water is positively correlated with the levels of fluorine, aluminum, sulfate, total dissolved solids, and total hardness, and negatively correlated with the levels of mercury and selenium.

Objective:To explore the β-amyloid (Aβ) deposition pattern of subjects with the preclinical Alzheimer′s disease (AD), community-derived amnestic mild cognitive impairment (aMCI) and normal cognition (NC) from communities of Shanghai.Methods:According to the inclusion and exclusion criteria, 273 subjects (104 males, 169 females; age (64.2±7.6) years) were recruited from Shanghai community and memory clinics from December 2018 to July 2020. All subjects underwent MRI, 18F-AV45 PET imaging and neuropsychological scale tests and were grouped into AD, aMCI and NC groups based on clinical diagnosis. Differences in demographic information, the neuropsychological scale tests′ scores and positive rate of Aβ deposition among each group were analyzed by one-way analysis of variance or χ2 test. Aβ deposition patterns of AD and MCI groups were analyzed at voxel level, and the differences of Aβ deposition among different groups were compared. Results:Among 273 patients, the positive rates of Aβ deposition in AD, aMCI and NC groups were 84.4%(38/45), 36.4%(20/55) and 23.1%(40/173), respectively ( χ2=58.37, P<0.001). Among AD, aMCI, NC and NC (Aβ-) groups ( n=132), the education years of AD group was the lowest ((9.7±4.6) years; F=8.86, P<0.001). In addition, there were significant differences in the scores of several neuropsychological scale tests among AD, aMCI, NC groups and NC (Aβ-) group ( F values: 27.68-235.50, all P<0.001). Compared with subjects in NC(Aβ-) group, the Aβ depositions in the aMCI and AD groups were widely distributed in the whole cerebral cortex; and AD group had higher Aβ deposition in bilateral frontal, parietal, temporal, occipital lobe, cingulate gyrus and precuneus than aMCI group. Conclusions:The positive rate of Aβ deposition in the preclinical AD population from the Shanghai community is obtained. There are significant different Aβ deposition patterns in subjects at different stages of AD.

Objective:To explore the association of the impaired cognition and the deposition of β-amyloid (Aβ) in normal cognitive (NC) and mild cognitive impairment (MCI).Methods:From December 2018 to January 2021, 305 subjects (113 males, 192 females; age (64.0±7.7) years) who completed neuropsychological tests and MRI in Shanghai Sixth People′s Hospital, Shanghai Jiao Tong University and 18F-florbetapir (AV45) PET imaging in Huashan Hospital, Fudan University were retrospectively analyzed. The subjects were divided into MCI group and NC group based on neuropsychological tests, and each group was further divided into Aβ-positive and Aβ-negative based on PET imaging results. Independent-sample t test, Mann-Whitney U test and χ2 test were used to analyze the data. Results:There were 118 subjects in MCI group and 187 subjects in NC group. The Aβ-positive rate in MCI group (37.3%, 44/118) was higher than that in NC group (26.2%, 49/187; χ2=4.19, P=0.041). The assessment performances of MCI group in general cognitive function, memory function, language function and executive function were inferior to those of NC group ( t values: from -10.63 to -6.31, z values: from -11.01 to -6.03, all P<0.001). The Auditory Verbal Learning Test-Long Delay Recall (AVLT-LDR) score of Aβ-positive subjects was lower than that of Aβ-negative subjects in MCI group (1.00(0.00, 3.00) and 3.00(1.00, 4.00); z=-2.49, P=0.013). The Montreal Cognitive Assessment Basic (MoCA-B) score of Aβ-positive subjects was lower than that of Aβ-negative subjects in NC group (25.29±2.67 and 26.36±2.42; t=-2.61, P=0.010). Conclusion:Compared to Aβ-negative subjects, MCI patients with Aβ-positive perform worse on memory tests, and NC subjects with Aβ-positive perform worse on general cognitive function.

ObjectiveTo explore the mechanism of aerobic capacity on depression in school-age children, and the multiple mediators of the five dimensions of psychosocial functioning (emotional symptoms, conduct problems, peer problems, prosocial behavior and hyperactivity) between aerobic capacity and depression. MethodsFrom October to December, 2021, pupils of Grade two to Grade five from two primary schools were chester-sampled and investigated using 20-meter multistage shuttle run test, Depression Self-Rating Scale for Children, Self-reported Strengths and Difficulties Questionnaire. ResultsA total of 391 pupils underwent 20-meter multistage shuttle run test, and 312 out of them answering the questionnaires, and 294 questionnaires were valid. Aerobic capacity, depression, emotional symptoms, peer problems, prosocial behavior and hyperactivity were significantly correlated with each other (|r| > 0.127, P < 0.05) (except aerobic capacity and peer problems, and emotional symptoms and prosocial behavior). The results of the multiple mediation effect model showed that aerobic capacity could directly and negatively predict depression, and the mediating effects of emotional symptoms, peer problems, prosocial behavior and hyperactivity were significant, accounting for 34.37%, 12.54%, 34.06% and 17.80% of the total mediating effect, respectively. ConclusionThe aerobic capacity could not only directly affect depression of school-age children, but also improve their psychosocial functioning by reducing emotional symptoms, peer problems and hyperactivity, and increasing prosocial behavior, to indirectly affect their depression.

Objective:To investigate the tau deposition pattern in Alzheimer′s disease (AD) and its correlation with cognition by 18F-MK6240 PET imaging. Methods:From August 2021 to February 2022, 46 elderly people over 55 years old (16 males, 30 females; age (68.9±7.7) years) were included from outpatient and community in Shanghai. Structural brain MRI, β-amyloid (Aβ) PET imaging, tau-PET imaging and comprehensive neuropsychological tests batteries were conducted. The subjects were divided into AD group ( n=16) and normal cognition (NC) group ( n=30) according to the 2018 National Institute on aging and Alzheimer′s Association (NIA-AA) diagnostic criteria. Quantitative analysis was conducted to investigate the tau deposition pattern in AD after preprocessing 18F-MK6240 PET images with MRI images. SUV ratio (SUVR) of brain regions such as entorhinal cortex, hippocampus, parahippocampal gyrus, amygdala, insular lobe, frontal lobe, precuneus, occipital lobe, thalamus and putamen were analyzed, with cerebellum as reference region. The differences of tau deposition in brain regions between AD and NC groups were analyzed by independent-sample t test. The associations between SUVR and Mini-Mental State Examination (MMSE) score and Montreal Cognitive Assessment-Basic (MoCA-B) score were analyzed by Pearson correlation analysis. Results:AD displayed a significant tau deposition in frontal lobes, temporal lobes and parietal lobes compared with NC. SUVR of brain regions in AD group were higher than those in NC group ( t values: 3.37-9.61, all P<0.05). SUVR in brain regions were negatively correlated with MMSE score ( r values: from -0.735 to -0.350, all P<0.05) and MoCA-B score ( r values: from -0.723 to -0.367, all P<0.05). Conclusion:18F-MK6240 PET can demonstrate the tau deposition in the brain of AD patients, and the tau deposition is related to cognitive function.

Objective:To analyze the influencing factors of HIV-testing awareness of Myanmar people in Dehong, and provide references for early discovery of HIV infection among Myanmar people.Methods:A cross-sectional study was conducted among 400 Myanmar people in Dehong through convenient sampling method in a form of face-to-face questionnaire interview. HIV-testing awareness and influencing factors were described and analyzed with χ2 and logistic regression method . Results:A total of 393 participants were included in the study, in whom 241 (61.3%) were males, 256(65.1%) were Myanmar people, 233(59.3%) received 1-7 years of education in Myanmar, 348 do not speak Chinese (88.5%), 226(57.5%) were married, 378(96.2%) were non-solitary, 257(65.4%) were workers, the main purpose of etering into China in 273(69.5%) was working in China, 273(69.5%) had stayed in China for more than 2 years, 573(95.4%) had a clear plan on whether to stay in China for a long time in the future. In these Myanmar people, the rate of AIDS-related knowledge awareness was 75.6%; 165(42.0%) had sexual behavior in the last 6 months. 221(56.2%) knew that they had been tested for HIV. Ninety (22.9%) were more likely to actively seek HIV-testing after they had unprotected behaviors such as homosexual sex, injecting drug use, and sex without using condom. Multivariate logistic regression analysis showed that the influencing factors of HIV-testing awareness included being women (compared with men, OR=2.155, 95% CI: 1.088-4.270), educational level of 8-10 years or more than 10 years (compared with education level 1-4 years, OR=5.207, 95% CI: 2.094-12.950; OR=19.780, 95% CI: 5.800-67.457), having sexual behavior in the last 6 months (compared with those who had not, OR=2.534, 95% CI: 1.343-4.779), having received AIDS-related education in the last 6 months (compared with those who had not, OR=2.462, 95% CI: 1.303-4.654), AIDS-related knowledge awareness (compared with those who had no awareness, OR=8.126, 95% CI: 2.936-22.491). Conclusions:Myanmar people have low awareness of HIV-testing in Dehong. Those who were women and high literacy levels, having received AIDS-related knowledge education in the past 6 months, and having AIDS-related knowledge have an active HIV-testing awareness, after they have high-risk behaviors. AIDS-related knowledge should be publicized to promote early discovery of high-risk behaviors.