With the emergence of deep learning techniques based on convolutional neural networks, artificial intelligence (AI) has driven transformative developments in the field of medical image analysis. Recently, large language models (LLMs) such as ChatGPT have also started to achieve distinction in this domain. Increasing research shows the undeniable role of AI in reshaping various aspects of medical image analysis, including processes such as image enhancement, segmentation, detection in image preprocessing, and postprocessing related to medical diagnosis and prognosis in clinical settings. However, despite the significant progress in AI research, studies investigating the recent advances in AI technology in the aforementioned aspects, the changes in research hotspot trajectories, and the performance of studies in addressing key clinical challenges in this field are limited. This article provides an overview of recent advances in AI for medical image analysis and discusses the methodological profiles, advantages, disadvantages, and future trends of AI technologies.
Artificial Intelligence ; Humans ; Image Processing, Computer-Assisted/methods* ; Neural Networks, Computer ; Deep Learning ; Diagnostic Imaging/methods*

OBJECTIVE To explore the changes of serum Toll-like receptor(TLR)2,TLR4 and 25 hydroxyvitamin D[25(OH)D]in the general anesthesia orthopedic surgery patients with postoperative pulmonary infections and analyze the distribution and drug resistance of pathogens.METHODS A total of 59 general anesthesia orthopedic surgery patients with postoperative pulmonary infections who were treated in Tianjin Fifth Central Hospital from Jul.2019 to Jul.2024 were chosen as the infection group,meanwhile,65 patients who underwent the same surger-y but did not have postoperative pulmonary infection or other acute and chronic infections were chosen as the non-infection group.The etiological characteristics of the postoperative pulmonary infections and the drug re-sistance rates were analyzed.The clinical data and the levels of serum TLR2,TLR4 and 25(OH)D were observed and compared between the two groups of patients.The values of the above indicators in diagnosis of the postopera-tive pulmonary infections in the general anesthesia orthopedic surgery were analyzed.RESULTS Totally 63 strains of pathogens were isolated from the 59 general anesthesia orthopedic surgery patients with postoperative pulmona-ry infections,36(57.14％)of which were gram-negative bacteria.3 of 15 Pseudomonas aeruginosa strains were resistant to meropenem;4 of 12 Klebsiella pneumoniae strians were resistant to meropenem;all of the 11 Staph-ylococcus aureus strains were methicilin-resistant S.aureus.The proportions of patients with age,hypoproteine-mia and diabetes mellitus,the serum TLR2 and TLR4 levels were higher in the infection group than in the non-infection group(P＜0.05),while the 25(OH)D level of the infection group was lower than that of the no in-fection group(P＜0.05).The area under the curves(AUCs)of serum TLR2,TLR4,25(OH)D and the joint de-tection of the above indicators were respectively 0.817,0.801,0.787 and 0.917 in diagnosis of postoperative pul-monary infections in the general anesthesia orthopedic surgery patients,and the AUC of the joint detection of the indicators was greater than that of the single detection(P＜0.05).CONCLUSIONS The general anesthesia or-thopedic surgery patients with postoperative pulmonary infections show high expressions of serum TLR2 and TLR4 and the low expression of 25(OH)D.The joint detection of the above indexes has higher diagnostic value.The gram-negative bacteria are dominant among the pathogens causing the infections.The pathogens vary in the drug resistance.

Objective:To elucidate the clinical and genetic characteristics of familial cases with Glucose transporter type 1 deficiency syndrome (Glut1DS).Methods:A survey of family history was conducted on children (proband) with Glut1DS who had visited Peking University First Hospital between November 2008 and April 2024 by focusing on the clinical manifestations of family members. Peripheral venous blood (2 mL) was collected from the pediatric patients and their parents. Genomic DNA was extracted and sequenced subsequently. Sanger sequencing was performed to validate the identified variant sites of the SLC2A1 gene in the probands and their family members. The pathogenicity of suspected variants was analyzed according to the 2015 American College of Medical Genetics and Genomics (ACMG) Standards and Guidelines for the Interpretation of Sequence Variants. The clinical features, auxiliary examinations, and mutational characteristics of family members with SLC2A1 variants were analyzed. This study has been approved by the Clinical Research Ethics Committee of Peking University First Hospital (Ethics No. 2021 Research 332). Results:Among 87 cases with Glut1DS, 10 families with autosomal dominate inherited cases were identified, accounting for 11.0% of the cases. Of the 11 children, 8 were boys and 3 were girls. The onset of the disease had ranged from 3 months to 120 months (median 6 months), with 4 cases of early-onset classic type, 2 cases of late-onset classic type, and 5 cases of non-classic type. Six children had seizures, and 7 exhibited movement disorders. Seven children underwent developmental assessment, of which 3 had mild developmental delay, 2 were borderline, and 2 were normal. Nine children underwent lumbar puncture. The cerebrospinal fluid glucose levels ranged from 1.45 to 2.25 mmol/L (median 1.86 mmol/L), and the cerebrospinal fluid to blood glucose ratios ranged from 0.29 to 0.44 (median 0.35). Among the 8 fathers with SLC2A1 gene variants, 4 were asymptomatic, 2 developed paroxysmal exercise-induced movement disorders (PED) in childhood and adulthood, respectively, 1 had poor memory since childhood, 1 developed migraines during adolescence, and his sister was an asymptomatic carrier. The father with childhood-onset PED had a cerebrospinal fluid test with CSF glucose of 1.85 mmol/L. Of the 3 mothers with SLC2A1 gene mutations, 1 was an asymptomatic carrier; 2 developed PED in childhood and after the age of 20 respectively. The mother who developed PED in childhood also had psychomotor developmental delay. Genetic testing results revealed that among 10 families, 8 carried missense variants, 1 carried a nonsense variant, and 1 carried a small fragment insertion leading to a frameshift variant. Among the 11 cases, SLC2A1 gene variants in 8 children were inherited from their fathers, while in 3 cases, the variants were inherited from their mothers. The pathogenicity of the genetic variants was evaluated according to the Standards and Guidelines for the Interpretation of Sequence Variants published by the ACMG. Among the 8 variants identified in the 10 families, 4 were classified as pathogenic variants, 1 as likely pathogenic, and 3 as variants of uncertain significance (VUS). Four variant sites induding, c. 204_205insTCTC (p.V69fs), c. 412G>C (p.G138R), c. 431T>G (p.V144G), and c. 875A>G (p.Y292C), were not previously reported in the literature. Among these, the latter three were categorized as VUS. Conclusion:Familial Glut1DS account for 11.0% of the cases in China, with the majority of SLC2A1 gene variants inherited from the fathers, predominantly missense mutations, and with an autosomal dominant inheritance pattern. Probands tend to have earlier onset and more severe symptoms than their parents, who often present with mild or no symptoms.

Objective:To compare the effects of using cone beam CT (CBCT) and oral scanning registration on implant positional accuracy during robot-assisted implant surgery, and to provide a basis for selecting the appropriate registration for robot-assisted implant surgical options.Methods:One patient with dentition defect, specifically missing teeth at positions 21 and 26 and having natural teeth adjacent mesially and distally to the edentulous area, who visited the Department of Oral Implantology, School of Hospital of Stomatology Wuhan University in 2024 were selected. Based on reconstructed imaging data, 30 identical jaw models were printed. These models were divided into a CBCT registration group and an intra-oral scanning registration group (15 models per group). An associate chief physician with extensive experience in implant surgery performed preoperative registration using the implant robot and completed the implant surgeries. Postoperative CBCT scans were used to determine the three-dimensional position of the implants. The deviations between the planned implant position and the actual position were evaluated, including deviations at the implantation point, apical point, and angular deviation. The differences between the two groups were compared.Results:The implantation deviation was 0.675 (0.490) mm, apical deviation was (0.680±0.272) mm, and the angular deviation was 0.566°±0.147° in the CBCT registration group, and in the intra-oral scanning registration group, implantation deviation was 0.695 (0.313) mm, apical deviation was (0.667±0.217) mm, and the angular deviation was 0.523°±0.168°. There was no statistically significant error in implant precision between the two groups ( P>0.05). Conclusions:This in vitro experiment found that the use of intra-oral scanning registration in robot-assisted implant surgery can achieve similar implant placement accuracy as CBCT registration.

Objective:To compare minimally invasive surgery with traditional open surgery, analyze the current application status of health economic evaluations in the treatment of digestive tract cancers, such as esophageal cancer, gastric cancer, and colorectal cancer by minimally invasive surgery and provide evidence for the rational selection of clinical treatment, alleviation of disease-related economic burdens, and rational allocation of healthcare resources.Methods:By using five databases, i.e. China National Knowledge Infrastructure, Wanfang data, Chinese Biomedical Literature Database, PubMed, and Embase, a database was established to retrieve all the papers about health economic studies of minimally invasive surgery for esophageal cancer, gastric cancer, and colorectal cancer published until December 31, 2023. Literature was analyzed by using software NoteExpress 3.8, and data were processed using Excel 2021. The quality of included papers was evaluated using the CHEERS 2022 checklist, and Meta-analysis was conducted by using software Stata 17.0.Results:A total of 10 919 relevant papers were retrieved, and 59 studies were included. Only 14 studies (23.7%) used standard health economic evaluation methods. Meta-analysis results revealed no significant differences in direct medical expenditure and total expenditure between minimally invasive surgery and open surgery. However, the expenditure for minimally invasive surgery exhibited a significant increase [mean difference ( MD)=5 973.12 yuan, P<0.001], while hospital stay and indirect expenditure significantly decreased ( MD: -4.85 days and -733.79 yuan, P<0.001). In China, for gastric cancer, the direct medical expenditure of endoscopic surgery was lower than that of open surgery ( MD=-33 000.00 yuan) with no significant difference ( P<0.001). In colorectal cancer cases, the direct medical and surgical expenditures for laparoscopic surgery were higher than those for open surgery ( MD: 4 277.94 yuan and 4 267.80 yuan, P<0.001), while the indirect and total medical expenditures decreased ( MD: -768.34 yuan and -159.10 yuan). Hospital stays in patients who had minimally invasive surgery for all three types of cancer were shorter than those who had open surgery ( P<0.001). Conclusions:In the treatment of gastrointestinal cancer, compared with open surgery, minimally invasive surgery shows higher expenditure, but has advantages, such as shorter hospital stay and lower indirect expenditure, and there were no significant differences in direct medical and total expenditures between the two approaches. When conducting health economic evaluation, factors such as postoperative complications, hospital stay, and patient's economic status should be considered for their impact on total medical expenditure. It is necessary to pay attention to the application of health economic evaluations in healthcare decision-making.

Objective:To study the association between subtypes of hypertension and risk for all-cause mortality and cardio-cerebrovascular mortalities in the elderly in communities of Beijing.Methods:The data were collected from the Beijing Healthy Aging Cohort Study. The elderly in 5 urban areas (former Xicheng, former Xuanwu, Fangshan, Haidian and Tongzhou) and 4 rural areas (Yanqing, Miyun, Huairou and Daxing) in Beijing were selected as the study subjects by multi-stage stratified cluster random sampling. The baseline survey was conducted from July 2009 to September 2015. The follow-up was conducted until March 31, 2021, and a total of 6 326 participants were enrolled. Cox proportional hazard regression model was used to analyze the association of SBP, DBP, normal blood pressure, high normal blood pressure, simple systolic hypertension, simple diastolic hypertension and systolic diastolic hypertension with all-cause mortality and cardio-cerebrovascular mortality.Results:By March 31, 2021, the median follow-up time was 6.30 years, the all-cause mortality density was 201.67/10 000 person-years, and the cardio-cerebrovascular mortality density was 90.20/10 000 person-years. Multivariate Cox proportional hazard regression model analysis showed that the risk for all-cause mortality increased by 5.6% ( HR=1.056, 95% CI: 1.020-1.092), and the risk for cardio-cerebrovasculvar mortality increased by 12.5% ( HR=1.125, 95% CI: 1.071-1.182) for every 10 mmHg increase in SBP. The risk for all-cause mortality increased by 8.6% ( HR=1.086, 95% CI: 1.023-1.152), and the risk for cardio-cerebrovascular mortality increased by 19.9% ( HR=1.199, 95% CI: 1.101-1.306) for every 10 mmHg increase in DBP. Compared with the normal blood pressure group, the risk for all-cause mortality increased by 64.8% ( HR=1.648, 95% CI: 1.049-2.591) and the risk for cardio- cerebrovascular mortality increased by 112.8% ( HR=2.128, 95% CI: 1.069-4.233) in the simple diastolic hypertension group. The risk for all-cause mortality increased by 34.4% ( HR=1.344, 95% CI: 1.023-1.467) and the risk for cardio-cerebrovascular mortality increased by 111.3% ( HR=2.113, 95% CI: 1.384-3.225) in the systolic-diastolic hypertension group. Conclusions:In the elderly in communities of Beijing, beside systolic-diastolic hypertension. It is necessary to pay close attention to the simple diastolic hypertension, which has lower prevalence, and give targeted prevention and treatment.

Objective:To investigate the association of urinary albumin-to-creatinine ratio (UACR) in normal range and cardiovascular health (CVH) score with cardio-cerebroascular and all-cause mortality in urban elderly residents in Beijing.Methods:Based on Beijing Healthy Aging Cohort Study, data from 1 817 elderly participants with normal UACR were analyzed, and UACR was used as both continuous and categorical variable. "Life's Essential 8" was used to develop a CVH score. Statistical analysis was performed by using software SPSS 26.0 and R 4.2.1, including two independent samples t-test, χ2 test and non-parametric test. Multivariable Cox proportional hazard regression models stratified by CVH score and the restricted cubic spline were used to analyse the association of UACR with the risk for cardio-cerebrovascular and all-cause mortality. Results:By March 31, 2021, the median follow-up time M ( Q1, Q3) was 11.28 (10.84, 11.36) years, a total of 308 deaths were recorded during follow-up, and the mortality density was 163.87/10 000 person-years. The age of the participants was (71.4±6.6) years, and 1 070 participants were women. The results showed that the risk for both cardio-cerebrovascular disease and all-cause mortality decreased with the decrease of UACR in the low CVH score group ( HR=0.500, 95% CI: 0.341-0.734; HR=0.793, 95% CI: 0.647-0.971), and in the high CVH score group, there was a decreasing trend in the risk for cardio-cerebrovascular mortality with the decrease of UACR ( HR=0.665, 95% CI: 0.447-0.990). Compared with the participants with low CVH score and higher UACR, the risk for cardio-cerebrovascular and all-cause mortality decreased by 68.9% and 45.6%, respectively, in the participants with high CVH score and lower UACR ( HR=0.311, 95% CI: 0.131-0.739; HR=0.544, 95% CI: 0.360-0.822), and the risk for all-cause mortality decreased by 26.7% in the participants with high CVH score and higher UACR ( HR=0.733, 95% CI: 0.542-0.993). Conclusions:In urban elderly residents in Beijing, higher UACR were associated with a significantly increased risk for cardio-cerebrovascular and all-cause mortality, and in the low-CVH score group, decreased UACR was protective factor against cardio-cerebrovascular and all-cause mortality. The combined effect of cardiovascular health status and normal UACR had a greater protection against the risk for cardio-cerebrovascular and all-cause mortality.

Objective:To describe and analyze the distribution characteristics of a body shape index (ABSI) and its association with all-cause mortality in the elderly in communities of Beijing.Methods:The baseline and follow-up data of 10 423 subjects in Beijing Healthy Aging Cohort Study were used. The endpoint outcome was all-cause mortality in follow-up. The associations of ABSI, BMI and waist circumference with all-cause mortality were analyzed with a Cox proportional hazard regression model.Results:The mean follow-up time in this cohort was 6.36 years and the all-cause mortality was 15.27%. The restricted cubic spline results showed that ABSI showed a "U" association with all-cause mortality in the total population. After adjustment for confounding factors, compared with the normal ABSI group, the risk for all-cause mortality was higher in the low ABSI group ( HR=1.69, 95% CI: 1.32-2.17) and high ABSI group ( HR=1.75, 95% CI: 1.40-2.18). Compared with the normal BMI group, low BMI was associated with an increased risk for all-cause mortality ( HR=1.70, 95% CI: 1.40-2.08). Compared with the normal waist circumference group, central obesity was associated with a decreased risk for all-cause mortality ( HR=0.89, 95% CI: 0.80-0.99). The results of gender specific analysis on ABSI association with all-cause mortality were similar to that in total population, but the strength of association varied among different age groups. Cross-grouping results showed that older people with normal BMI and waist circumference and abnormal ABSI had an increased risk of all-cause mortality ( HR=1.91, 95% CI: 1.22-2.99). Conclusions:In Beijing, ABSI is correlated with the risk for all-cause mortality in the elderly population, which is more sensitive and specific than the traditional indicators (BMI and waist circumference). They can be used as one of the physical measurement indicators for all-cause mortality risk prediction in the health risk assessment and management in the elderly.

Objective:To explore the association between physical activity (PA) level and all-cause mortality in the elderly with different obesity status in Beijing.Methods:The study subjects were from the Cardiovascular and Cognitive Healthy Study in Middle-Aged and Elderly Residents of Beijing, a total of 3 746 individuals aged ≥60 years in the baseline survey between 2013 and 2015 were included in the study. Questionnaire survey and physical examination were conducted to collect the basic information of the individuals. The metabolic equivalent of the elderly was used to calculate the PA level, and an international PA questionnaire was used to determine the PA intensity. BMI, waist-to-height ratio (WHtR) and a body shape index (ABSI) were used to evaluate individuals' obesity status. The distribution of different PA levels under different obesity states was described by using bar chart and cumulative percentage bar chart. Cox proportional hazard regression model was used to analyze the correlation between the level of PA and all-cause mortality risk in different types of obesity status.Results:By December 31, 2019, the median follow-up time was 5.46 years, and the mortality density was 244.55/10 000 person-years. Compared with the individuals in high-PA intensity group, the all-cause mortality risk increased by 41% ( HR=1.41, 95% CI: 1.14-1.76) and 122% ( HR=2.22, 95% CI: 1.76-2.81), respectively, in moderate and low PA intensity groups. Compared with the individuals in high-PA intensity-high-obesity group, based on the BMI, the all-cause mortality risk increased by 85% ( HR=1.85, 95% CI: 1.08-3.16) and 250% ( HR=3.50, 95% CI: 2.01-6.10) in those in moderate-PA intensity-high-obesity group and in low-intensity-high-obesity group. Based on the WHtR, the all-cause mortality risk increased by 53% ( HR=1.53, 95% CI: 1.02-2.29) and 218% ( HR=3.18, 95% CI: 2.09-4.86), respectively, in those with moderate-PA intensity-high-obesity and those with low-PA intensity-high-obesity. According to the ABSI, the all-cause mortality risk increased by 120% in those in low-PA intensity-high-obesity group ( HR=2.20, 95% CI: 1.53-3.18). Based on any one of the indicators BMI, WHtR, or ABSI, there was no significant difference in all-cause mortality risk between high-intensity-moderate-obesity group and high-intensity-low-obesity group (all P>0.05). Conclusions:PA level is closely associated with the all-cause mortality risk in the elderly in communities of Beijing. Increasing PA level can not only reduce the all-cause mortality risk but also reduce even eliminate the excess all-cause mortality risk associated with obesity.

Objective:To analyze the association between different lipid indicators and cardio-cerebrovascular mortality in the elderly, identify a better lipid indicator for the risk for cardio-cerebrovascular mortality in the elderly in communities of Beijing.Methods:The elderly aged 60 years and above were selected from the Beijing Healthy Aging Cohort Study from July 2009 to September 2015. Remnant cholesterol (RC) was derived by calculation. The Cox proportional hazard regression models determined the HR and 95% CI incidating the associations of baseline different lipid indicators with cardio-cerebrovascular and all-cause mortality. Results:By March 31, 2021, with a mean follow-up duration of 6.88 years in the study population, a total of 492 cardio-cerebrovascular deaths and 1 056 all-cause deaths wre recorded. The HR values indicating the association between LDL-C, HDL-C, and RC were 0.87 (95% CI: 0.78-0.97), 0.46 (95% CI: 0.35-0.62), and 1.29 (95% CI: 1.14-1.45) for cardio-cerebrovascular mortality, respectively, and 0.93 (95% CI: 0.86-1.00), 0.66 (95% CI: 0.55-0.80) and 1.22 (95% CI: 1.12-1.33) for all-cause mortality. The associations of RC and HDL-C with cardio-cerebrovascular mortality were consistent regardless of hyperlipidemia in the elderly. Subgroup analyses showed that elevated RC was associated with increased risk for cardio-cerebrovascular mortality, and elevated HDL-C was associated with decreased risk for cardio-cerebrovascular mortality in the elderly in different gender, age, smoking status, drinking status, and diabetes status groups in communities of Beijing. Conclusion:RC might be a better potential lipid indicator for the risk for cardio-cerebrovascular mortality in the elderly in communities compared with traditional cholesterol indicator.