Objective:To investigate the potential causal relationships between gut microbiota composition and the risk of developing various subtypes of breast cancer by using bidirectional two-sample Mendelian randomization(MR).Methods:The research utilized genome-wide association studies(GWAS) data on gut microbiota from the MiBioGen database and GWAS data on breast cancer from the Breast Cancer Association Consortium (BCAC). In this MR study, inverse variance weighted (IVW), weighted median, MR Egger, and MR-PRESSO methods were used. Additionally, reverse MR and stratified analyses were conducted to assess reverse causality and the impact on different subtypes of breast cancer.Results:Adlercreutzia (IVW OR=0.92, 95% CI: 0.87-0.98, P=0.01) and Parabacteroides (IVW OR=0.87, 95% CI: 0.79-0.96, P=0.007) exhibited a statistically significant protective effect on breast cancer. Conversely, Sellimonas (IVW OR=1.05, 95% CI: 1.01-1.09, P=0.01) was significantly associated with an increased risk of breast cancer. Desulfovibrio (IVW OR=0.94, 95% CI: 0.88-1.00, P=0.04) and Ruminococcaceae (UCG013) (IVW OR=0.92, 95% CI: 0.86-0.99, P=0.03) presented suggestive protective effects against breast cancer. Furthermore, stratified analysis revealed that the protective effect of Adlercreutzia against breast cancer persisted in the estrogen receptor(ER)-positive subtypes, while Desulfovibrio persisted in the ER-negative subtypes. Sellimonas was causally associated with the risk of ER-positive subtypes. CACNA1S was identified as the functional gene of Adlercreutzia, and associated with favorable prognosis in breast cancer, while ERBB4 was identified as the functional gene of Sellimonas and associated with poor prognosis in breast cancer. Conclusions:This study identifies the causal relationships between gut microbiota and breast cancer, suggesting a novel target for early clinical intervention and treatment, with potential implications for future functional analysis.

Objective To evaluate the value of deep learning(DL)-based artificial intelligence(AI)automatic reconstruction for evaluation of grafts in patients who underwent coronary artery bypass grafting(CABG).Methods Coronary CT angiography data of 90 patients who underwent CABG with a total of 197 grafts were retrospectively analyzed.Taken manual evaluation results(manual group)as the standards,the efficacy of AI(AI group)for evaluating the degree of stenosis of graft and distal autologous blood vessels were assessed.The consistency between calculating unprotected coronary territory(UCT)and the total time for image post-processing and diagnosis were compared between groups.Results AI group showed average consistency with manual group for evaluating the number of grafts([intra-class correlation coefficient,ICC]=0.743,P＜0.05),average to excellent for evaluating the maximum degree of graft stenosis(Kappa=0.310-1.000,all P＜0.05),also average to good consistency for evaluating the maximum degree of stenosis of the native vessel distal to the graft insertion(Kappa=0.292-0.795,all P＜0.05).AI group had moderate consistency with manual group for UCT(ICC=0.469,P＜0.05),achieved an area under the curve of 0.811.The overall time of image post-processing and diagnosis in AI group were both significantly shorter than that in manual group(P＜0.05).Conclusion Having acceptable consistency with manual evaluation and ability for assistant,AI was efficient for automatic reconstructing coronary artery bypass graft and quantifying the degree of graft stenosis.

Objective To propose and validate a decomposition method based on half-projection reconstruction for dual-energy cone beam CT(DE CBCT),thereby providing a potentially feasible low-dose imaging solution for anatomical monitoring and dose reconstruction optimization in adaptive radiotherapy.Methods Dual-energy scans were performed on a Gammex phantom using the on-board kilovoltage CBCT system of a VitalBeam accelerator at acquisition frame rates of 15 and 7 frames per second(f/s).Images were reconstructed from the projection data,and dual-energy decomposition was applied to the 7 f/s dual-energy images to derive relative electron density(RED)and stopping power ratio(SPR)using weighted formulas and empirical functions,followed by accuracy evaluation.Additionally,the weighted CT dose index was calculated for different scanning parameters.Results Dual-energy decomposition effectively suppressed image artifacts,with RED and SPR errors remaining below 2.82%and 2.56%,respectively.Compared with the traditional dual-scan method which required high-and low-energy acquisitions,the weighted CT dose index of the half-projection DE CBCT was reduced by 11.60 mGy(a 52.90%reduction).Furthermore,it was 2.58 mGy lower than the dose of the full-projection high-energy CBCT alone(a 19.98%reduction)and only 1.31 mGy higher than that of the low-energy CBCT(a 14.52%increase).Conclusion The proposed method effectively suppresses image artifacts while maintaining high accuracy in RED and SPR under low radiation dose conditions,demonstrating its potential value for scenarios requiring frequent image guidance,such as adaptive radiotherapy.

Objective To validate the diagnostic performance and risk stratification ability of an AI-based recognition system(PE-AI)for pulmonary embolism(PE)using computed tomography pulmonary angiography(CTPA)so as to analyze its diagnostic value in clinical practice.Methods A total of 416 patients with suspected PE who underwent CTPA from January 1,2023 to December 10,2023 at our hospital were included in this study.Two junior radiologists and PE-AI separately detected and diagnosed emboli in the collected cases by double-blind method,and recorded the diagnosis time respectively.Three senior radiologists reviewing with clinical follow-up results were used as the gold standard in this study.Diagnostic performance was evaluated by using the receiver operating characteristic(ROC)curve analysis and Delong-t test.For positive cases,the pulmonary artery obstruction index(PAOI)calculated by AI and manually were collected respectively and consistency analysis was performed.Results The area under the curve(AUC)of PE-AI,manual and combined diagnosis was 85.6％,90.8％and 95.1％,respectively,which differed significantly(P＜0.05).The reading time of PE-AI[(0.16±0.07)min]was significantly lower than the time of manual[(4.42±1.85)min,P＜0.001]and combined diagnosis[(4.58±1.84)min,P＜0.001].The PAOI measured by PE-AI and manually had high consistency(intraclass correlation efficient,ICC=0.80)in the subgroup analysis of confirmed cases.Conclusion AI can quickly identify pulmonary artery emboli in a short time and assist radiologists to improve diagnostic efficiency.At the same time,through the intelligent detection of PAOI,it is helpful for the risk stratification of patients with PE and optimizing the diagnosis and treatment pathway for pulmonary embolism.

Objective To observe the value of training about ovarian-adnexal reporting and data system((O-RADS)for risk stratification of adnexal lesions diagnostic efficacy of different seniority physicians before and after training.Methods A total of 575 O-RADS 1-5 point lesions from 470 patients who received non-contrast enhanced pelvic MR and dynamic contrast-enhanced MRI(DCE-MRI)were retrospectively included.The lesions were scored by 1 junior radiologist(radiologist A)and 1 senior radiologist(radiologist B)independently according to O-RADS risk stratification,and the results were recorded as R1 and R2,respectively.Three months later,the lesions were rescored by radiologist A and B after receiving systematic training from gynecological imaging experts,and the results were recorded as R11 and R22,respectively.Two gynecological imaging experts conducted a consensus scoring,and the results were recorded as R0.Taken O-RADS score＞3 as the criterion for malignant,the diagnostic efficacy of radiologist A and B before and after training were evaluated,while taken R0 as the reference,the intra-observer,inter-observer consistency between radiologist A and B,as well as their consistency with R0 were calculated.Results The diagnostic sensitivity and specificity of R0 was 95.21%and 93.14%,respectively.The diagnostic sensitivity of radiologist A before and after training was 92.22%and 95.21%,with specificity of 83.33%and 89.46%,respectively.For radiologist B,the sensitivity before and after training was 95.81%and 95.21%,with specificity of 92.89%and 91.91%,respectively.Good intra-observer consistency of O-RADS score was observed both in radiologist A and B,with Kappa value of 0.845 and 0.884,respectively,which also noticed between radiologist A and B,with Kappa value of 0.761,and the Kappa value of R1,R2 and R0 was 0.781 and 0.911,respectively.After training,the inter-observer consistency of radiologist A and B increased,and Kappa value of R11,R22 and R0 was 0.844 and 0.915,respectively.Conclusion Training about O-RADS for risk stratification was helpful to improving diagnostic specificity of benign and malignant adnexal lesions,especially for junior radiologists.

Objective To explore the value of apparent diffusion coefficient(ADC)value combined with ovarian-adnexal reporting and data system(O-RADS)MRI score in differentiating benign and malignant adnexal lesions with score 3-5.Methods The imaging data of 241 adnexal lesions with O-RADS MRI score 3-5 proved by pathology were analyzed retrospectively.The ADC values of all lesions were measured,and the optimal thresholds were determined by receiver operating characteristic(ROC)curve analysis,the comprehensive model was established using binary logistic regression analysis,the corresponding diagnostic efficacy was calculated.Results The median ADC values of the benign,borderline and malignant groups were 2.166 × 10-3 mm2/s,1.383 ×10-3 mm2/s and 0.839× 10-3mm2/s,respectively(P＜0.05).The area under the curve(AUC)of the combination of ADC value with O-RADS MRI score for differentiating benign and malignant adnexal lesions was 0.928[95％ confidence interval(CI)0.888-0.958],which was higher than that of O-RADS MRI score and ADC value alone(P＜0.05).The sensitivity and specificity of the three models in differ-entiating benign and malignant adnexal lesions were 93.5％,98.9％,83.9％,respectively;and 79.1％,58.8％,79.1％,respectively.Conclusion ADC value combined with O-RADS MRI score can be the most effective in the diagnosis of benign and malignant adnexal lesions,which is higher than O-RADS MRI score and ADC value.Compared with O-RADS MRI score,ADC value combined with O-RADS MRI score maintained good sensitivity and increased diagnostic specificity.

Objective:To achieve the conversion from dual-energy cone-beam CT (DECBCT) at the kilovolt (KV) level to projections at the megavolt (MV) level using an improved CycleGAN network, in order to provide a potential reference benchmark and real-time monitoring of in vivo doses delivered by exit beams for the safe implementation of advanced techniques such as online adaptive radiotherapy. Methods:Simulated patient data were generated using a 4D extended cardiac torso (XCAT) model, and projections were generated based on the geometric parameters of Varian′s onboard cone-beam CT. Furthermore, relative electron density (RED) images were derived from DECBCT images using an iterative dual-energy decomposition algorithm. The SE-CycleGAN and CycleGAN networks were trained to generate MV projection images using DECBCT projections and RED images, respectively. The performance of both methods was evaluated using metrics including structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), and root mean square error (RMSE).Results:SE-CycleGAN significantly outperformed CycleGAN in all evaluation metrics ( Z = -23.92, -26.17, -25.54, -26.80, -11.54, -11.21, P<0.05), particularly in learning global information. Besides, although both methods generated satisfactory MV projections, training using DECBCT projections as input yielded better effects than training using RED images. For all the 3 636 sets of projections in the test set, the SE-CycleGAN and CycleGAN networks using DECBCT projections as input respectively yielded SSIMs of 0.997 7±0.000 7 and 0.997 1±0.001 6, PSNRs of 39.625 0±4.684 4 and 36.272 2±5.566 3, and RMSEs of 0.004 1±0.002 7 and 0.006 3±0.0043, respectively. In contrast, the SE-CycleGAN and CycleGAN networks using RED projections as input respectively yielded SSIMs of 0.996 8±0.001 0 and 0.996 2±0.001 5, PSNRs of 38.548 7±3.637 4 and 36.007 3±4.437 8, and RMSEs of 0.004 3±0.002 2 and 0.006 1±0.0037, respectively. Conclusions:This study proposed a new method to establish reference benchmarks for in vivo dose monitoring based on DECBCT and deep learning technologies. This method is accurate and effective according to the preliminary validation using virtual simulation experiments.

Objective:To apply statistical process control (SPC) techniques to the quality assurance of non-normal radiotherapy plans through Johnson transformation, establishing patient-specific tolerance and action limits based on treatment sites and dose/distance assessment criteria, thereby enhancing the intensity-modulated radiation therapy (IMRT) verification accuracy and dose delivery precision.Methods:In this study, 951 gamma analysis data of patient-specific quality assurance (PSQA) executed on the Halcyon accelerator platform were selected and categorized into six groups based on treatment sites, including brain (102 cases), head and neck (100 cases), breast (229 cases), lung (154 cases), esophagus (223 cases), and pelvic (143 cases) groups. The six groups of data were statistically analyzed through Anderson-Darling normality tests ( α = 0.05) using Minitab 21 software. Non-normal data were transformed into normal data through Johnson transformation and then were used to establish treatment site-specific tolerance and action limits, which were compared with the Shewhart control charts based on normal distributions. Results:The PSQA result of the six groups all exhibited non-normal distributions ( P < 0.05). Through Johnson transformation, the tolerance and action limits for the head and neck, breast, lung, esophagus, and pelvic areas under the 3%/2 mm criterion ranged from 95.13% to 96.16% and 94.19% to 95.91%, respectively. In contrast, the tolerance and action limits ranged from 91.15% to 94.86% and 89.94% to 94.78% under the 2%/2 mm criterion. Directly applying Shewhart control charts without normality assumptions yielded higher tolerance limits compared to the application of Johnson transformation, increasing the false positive rate in the non-normal PSQA process. Conclusions:Applying the SPC techniques directly to a non-normal process can lead to an increased false alarm rate and wrong process interpretation. The SPC techniques combined with Johnson transformation enable more effective monitoring of a non-normal PSQA process, facilitating timely identification of potential factors that may lead to an out-of-control process based on the treatment site-specific limits.

OBJECTIVE: Humans are exposed to complex mixtures of environmental chemicals and other factors that can affect their health. Analysis of these mixture exposures presents several key challenges for environmental epidemiology and risk assessment, including high dimensionality, correlated exposure, and subtle individual effects. METHODS: We proposed a novel statistical approach, the generalized functional linear model (GFLM), to analyze the health effects of exposure mixtures. GFLM treats the effect of mixture exposures as a smooth function by reordering exposures based on specific mechanisms and capturing internal correlations to provide a meaningful estimation and interpretation. The robustness and efficiency was evaluated under various scenarios through extensive simulation studies. RESULTS: We applied the GFLM to two datasets from the National Health and Nutrition Examination Survey (NHANES). In the first application, we examined the effects of 37 nutrients on BMI (2011-2016 cycles). The GFLM identified a significant mixture effect, with fiber and fat emerging as the nutrients with the greatest negative and positive effects on BMI, respectively. For the second application, we investigated the association between four pre- and perfluoroalkyl substances (PFAS) and gout risk (2007-2018 cycles). Unlike traditional methods, the GFLM indicated no significant association, demonstrating its robustness to multicollinearity. CONCLUSION GFLM framework is a powerful tool for mixture exposure analysis, offering improved handling of correlated exposures and interpretable results. It demonstrates robust performance across various scenarios and real-world applications, advancing our understanding of complex environmental exposures and their health impacts on environmental epidemiology and toxicology.
Humans ; Environmental Exposure/analysis* ; Linear Models ; Nutrition Surveys ; Environmental Pollutants ; Body Mass Index