Objective:To develop and validate machine learning-based radiomics models using preoperative CT images for individualized prediction of mitotic index (MI) in patients with gastrointestinal stromal tumors (GIST).Methods:The study was a case-control study. The data of 348 GIST patients confirmed by pathology were retrospectively collected from two independent medical centers: the Affiliated Hospital of Qingdao University (center 1) and Shandong Provincial Hospital Affiliated to Shandong First Medical University (center 2), covering the period from January 2013 to June 2018. Patients from center 1 were divided into a training cohort (176 cases) and an internal validation cohort (75 cases) at a ratio of 7∶3 using random sampling. Patients from center 2 served as an independent external validation cohort (97 cases). The primary endpoint was MI, categorized into high MI (145 cases) and low MI (203 cases) groups. Radiomic features were extracted from the portal venous phase images of preoperative contrast-enhanced CT scans. Five machine learning algorithms, including logistic regression, support vector machine, random forest, decision tree, and extreme gradient boosting (XGBoost),were employed to construct MI prediction models. The optimal model was identified using receiver operating characteristic curves. An individualized prediction model was developed by integrating the the optimal machine learning model combined with selected independent clinical factors, and the importance of features was visualized using Shapley Additive Explanation (SHAP) analysis. Patients were followed up, and Kaplan-Meier curves along with log-rank tests were used to evaluate recurrence-free survival (RFS) differences between the predicted high MI and low MI groups.Results:Among the five constructed machine learning models, the XGBoost model demonstrated the best predictive performance, with area under the curve (AUC) of 0.809 (95% CI 0.738-0.872), 0.693 (95% CI 0.571-0.809), and 0.718 (95% CI 0.605-0.822) in the training cohort, internal validation cohort, and external validation cohort, respectively. An individualized prediction model combining the XGBoost model with independent clinical factors (tumor location and tumor size) was developed. The model achieved AUC of 0.843 (95% CI 0.785-0.899), 0.791 (95% CI 0.680-0.894), and 0.777 (95% CI 0.678-0.861) in the training cohort, internal validation cohort, and external validation cohort, respectively. SHAP analysis indicated that radiomic features had the highest predictive impact. In both the training cohort and internal validation cohort, the RFS of patients predicted to be in the high MI group was lower than that of the low MI group, with statistically significant differences ( χ2=14.58, 9.52, both P<0.001). However, there was no statistically significant difference in RFS in the external validation set ( χ2=6.18, P=0.080). Conclusions:The optimal XGBoost model based on radiomic features extracted from preoperative portal venous phase CT images, when combined with clinical factors, can effectively predict the MI of GIST patients.

Objective:To explore the value of the vesical imaging-reporting and data system (VI-RADS) score based on multiparametric MRI (mpMRI) combined with quantitative tumor MRI parameters in assessing the muscle invasion of bladder cancer.Methods:The study was a case-control study. The data of 87 bladder cancer patients confirmed by pathology who underwent mpMRI of the bladder were retrospectively collected from the First Medical Center of Chinese PLA General Hospital between January 2019 and April 2023 The pathological findings were used as the gold standard to categorize them into the muscle invasive bladder cancer (MIBC) group (29 cases) and non-muscle invasive bladder cancer (NMIBC) group (58 cases). Quantitative parameters were measured based on preoperative mpMRI images, including the length of tumor bladder wall contact, the perpendicular distance between the bladder tumor and the tangent of the bladder wall, the maximal diameter of the bladder tumor, and the volume of the bladder tumor. Bladder cancer was classified according to the VI-RADS scoring criteria. The Mann-Whitney U test was used for intergroup comparisons. Multivariate logistic regression analysis was performed to obtain the independent risk factors related to muscle invasion of bladder cancer and to establish the model. The receiver operating characteristic curves were analyzed for MRI quantitative parameters and logistic regression models, and area under the curve (AUC) comparisons were performed using the DeLong test. Results:The differences in tumor bladder wall contact length, perpendicular distance from the tumor to the tangent line of the bladder wall, maximum diameter, bladder tumor volume, and the VI-RADS scores were statistically significant between the MIBC group and the NMIBC group ( P<0.05). Multifactorial logistic regression analysis showed that tumor bladder wall contact length ( OR=21.07, 95% CI 3.56-124.89, P=0.001) and VI-RADS score ( OR=11.90, 95% CI 3.53-40.12, P<0.001) were the independent risk factors for evaluating the muscle invasion of bladder cancer. The difference between the VI-RADS score and the tumor bladder wall contact length for assessing muscular infiltration of bladder cancer had AUCs of 0.802 (95% CI 0.704-0.899) and 0.759 (95% CI 0.652-0.865). The combined model of VI-RADS score combined with tumor bladder wall contact length had an AUC of 0.891 (95% CI 0.812-0.970), which was higher than the diagnostic efficacy of applying tumor bladder wall contact length or VI-RADS score alone ( Z=3.05, 2.37, P=0.002, 0.018). Conclusion:Tumor contact length with the bladder wall is an independent risk factor for assessing muscle invasion of bladder cancer and the combination of VI-RADS score may enhances diagnostic accuracy.

Objective:To evaluate the short-and medium-term therapeutic efficacy of shockwave balloon therapy for TASCⅡ C/D femoropopliteal artery atherosclerosis obliteration.Methods:This retrospective cohort study included 25 patients who received shockwave balloon therapy in five vascular centers from August 2022 to June 2023. All patients were diagnosed with TASC Ⅱ C/D femoropopliteal arteriosclerosis obliterans (13 cases of type C and 12 cases of type D), and underwent intravascular shock wave lithotripsy (IVL) to treat calcified lesions. The immediate effectiveness (residual stenosis<30% and no flow-limiting dissection), safety (whether there were adverse vascular events during the operation) and the rate of salvage stent implantation were recorded. The observation indexes of patients before operation, early postoperative period (immediately after operation or before discharge) and postoperative follow-up period (3, 6, 12 months after operation) were collected. The observation indexes included ankle-brachial index (ABI), Rutherford classification, and minimum lumen diameter (MLD). Repeated measures ANOVA was used to evaluate the changes of observation indexes in the early postoperative and follow-up stages compared with those before operation; Kaplan-Meier survival analysis was used to evaluate the one-stage patency rate at follow-up and the target lesion revascularization rate free from clinical drive.Results:The immediate effectiveness of surgery was 100% in all patients, with no vascular related adverse events occurred, and no remedial stent implantation was performed. The ABI, Rutherford grade and MLD of the patients in the early postoperative period and each follow-up stage were improved compared with those before operation, with statistically significant differences ( P<0.05). Kaplan-Meier survival analysis showed that the primary patency rate at 12 months after surgery was 0.78 (95% CI 0.64-0.84), and the revascularization rate of target lesions free from clinical drive was 0.87 (95% CI 0.85-0.95). Conclusion:Shockwave balloon therapy for complex calcified femoropopliteal artery lesions is safe and reliable, with satisfactory short-and medium-term efficacy.

Objective:To compare the clinical efficacy of catheter-directed thrombolysis (CDT) via the contralateral femoral vein approach (CFVA-CDT) and the calf deep vein approach (CVA-CDT) in the treatment of acute mixed-type lower extremity deep vein thrombosis (DVT).Methods:Patients treated with CFVA-CDT and CVA-CDT for acute mixed-type DVT were retrospectively collected from January 2018 to December 2021, totaling 49 and 32 patients, respectively. The relevant technical indicators, thrombolysis rates in the iliac-femoral vein segment and femoral-popliteal vein segment, clinical efficacy, and the incidence of lower extremity deep vein patency, venous valve insufficiency, and post-thrombotic syndrome (PTS), as well as the severity of chronic venous disease in the affected limb (VCSS score) during a 2-year follow-up period were retrospectively compared between the two venous access CDT groups. The t-test was used for comparing quantitative data, while the chi-square test or Fisher′s exact test was used for categorical data.Results:During CFVA-CDT procedure, 6-8 F vascular sheaths were used, and balloon dilation of 2~6 mm was more frequently employed (65.31%, 32/49) to expand venous stenosis/occlusion segments before successful sheath placement compared to the CVA-CDT group (37.50%, 12/32), and the difference was statistically significant ( P=0.014). In the CVA-CDT group, 31.25% (10/32) of patients had a maximum sheath size of 6 F, while the remainder used 4 or 5 F sheaths. Among them, 34.38% (11/32) of patients required re-puncture of the popliteal or femoral vein for larger sheaths (≥8 F) for thrombus aspiration and subsequent endovascular treatment during or after thrombolysis. The effective thrombolysis rates (≥50%) in the iliac-femoral vein segment were not significantly different between the two groups ( P=0.778). The effective thrombolysis rate of the femoral-popliteal venous segment is related to the presence or absence of popliteal vein opacification on lower extremity venous antegrade venography. There was no significant difference between the groups when the popliteal vein was visualized ( P=1.000). While the popliteal vein was not visualized, the CVA-CDT group (75.0%, 15/20) was significantly better than the CFVA-CDT group (34.38%, 11/32), and the difference was statistically significant ( P=0.004). There was no significant difference in clinical efficacy between the two groups ( P=0.819). During follow-up, the femoral-popliteal vein patency rate in the CVA-CDT group (87.50%, 28/32) was significantly higher than in the CFVA-CDT group (44.90%, 22/49), the difference was statistically significant ( P<0.001). Conclusions:Successful CFVA-CDT requires the assistance of more ancillary devices, while the use of larger sheaths is more limited in CVA-CDT due to the smaller caliber of the calf deep veins. The presence or absence of popliteal vein opacification on lower extremity venous antegrade venography may influence the effective thrombolysis of the femoral-popliteal venous segment thrombus in patients with acute mixed deep vein thrombosis (DVT) treated with CFVA-CDT and CVA-CDT. Compared to CFVA-CDT, CVA-CDT can improve the patency rate of the femoral-popliteal venous segment.

Objective:To explore age-related variations in the angle of the cochlear basal turn using temporal bone CT, providing a reference for selecting the optimal Stenvers position radiographic projection angle in children and adults.Methods:The retrospective study included children and adults who underwent temporal bone CT scans at Beijing Tongren Hospital from November 2014 to April 2023. A total of 620 participants were included, including 368 males and 252 females. Patients were divided into 20 age-ralated groups: infants under one year old (3 to 11 months) were divided into monthly subgroups (9 groups); children and adolescents aged 1 to 18 years were grouped biennially (9 groups); adults were divided into two groups: 19 to 29 years and 30 to 40 years. Using multiplanar reconstruction (MPR) techniques, the CT images of the temporal bone were reformatted into oblique transverse sections to maximize the visibility of the cochlear basal turn.The cochlear basal turn angle was defined as the angle between the vertical axis of the cochlear basal turn and the mid-sagittal plane of the skull. Statistical analysis was performed to compare age-related differences in cochlear basal turn angles. Two additional patients were included to compare Stenvers position X-ray images with corresponding temporal bone CT scans, assessing the visibility of cochlear implant electrodes post-implantation.Results:Among infants aged 3 to 11 months, the cochlear basal turn angle was 29.4°±4.5°, with no significant differences observed between subgroups ( P>0.05). However, significant differences were found between infants (<1 year old) and the 1-2-year-old group compared to each age group from 3 to 40 years ( P<0.05). Additionally, the angles differed significantly between the 3-14-year-old groups and the 19-40-year-old groups ( P<0.05), whereas no significant differences were found among the remaining groups ( P>0.05). The visibility of the cochlear implant electrodes, appearing round in shape on standard Stenvers position X-ray images, closely resembled that observed in temporal bone CT scans. Conclusion:Age-related variations in the cochlear basal turn angle provide a valuable reference for optimizing Stenvers position radiography angles after cochlear implantation, improving the accuracy and quality of post-implantation imaging.

Objective:To evaluate the efficacy of the “double-low” scanning protocol (low tube voltage and low-concentration contrast agent) combined with the artificial intelligence iterative reconstruction (AIIR) algorithm for abdominal CT enhancement in patients with obesity and to identify the optimal AIIR reconstruction algorithm level.Methods:From April 2024 to July 2024, patients with a body mass index≥30.00 kg/m2 who underwent abdominal CT enhancement at the Affiliated Hospital of Hebei University were prospectively included. All patients were randomly assigned to groups A or B. Patients in Group A accepted the conventional scanning protocol (automatic tube voltage selection and a contrast agent concentration of 350 mg/ml) with reconstruction using the Karl 3D iterative reconstruction algorithm at levels 3-5. The “double-low” protocol (a fixed tube voltage of 80 kVp and a contrast agent concentration of 320 mg/ml) with AIIR algorithm reconstruction at levels 1-5 were performed in Group B. CT values and image noises were measured, including the right posterior liver lobe at the level of the first porta hepatis and subcutaneous fat at the third lumbar level during arterial and portal venous phases, abdominal aorta at the third lumbar vertebra during the arterial phase, and portal vein trunk during the portal-venous-phase. Radiation dose, total iodine intake, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and subjective image quality scores were recorded. The optimal reconstruction levels for arterial-phase and portal-venous-phase images were identified for each group by repeatedly measured ANOVA. The figure of merit (FOM) was calculated for the best images in both groups, and comparisons were made in terms of image quality, radiation dose, and iodine intake using an unpaired t-test or Wilcoxon test. Results:Overall, 150 patients with obesity were collected, and each group included 75 cases. In group A, compared with levels 3 and 4 ( P<0.001), the Karl 3D level 5 algorithm yielded significantly higher SNR, CNR values, and subjective scores, designating level 5 as the optimal reconstruction level. In group B, the AIIR level 4 algorithm achieved higher SNR and CNR values than level 5 and achieved higher subjective scores than levels 3 and 5 ( P<0.001), which means that level 4 was the optimal reconstruction level. Images reconstructed with AIIR level 4 in group B exhibited significantly higher CT, SNR, CNR, FOM values, and subjective scores than those reconstructed with Karl 3D level 5 in group A ( P<0.001). Compared with group A, the volume CT dose index values, dose-length product, and size-specific dose estimate based on water equivalent diameter in Group B were reduced by 56.75%, 58.29%, and 56, 71% during the arterial phase, and 56.70%, 58.27%, and 56.88% during the portal venous phase, respectively. Total iodine intake was significantly reduced by 10.71% in group B ( P<0.001). Conclusions:The “double-low” scanning protocol combined with AIIR algorithm significantly reduced radiation dose and iodine intake during abdominal CT enhancement in patients with obesity, without compromising image details, increasing noise, or altering image quality. AIIR level 4 was the optimal image reconstruction level for arterial-phase and portal-venous-phase in obese patients.

Objective:To investigate the value of breast MRI imaging features combined with serological indicators in predicting the metastatic burden of axillary lymph nodes (ALN) in breast cancer.Methods:This cross-sectional study retrospectively enrolled 146 female patients diagnosed with breast cancer at the First Affiliated Hospital of Shihezi University from January 2020 to November 2024. Patients′ pre-treatment clinical data, serological indices, breast MRI image features, and post-surgical pathologic features were analyzed. Patients were divided into low-burden (<3 metastatic lymph nodes) group and high-burden (≥3 metastatic lymph nodes) group based on pathological ALN confirmation. Group comparisons of clinical variables were analyzed using independent samples t-tests, Mann-Whitney U tests, or χ2 tests. Indicators with statistically significant differences were included in a multivariable logistic regression analysis to screen for independent influences predicting high ALN load and construct multiple logistic regression models. The performance of these models was evaluated using receiver operating characteristic curves and area under the curve (AUC), while net clinical benefit was assessed using decision curve analysis (DCA). Results:Significant differences were observed between low and high ALN burden groups in carcinoembryonic antigen levels, CA153 levels, tumor diameter, margins, enhancement characteristics, number of peritumoral thick blood vessels (TBVs), MRI-reported ALN loading status (MRI-ALN), and lymphovascular invasion status ( P<0.05). Multivariable logistic regression analysis showed that serum CA153 level ( OR=1.056, 95% CI 1.007-1.108, P=0.024), tumor margins ( OR=3.977, 95% CI 1.561-10.131, P=0.004), TBVs ( OR=3.058, 95% CI 1.217-7.684, P=0.017), and MRI-ALN ( OR=9.424, 95% CI 3.531-25.155, P<0.001) were independent risk factors predicting high ALN load in breast cancer ( P<0.05). The logistic regression model incorporating these four risk factors yielded optimal predictive performance for high ALN burden in breast cancer (AUC=0.854). DCA demonstrated optimal net clinical benefit within the threshold probability range of 13.3% to 72.7%. Conclusions:Tumor margins, TBVs, MRI-ALN, and CA153 levels are significantly associated with high ALN metastatic burden in breast cancer. Constructing a predictive model incorporating these features can significantly improve the accuracy of identifying high ALN burden.

Objective:To investigate the value of MRI radiomics model in evaluating the pathological complete response (pCR) status of human epidermal growth factor receptor 2(HER-2) positive breast cancer after neoadjuvant therapy.Methods:The study was a cross-sectional study. The clinical, pathological, and MRI data of 243 HER-2 positive breast cancer patients who received neoadjuvant therapy in Shanxi Province Cancer Hospital from January 2021 to June 2023 were retrospectively analyzed. All patients were female, aged 26?75 years. All patients were randomly divided into training set (146 cases) and validation set (97 cases) at a ratio of 6∶4 according to the simple random sampling method. Univariate and multivariate logistic regression were used to screen independent predictors of pCR. Radiomics features were extracted from the early-phase (the 2nd phase) images of breast dynamic contrast-enhanced-MRI after neoadjuvant therapy.The four-step procedure was adopted for feature screening. The radiomics model was constructed by logistic regression. A combined model was constructed by integrating radiomics features and independent predictors. Two radiologists (Reader 1 with 10 years experience and Reader 2 with 13 years experience) who major in breast MRI visually evaluated the pCR status of breast cancer after neoadjuvant therapy. The receiver operating characteristic curve and area under the curve (AUC) were used to evaluate the efficacy of Reader 1, Reader 2, the radiomics model, and the combined model in predicting pCR status. The Hosmer-Lemeshow goodness-of-fit test was used to evaluate the calibration of the model.Results:Among 243 HER-2 positive breast cancer patients, totally 118 achieved pCR. In clinical and pathological features, HER-2 3+ was an independent predictor of pCR ( OR=2.71, 95% CI 1.03?7.12, P=0.043). In the training set and validation set, the AUCs of the radiomics model in predicting pCR status were 0.899 and 0.853, respectively.The AUCs of the combined model were 0.917 and 0.890, respectively. In the validation set, the AUC value of the radiomics model in predicting pCR status was higher than that of Reader 1 and Reader 2. Hosmer-Lemeshow goodness-of-fit test showed that there was no significant difference between the prediction of pCR status by the combined model and radiomics model and the actual results in the training set and validation set, and the fitting was good ( P>0.05). Conclusion:The MRI-based radiomics model can be used to predict pCR status in HER-2 positive breast cancer and outperforms the visual qualitative assessments of radiologists.

Objective:To develop an artificial intelligence (AI)-based automatic scoring model for magnetic resonance imaging-detected extramural vascular invasion (AI-mrEMVI) and evaluate its performance and prognostic value in patients with rectal cancer.Methods:In this multicenter retrospective cohort study, a total of 2 501 rectal cancer patients from seven centers between November 2012 and December 2020 were included and divided into completely independent training ( n=1 830) and validation ( n=671) cohorts. A nnUNet-based AI-mrEMVI scoring model was constructed. Manual mrEMVI scores assigned by two radiologists served as the reference standard for accessing the accuracy of the AI-mrEMVI scoring. Kaplan-Meier survival analysis and Cox regression were used to evaluate the prognostic stratification ability of the AI-mrEMVI scores. The concordance index (C-index) was calculated to evaluate prognostic performance. Results:In the validation cohort, the manual mrEMVI scores were 0-2 in 425 patients (63.3%), 3 in 89 (13.4%), and 4 in 157 (23.4%). The AI-mrEMVI model identified 0-2 in 375 patients (55.9%), 3 in 95 (14.2%), and 4 in 201 (30.0%), with an overall accuracy of 81.1% (544/671, 95% CI 77.9%-84.0%). The 3-year disease-free survival (DFS) rates for patients with AI-mrEMVI scores of 0-2, 3, and 4 were 85.2%, 70.0%, and 58.2%, respectively, and the 5-year overall survival (OS) rates were 87.2%, 81.6%, and 62.6%, respectively (DFS: χ2=48.74, P<0.001; OS: χ2=30.04, P<0.001). Multivariable Cox regression showed that for DFS, AI-mrEMVI scores of 3 and 4 were associated with hazard ratios ( HR) of 1.75 (95% CI 1.11-2.77, P=0.016) and 2.65 (95% CI 1.86-3.78, P<0.001), respectively. For OS, an AI-mrEMVI score of 4 was associated with an HR of 2.56 (95% CI 1.62-4.03, P<0.001). The C-index values of the AI-mrEMVI scoring model for predicting DFS and OS were 0.647 (95% CI 0.608-0.686) and 0.650 (95% CI 0.598-0.702), respectively. Conclusion:The proposed AI-mrEMVI automatic scoring model demonstrated high diagnostic accuracy and performed favorably in predicting DFS and OS prognostic risk in patients with rectal cancer.

Objective:To investigate the value of habitat radiomic features based on gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI in establishing a predictive model for cytokeratin 19 (CK19) expression in hepatocellular carcinoma (HCC) and to evaluate its role in prognostic risk stratification.Methods:This multicenter case-control study retrospectively enrolled 489 patients with pathologically confirmed HCC who underwent Gd-EOB-DTPA-enhanced MRI between June 2016 and June 2024. Among them, 346 patients from the First Affiliated Hospital of Soochow University were divided into a training cohort ( n=245) and an internal test cohort ( n=101) via stratified sampling at a 7∶3 ratio. And 143 patients from Nantong Third Hospital Affiliated to Nantong University served as an external validation cohort. The training cohort included 53 CK19-positive and 192 CK19-negative patients. The internal test cohort included 21 CK19-positive and 80 CK19-negative patients. The external validation cohort included 30 CK19-positive and 113 CK19-negative patients. Univariate logistic regression analysis was performed to identify potential factors associated with CK19 expression, and a clinical-radiologic model was constructed. The k-means clustering algorithm was applied to segment target HCC lesions into 3 subregions. Radiomic features were extracted and selected from these habitat subregions. Habitat radiomics models were constructed for the arterial phase (AP), portal venous phase, hepatobiliary phase (HBP), and combined phases (CP). Multivariate logistic regression analysis identified independent clinical and radiologic predictors of CK19 expression, and the optimal habitat model score was integrated to build a clinical-radiologic-habitat combined model. The area under the receiver operating characteristic curve (AUC) was used to evaluate model predictive performance. Recurrence-free survival (RFS) was analyzed using the Kaplan-Meier method and the differences in survival curves were compared with the log-rank test. Results:Univariate logistic regression analysis revealed that alpha-fetoprotein (AFP) ( OR=2.629, 95% CI 1.412-4.896, P=0.002), AP enhancement ( OR=3.636, 95% CI 1.642-8.052, P=0.001), AP peritumoral enhancement ( OR=2.219, 95% CI 1.084-4.542, P=0.029), and HBP peritumoral hypointensity ( OR=2.010, 95% CI 1.004-4.021, P=0.049) were potential factors associated with CK19 expression, which were incorporated into the clinical-radiologic model. In the internal and external validation cohorts, the AUC of the clinical-radiologic model was 0.690 (95% CI 0.590-0.778) and 0.650 (95% CI 0.565-0.727), respectively. The habitat radiomics model based on CP images demonstrated the highest performance. It achieved AUC of 0.729 (95% CI 0.622-0.836) and 0.725 (95% CI 0.607-0.842) in the internal and external validation cohorts, respectively. Multivariate analysis identified AFP ( OR=2.494, 95% CI 1.163-5.348, P=0.019), AP enhancement ( OR=5.230, 95% CI 1.868-14.643, P=0.002) and habitat radiomics model score ( OR=4.105, 95% CI 2.643-6.368, P<0.001) as independent predictors of CK19 positivity. Based on these factors, a combined clinical-radiologic-habitat combined model was established. The clinical-radiologic-habitat combined model achieved AUCs of 0.767 (95% CI 0.671-0.846) and 0.730 (95% CI 0.649-0.801) in the internal and external validation cohorts, respectively. Significant differences in RFS were observed between the CK19-positive group (25.1 month) and CK19-negative group (51.0 month) as predicted by the clinical-radiologic-habitat model ( χ2=4.17, P=0.041). Conclusion:The clinical-radiologic-habitat combined model based on Gd-EOB-DTPA-enhanced MRI habitat radiomics demonstrates good predictive performance for CK19 expression in HCC and offers valuable prognostic stratification for clinical practice.