Objective To evaluate the efficiency of intra-tumor and peri-tumor ultrasound radiomics models based on machine learning algorithms for predicting benign and malignant Breast Imaging Reporting and Data System (BI-RADS) 4 breast lesions, and provide insights into early diagnosis of breast cancer. Methods A retrospective analysis was conducted based on the medical records of 450 female patients who underwent breast ultrasound examination in the Affiliated Cancer Hospital of Xinjiang Medical University from June 2020 to April 2022. The patients were divided into the benign (n = 199) and malignant (n = 195) groups according to pathological examination, and randomized into the training (n = 275) and validation (n = 119) sets at a ratio of 7∶3. Radiomics features were extracted and screened. Intra-tumor, peri-tumor, and intra-tumor + peri-tumor ultrasound radiomics models were constructed based on three machine learning algorithms, including logistic regression (LR), support vector machine (SVM), and multi-layer perceptron (MLP). Receiver operating characteristics (ROC) curves, calibration curves, and decision curves were plotted to evaluate the efficacy of the radiomics models for prediction of benign and malignant breast lesions. Results A total of 17 intra-tumor, 16 peri-tumor, and 17 intra-tumor + peri-tumor radiomics features were selected for model construction. Based on LR, MLP, and SVM algorithms, the intra-tumor + peri-tumor radiomics models showed higher predictive efficacy than intra-tumor and peri-tumor radiomics models. The predictive efficacy of intra-tumor, peri-tumor, and intra-tumor + peri-tumor radiomics models were higher based on the SVM algorithm than based on LR and MLP algorithms. For the intra-tumor radiomics model based on the SVM algorithm, the area under the ROC curve (AUC), accuracy, sensitivity, and a specificity were 0.909, 0.851, 0.860, and 0.842, respectively, in the training set and 0.866, 0.832, 0.847, and 0.817, respectively, in the validation set. For the peri-tumor radiomics model based on the SVM algorithm, these values were 0.899, 0.855, 0.882, and 0.827, respectively, in the training set and 0.844, 0.815, 0.847, and 0.783, respectively, in the validation set. For the intra-tumor + peri-tumor radiomics model based on the SVM algorithm, these values were 0.943, 0.876, 0.860, and 0.892, respectively, in the training set and 0.881, 0.849, 0.915, and 0.783, respectively, in the validation set. Conclusion The intra-tumor and peri-tumor ultrasound radiomics models based on machine learning algorithms are highly valuable for prediction of benign and malignant BI-RADS 4 breast lesions. The intra-tumor + peri-tumor ultrasound radiomics model based on the SVM algorithm has the optimal efficacy for prediction of benign and malignant BI-RADS 4 breast lesions.

Objective:To explore the correlation between serum ferritin (SF) and risk of lean non-alcoholic fatty liver disease (NAFLD), so as to provide the basis for the prevention and treatment of lean NAFLD.Methods:A total of 7 187 people without NAFLD at baseline who took at least 2 physical examinations in the Health Management Center of the Second Hospital of Dalian Medical University from January 2014 to December 2023 and met the selection criteria were selected as the research subjects, and all the subjects had no NAFLD at baseline. Subjects were divided into four groups according to baseline SF quartiles: 1 797 cases in the first quartile ( Q1) group, 1 797 cases in the second quartile ( Q2) group, 1 797 cases in the third quartile ( Q3) group, and 1 796 cases in the fourth quartile ( Q4) group. The incidence of lean NAFLD in each group were observed. Kaplan-Meier curve was plotted to calculate the cumulative incidence of lean NAFLD which compared by log-rank test. Cox proportional hazard regression model was used to analyze the correlation between SF and new-onset lean NAFLD, Q1, Q2, Q3 and Q4 of SF were taken as continuous variables into the model for trend test.The stability of the results was verified by two item sensitivity analyses. Time-dependent receiver operating characteristic curve (ROC) was plotted to evaluate the predictive value of SF for the onset of lean NAFLD. Results:The cumulative follow-up were 25 076 person-years. There were 230 new cases of lean NAFLD, and the incidence density was 9.172/1 000 person-years. The incidence densities of lean NAFLD in Q1, Q2, Q3 and Q4 groups were 6.915/1 000 person-years, 8.552/1 000 person-years, 9.641/1 000 person-years, 12.003/1 000 person-years, respectively. Kaplan-Meier curve indicated that the incidence of lean NAFLD was increased with the increment of SF, and the difference was statistically significant (log-rank test, χ2=9.92, P=0.019). Cox proportional hazard regression model results showed that the risk of developing lean NAFLD in Q4 group increased by 72.8% ( HR=1.728, 95% confidence interval (95% CI): 1.059 to 2.820) compared with Q1 group. Trend analysis revealed that the risk of lean NAFLD increased by 18.9% for each one-quartile increase of SF( HR=1.189, 95% CI: 1.012 to 1.396). Two sensitivity analyses indicated that the risk of NAFLD in Q4 group was 1.795 times ( HR=1.795, 95% CI: 1.083 to 2.975) or 1.654 times ( HR=1.654, 95% CI: 1.022 to 2.678) higher than that in Q1 group. The area under the curve (95% CI) of SF for predicting the incidence of lean NAFLD at 2-, 3-, 7- and 8-year follow-up based on time-dependent ROC were 0.645 (0.593 to 0.698), 0.652 (0.603 to 0.700), 0.605 (0.539 to 0.672) and 0.716 (0.597 to 0.836), respectively. Conclusion:SF is an independent risk factor for lean NAFLD and has predictive value for the new-onset of lean NAFLD.

Purpose To explore the value of conventional ultrasound combined with ultrasound radiomics in detecting triple-negative breast cancer(TNBC).Materials and Methods In this study,we retrospectively collected the preoperative conventional ultrasound images and clinic-pathological data of 682 patients with lumpy breast cancer from the Affiliated Cancer Hospital of Xinjiang Medical University from January 2015 to December 2021.All patients included 325 cases with TNBC and 357 cases with non-TNBC.Features were extracted from preoperative ultrasound images of patients,screened,dimensionally reduced,and a combined model was constructed with ultrasound features and clinicopathological features.The performance of the three single models and combined models was evaluated by receiver operating characteristic curve analysis.Results The area under the curve of conventional ultrasound model,clinicopathological model,ultrasound imaging omics and the combined model in training group and validation suite were 0.730,0.718,0.982,0.985,0.726,0.683,0.981 and 0.982,respectively.The area under the curve of TNBC detected by the combined model was significantly higher than that of the single model(Z=7.311,7.024,3.883,all P<0.01).Conclusion The combined model based on ultrasonography has good diagnostic efficacy in detecting TNBC.As a non-invasive examination,it can avoid unnecessary biopsy to the patient.

Objective:To investigate the diagnostic value of different conventional ultrasound-based radiomics models and their combination with clinical ultrasound features in the pathological subtyping of renal cell carcinoma.Methods:Retrospective data from 286 patients diagnosed with renal cell carcinoma by pathology at the Tenth Affiliated Hospital of Southern Medical University between May 1，2017 and June 7，2024 were collected. Among the 286 patients，203 were clear cell carcinoma，44 were papillary renal cell carcinoma，and 39 were chromophobe renal cell carcinoma. The patients were randomly divided into a training group（201 cases）and a validation group（85 cases）in a ratio of 7 to 3. Regions of interest（ROI）were delineated on conventional ultrasound images，and the radiomics features were extracted. Feature selection was performed using Student's t-test，Pearson correlation，and the least absolute shrinkage and selection operator（LASSO）. Six different machine learning methods included category gradient boosting（CatBoost），light gradient boosting machine（LightGBM），Logistic regression（LR），random forest（RF），support vector machine（SVM）and extreme gradient boosting（XGBoost）were used to establish radiomics models. Weight balancing was applied to correct for sample imbalance，and an imaging genomics model was constructed after balancing the samples. Independent predictors of renal cell carcinoma subtyping were selected from clinical ultrasound features using univariate and multivariate logistic regression analyses，and a clinical imaging model was constructed. The best-performing radiomics model was combined with the clinical independent predictors to construct a combined model. Receiver operating characteristic curves and the obuchowski index were plotted to evaluate model performance. Results:Among the radiomics models，the model constructed using Random Forest（RS RF）after balancing the samples exhibited the best predictive performance，with area under the curve（AUCs）of 0.918（micro-average ROC）and 0.903（macro-average ROC），and the obuchowski index was 0.885 in the validation group. The long and short axes of ultrasound image tumor masses were used as imaging independent predictors to construct a clinical imaging model. In the validation group，the AUCs of the clinical model were 0.886（micro-average ROC）and 0.606（macro-average ROC），and the obuchowski index was 0.569. The combined model achieved AUCs of 0.888（micro-average ROC）and 0.967（macro-average ROC），with an obuchowski index of 0.933，outperforming any single model. Conclusions:The combination of conventional ultrasound-based radiomics models with clinical ultrasound features demonstrates high diagnostic value in differentiating clear cell carcinoma，papillary renal cell carcinoma，and chromophobe renal cell carcinoma. It may serve as an auxiliary tool for providing timely and effective clinical guidance.

Purpose To investigate the distribution characteristics of tumor-associated macrophages(TAM)in different regions of breast cancer with enhanced contrast-enhanced ultrasound(CEUS),and to further explore the relationship between TAM and CEUS indicators and clinicopathology in different regions of breast cancer.Materials and Methods A total of 119 patients with suspected breast cancer admitted to the Cancer Hospital Affiliated to Xinjiang Medical University from March 2021 to March 2023 were prospectively included.CEUS was applied to the tumor,and ultrasound-guided puncture biopsy was also taken.The lesions diagnosed as breast cancer by pathology was outlined the central area,marginal area and normal area next to the cancer,and was obtained the time intensity curves of different areas.The tissues were taken for immunohistochemistry and flow cytometry,and the TAM cells were stained and distinguished.The characteristics of TAM in different regions of breast cancer and its correlation with clinical pathology were analyzed,respectively.Results By immunohistochemistry and flow cytometry,there were significant differences in the number of TAM infiltration in the border area,central area and adjacent area of breast cancer(immunohistochemistry:F=382.326,P<0.05;flow cytometry:F=24.955,P<0.05).The characteristics of CEUS in three different regions showed that the TAM in the central region of breast cancer increased when filling defect appeared(t=2.631,P<0.05),but the TAM in the peripheral region was more(t=2.999,P<0.05).After angiography,lesions showed high perfusion,and there was significantly more TAM in the edge and central area of the lesion than that in normal area next to the cancer(t=5.529,P<0.05;t=2.584,P<0.05).Clinical stage was related to the TAM in three regions.When the clinical stage was high,there were more TAM in all three regions(t=6.658,2.367,2.400,all P<0.05).Histological grading was high,and TAM in all three areas was high(F=101.151,16.922,26.822,all P<0.05).Conclusion There was a decreasing trend of TAM in the marginal area,central area and adjacent tissues of breast cancer during CEUS.The edge region has more malignant CEUS characteristics than that in the central region and the normal region adjacent to cancer;and the number of TAM is more in breast cancer with late clinical grading,poor tissue differentiation and obvious contrast-enhanced ultrasound malignant characteristics.The distribution characteristics of TAM represent the malignant degree and metastatic probability of breast cancer to a certain extent,and TAM is a factor related to the invasion of breast cancer.

Purpose To investigate the distribution characteristics of tumor-associated macrophages(TAM)in different regions of breast cancer with enhanced contrast-enhanced ultrasound(CEUS),and to further explore the relationship between TAM and CEUS indicators and clinicopathology in different regions of breast cancer.Materials and Methods A total of 119 patients with suspected breast cancer admitted to the Cancer Hospital Affiliated to Xinjiang Medical University from March 2021 to March 2023 were prospectively included.CEUS was applied to the tumor,and ultrasound-guided puncture biopsy was also taken.The lesions diagnosed as breast cancer by pathology was outlined the central area,marginal area and normal area next to the cancer,and was obtained the time intensity curves of different areas.The tissues were taken for immunohistochemistry and flow cytometry,and the TAM cells were stained and distinguished.The characteristics of TAM in different regions of breast cancer and its correlation with clinical pathology were analyzed,respectively.Results By immunohistochemistry and flow cytometry,there were significant differences in the number of TAM infiltration in the border area,central area and adjacent area of breast cancer(immunohistochemistry:F=382.326,P<0.05;flow cytometry:F=24.955,P<0.05).The characteristics of CEUS in three different regions showed that the TAM in the central region of breast cancer increased when filling defect appeared(t=2.631,P<0.05),but the TAM in the peripheral region was more(t=2.999,P<0.05).After angiography,lesions showed high perfusion,and there was significantly more TAM in the edge and central area of the lesion than that in normal area next to the cancer(t=5.529,P<0.05;t=2.584,P<0.05).Clinical stage was related to the TAM in three regions.When the clinical stage was high,there were more TAM in all three regions(t=6.658,2.367,2.400,all P<0.05).Histological grading was high,and TAM in all three areas was high(F=101.151,16.922,26.822,all P<0.05).Conclusion There was a decreasing trend of TAM in the marginal area,central area and adjacent tissues of breast cancer during CEUS.The edge region has more malignant CEUS characteristics than that in the central region and the normal region adjacent to cancer;and the number of TAM is more in breast cancer with late clinical grading,poor tissue differentiation and obvious contrast-enhanced ultrasound malignant characteristics.The distribution characteristics of TAM represent the malignant degree and metastatic probability of breast cancer to a certain extent,and TAM is a factor related to the invasion of breast cancer.

Objective:To investigate the effect of changes in metabolic syndrome status and persistence on carotid plaque risk.Methods:This retrospective cohort study analyzed individuals who underwent routine health check-ups at the health management center of the Second Affiliated Hospital of Dalian Medical University from 2014 to 2023. Participants with at least three carotid ultrasound records meeting the inclusion criteria were classified into 4 groups based on changes in metabolic status: persistently metabolic health, transitioning from metabolic health to unhealth, transitioning from metabolic unhealth to health, and persistently metabolic unhealth. The cumulative incidence of carotid plaque in these groups was compared. A Cox proportional risk model was used to evaluate the relationship between changes in metabolic syndrome status, the number of metabolic syndrome components, and the risk of carotid plaque development. Restricted cubic spline analysis was applied to explore the association between changes in individual metabolic syndrome components and carotid plaque risk.Results:Compared to the persistently metabolic health group, the persistent unhealth group had the highest risk of developing carotid plaque( HR=1.35, 95% CI 1.05-1.74, P=0.021), followed by those who transitioned from metabolic health to unhealth and those who improved from metabolic unhealth to health. Furthermore, the risk of carotid plaque increased progressively with the number of metabolic syndrome components. Restricted cubic spline analysis revealed a nonlinear relationship between fasting blood glucose change and carotid plaque risk, while systolic blood pressure, diastolic blood pressure, waist circumference, triglycerides, and high-density lipoprotein-cholesterol showed a linear dose-response relationship with carotid plaque. Conclusions:The change of metabolic syndrome is associated with the risk of developing carotid plaque, and maintaining metabolic health, recovering from metabolic syndrome, or minimizing the number of metabolic syndrome components may be effective strategies to prevent carotid plaque formation.

Objective:To explore the relationship between thyroid hormone sensitivity and metabolic dysfunction-associated steatotic liver disease(MASLD) in a population with normal thyroid function, with a particular focus on sex-specific differences.Methods:This retrospective study included 41 355 euthyroid cases who underwent routine health examinations at the Health Management Centre of the Second Affiliated Hospital of Dalian Medical University from January 2014 to December 2023 were included. The free triiodothyronine(FT 3) to free thyroxine(FT 4) ratio(FT 3/FT 4) was calculated in order to reflect the peripheral sensitivity of the thyroid gland. Similarly, thyroid feedback quantile-based index(TFQI), thyrotrophic thyroxine resistance index(TT 4RI), and the FT 3-based TFQI-derived index(TFQI-FT 3) were calculated in order to reflect the central sensitivity of the thyroid gland. A Logistic regression was employed to analyse the effect of sex-specific thyroid hormone sensitivity indices on the prevalence of MASLD. The restricted cubic spline was used to analyse the non-linear relationship between the thyroid sensitivity hormone indices and MASLD. Furthermore, the correlation between the thyroid hormone sensitivity indices and MASLD in different subgroups was also analysed. Results:The prevalence of MASLD in the study population was 28.8%. After adjusting the model for confounders, the risk of MASLD increased by 7%, 3%, 10%, and 5% for each standard deviation increase in FT 3/FT 4, TFQI, TFQI-FT 3, and TT 4RI in the total population, respectively. The risk of MASLD increased by 6% and 5% for each standard deviation increase in FT 3/FT 4 and TFQI-FT 3 in men, respectively. For each standard deviation increase in FT 3/FT 4, TFQI, TFQI-FT 3, and TT 4RI in women, the risk of MASLD increased by 6%, 5%, 11%, and 5%, respectively. Higher FT 3/FT 4 and TFQI-FT 3 were positively associated with the risk of developing MASLD in men, and higher FT 3/FT 4, TFQI, TFQI-FT 3, and TT 4RI were positively associated with the risk of developing MASLD in women. There was a non-linear, inverted U-shaped relationship between TFQI and risk of MASLD in women. Subgroup analyses showed positive associations between FT 3/FT 4, TFQI, TFQI-FT 3, and MASLD. Conclusions:The thyroid hormone sensitivity indices may provide a basis for clinical prevention and management of MASLD in individuals with normal thyroid function. Additionally, FT 3/FT 4 and TFQI-FT 3 may indicate the risk of MASLD in the general population, while TFQI and TT 4RI are more suitable for assessing the risk of MASLD in women.

Objective:To investigate the diagnostic value of different conventional ultrasound-based radiomics models and their combination with clinical ultrasound features in the pathological subtyping of renal cell carcinoma.Methods:Retrospective data from 286 patients diagnosed with renal cell carcinoma by pathology at the Tenth Affiliated Hospital of Southern Medical University between May 1，2017 and June 7，2024 were collected. Among the 286 patients，203 were clear cell carcinoma，44 were papillary renal cell carcinoma，and 39 were chromophobe renal cell carcinoma. The patients were randomly divided into a training group（201 cases）and a validation group（85 cases）in a ratio of 7 to 3. Regions of interest（ROI）were delineated on conventional ultrasound images，and the radiomics features were extracted. Feature selection was performed using Student's t-test，Pearson correlation，and the least absolute shrinkage and selection operator（LASSO）. Six different machine learning methods included category gradient boosting（CatBoost），light gradient boosting machine（LightGBM），Logistic regression（LR），random forest（RF），support vector machine（SVM）and extreme gradient boosting（XGBoost）were used to establish radiomics models. Weight balancing was applied to correct for sample imbalance，and an imaging genomics model was constructed after balancing the samples. Independent predictors of renal cell carcinoma subtyping were selected from clinical ultrasound features using univariate and multivariate logistic regression analyses，and a clinical imaging model was constructed. The best-performing radiomics model was combined with the clinical independent predictors to construct a combined model. Receiver operating characteristic curves and the obuchowski index were plotted to evaluate model performance. Results:Among the radiomics models，the model constructed using Random Forest（RS RF）after balancing the samples exhibited the best predictive performance，with area under the curve（AUCs）of 0.918（micro-average ROC）and 0.903（macro-average ROC），and the obuchowski index was 0.885 in the validation group. The long and short axes of ultrasound image tumor masses were used as imaging independent predictors to construct a clinical imaging model. In the validation group，the AUCs of the clinical model were 0.886（micro-average ROC）and 0.606（macro-average ROC），and the obuchowski index was 0.569. The combined model achieved AUCs of 0.888（micro-average ROC）and 0.967（macro-average ROC），with an obuchowski index of 0.933，outperforming any single model. Conclusions:The combination of conventional ultrasound-based radiomics models with clinical ultrasound features demonstrates high diagnostic value in differentiating clear cell carcinoma，papillary renal cell carcinoma，and chromophobe renal cell carcinoma. It may serve as an auxiliary tool for providing timely and effective clinical guidance.

Purpose To explore the value of conventional ultrasound combined with ultrasound radiomics in detecting triple-negative breast cancer(TNBC).Materials and Methods In this study,we retrospectively collected the preoperative conventional ultrasound images and clinic-pathological data of 682 patients with lumpy breast cancer from the Affiliated Cancer Hospital of Xinjiang Medical University from January 2015 to December 2021.All patients included 325 cases with TNBC and 357 cases with non-TNBC.Features were extracted from preoperative ultrasound images of patients,screened,dimensionally reduced,and a combined model was constructed with ultrasound features and clinicopathological features.The performance of the three single models and combined models was evaluated by receiver operating characteristic curve analysis.Results The area under the curve of conventional ultrasound model,clinicopathological model,ultrasound imaging omics and the combined model in training group and validation suite were 0.730,0.718,0.982,0.985,0.726,0.683,0.981 and 0.982,respectively.The area under the curve of TNBC detected by the combined model was significantly higher than that of the single model(Z=7.311,7.024,3.883,all P<0.01).Conclusion The combined model based on ultrasonography has good diagnostic efficacy in detecting TNBC.As a non-invasive examination,it can avoid unnecessary biopsy to the patient.