Objective To explore the value of preoperative prediction of BRAF V600E mutation in ganglioglioma based on the full-volume histogram of apparent diffusion coefficient（ADC）. Methods The clinical and imaging data of 63 patients pathologically diagnosed with ganglioglioma（GG）in our hospital from January 2017 to December 2025 were consecutively and retrospectively collected.The patients were divided into the BRAF V600E wild-type group（n=33）and BRAF V600E mutant group（n=30）according to the gene mutation status.The FireVoxel software was used to segment the region of interest（ROI）along the lesion edge in the tumor on the ADC image layer by layer,and then 16 histogram features were extracted.Differences between the groups were compared using the independent samples t-test or Mann-Whitney U test.A P value≤0.05 indicated a statistically significant difference.The receiver operating characteristic curve（ROC）analysis was used to evaluate the diagnostic performance of histogram parameters for distinguishing BRAF V600E wild-type from BRAF V600E mutant GG. Results The ADCmean, ADC1th, ADC10th, ADC25th, ADC50th, ADC75th, ADC90th, and ADC99th in the BRAF V600E mutant group were significantly lower than those in the BRAF V600E wild-type group（all P values≤0.05）,while the skewness in the BRAF V600E mutant group was significantly higher than that in the BRAF V600E wild-type group（P values≤0.05）.The ROC analysis showed that ADCmean had the best diagnostic performance for distinguishing BRAF V600E wild-type and BRAF V600E mutant GG,with the area under the ROC curve（AUC）,sensitivity,specificity,cutoff value and Youden index being 0.848（0.736-0.926）,100.00%,63.64,≤1783.237 and 0.636,respectively. Conclusion ADC full-volume histogram parameters may serve as a non-invasive tool for preoperative differentiation of BRAF V600E wild-type and BRAF V600E mutant GG.
Objective To investigate the association between dysregulation of glymphatic system and cognitive impairment in presbycusis patients using structural magnetic resonance imaging(MRI). Methods The hearing ability,neuropsychological scales,and structural MRI data of 107 presbycusis patients and 99 healthy controls were collected from the Department of Otolaryngology-Head and Neck surgery,Nanjing First Hospital from January 2024 to January 2026.Diffusion tensor imaging along the perivascular spaces(DTI-ALPS)was analyzed using diffusion tensor imaging,and volumes of gray matter,white matter,choroid plexus,perivascular spaces,lateral ventricles,third ventricle,and fourth ventricle were extracted from 3D-T1WI to evaluate the properties of glymphatic system.Intergroup differences of the glymphatic system were compared between the two groups,and their correlations with clinical indicators were calculated. Results Pure tone audiometry tests showed that the average hearing thresholds of both left and right ears in presbycusis patients were significantly higher than those of healthy controls(P<0.05).Neuropsychological scale scores revealed abnormal performance in the digit span test,digit symbol substitution test,trail making test,and auditory verbal learning test in patients with presbycusis,indicating cognitive impairments.Compared to healthy controls,analysis of glymphatic system properties showed significantly increased volumes of choroid plexus,lateral ventricles,and third ventricle in presbycusis patients(P<0.05).Additionally,enlarged volume of choroid plexus was negatively associated with the digit symbol substitution test scores(r=-0.252,P=0.012). Conclusion sDysregulation of the glymphatic system in patients with presbycusis may be the potential mechanism underlying the development of cognitive impairments.
Objective This study investigates differences in spiculation features observed on cone-beam breast computed tomography（CBBCT）between invasive ductal carcinoma（IDC）and invasive lobular carcinoma（ILC）,along with their associated clinicopathological characteristics,and evaluates the diagnostic value of these spiculation differences in distinguishing IDC from ILC. Methods A retrospective review was conducted on the clinical,pathological,and CBBCT imaging data of 63 patients with pathologically confirmed IDC and 63 patients with pathologically confirmed ILC,all treated at our institution between July 2019 and May 2024.Group comparisons were performed using independent-samples t-tests,chi-square tests,and Fisher’s exact tests to assess differences in age,family history of breast cancer,lymph node metastasis,calcification status of the primary tumor,size of the primary tumor mass,molecular subtype,estrogen receptor（ER）status,progesterone receptor（PR）status,Ki-67 index,human epidermal growth factor receptor 2（HER-2）status,morphological characteristics of spiculations in the primary tumor,and degree of tumor-associated desmoplasia.Cohen’s Kappa statistic was used to assess inter-reader agreement between two radiologists regarding the subjective evaluation of spiculation features.Spearman rank correlation analysis was performed to evaluate associations between spiculation length,anterior segment width,tortuosity,and density and tumor histologic subtypes(IDC and ILC). Results Statistically significant differences were observed between the two groups in terms of calcification status,mass size,ER status,PR status,Ki-67 index,HER2 status,molecular subtype,morphological characteristics of spiculation,and degree of tumor-associated desmoplasia（all P<0.05）.High inter-reader agreement was achieved for spiculation morphological feature assessment in both IDC and ILC groups（kappa values ranging from 0.807 to 0.950,all P<0.05）.Spearman correlation analysis revealed significant associations between IDC and ILC and all measured spiculation morphological parameters. Conclusion sCBBCT can clearly visualize spiculations associated with breast masses.Spiculations in IDC typically manifest as short,rigid spiculations,whereas those in ILC more commonly appear as long,soft spiculations.This morphological distinction correlates with differences in the severity of tumor-induced desmoplastic reactions between the two subtypes.The short,rigid spiculation and long,soft spiculation patterns may aid in the differential diagnosis of IDC and ILC.
Objective To investigate the effect of iodine contrast medium on the image quality of contrast-enhanced mammography（CESM）. Methods Patients who underwent CESM examinations at the Radiology Department of the Fourth Hospital of Hebei Medical University from August 2019 to April 2023 were divided into three groups（29 individuals per group）depending on the concentration of the contrast agent injection：the iohexol group（320 mgI/ml）,iopamidol group（370 mgI/ml）,and iomeprol group（400 mgI/ml）.Prior to the CESM examination,a fixed injection rate of 3 ml/s was used,and the three different contrast agents were randomly injected for conventional imaging. Objective and subjective evaluations were performed on the images of the three groups to compare the influence of different iodine concentrations on the image quality of CESM.The Objective evaluation of the images was conducted by comparing the signal-to-noise ratio（SNR）,contrast-to-noise ratio（CNR）,lesion enhancement grayscale values,and image noise among the three groups to assess differences.Diagnostic physicians conducted double-blind subjective assessment to evaluate image quality.The diagnostic efficacy of three different contrast agents for breast diseases was compared using the ROC curve. Results There were statistically significant differences in the SNR and image noise among the three different iodine contrast agents(P=0.002,P=0.003).Iomeprol demonstrated a superior SNR and less image noise than did iohexol(P=0.001,P=0.01);iopamidol also had a greater SNR and less image noise than did iohexol(P=0.008,P=0.0001).However,there were no statistically significant differences in the CNR or lesion grayscale values(P=0.292,P=0.452).Subjective ratings showed statistically significant differences among the three groups(P=0.009), with iomeprol showing superior image quality compared to iohexol(P=0.007),with no statistically significant differences between the other two groups.The diagnostic efficacy of the high-concentration contrast agent was slightly higher than that of the other two lower-concentration contrast agents.The area under the curve(AUC)was 0.876、0.857、0.842.After pairwise comparisons,there was no statistically significant difference(P=0.909,P=0.783,P=0.860). Conclusion sHigh-concentration contrast agents have advantages in improving the SNR of CESM images,reducing image noise,and maintaining diagnostic efficacy.
Objective To develop a nomogram model incorporating dual-modality imaging features from mammography（MMG）and ultrasound（US）along with clinical factors,and to evaluate its predictive value for the malignancy risk of Breast Imaging Reporting and Data System（BI-RADS）category 4 lesions. Methods A retrospective analysis was conducted on 915 patients from two centers who were diagnosed as BI-RADS category 4 by MG.Patients from center 1 were randomly divided into a training set（n=481）and an internal test set（n=206）at a 7∶3 ratio,while patients from center 2 served as an external test set（n=228）.Feature selection was performed using the Boruta algorithm and LASSO regression.A multivariate Logistic regression model was constructed to predict malignancy risk and subsequently visualized as a nomogram.The model's discriminative ability was assessed using the area under the receiver operating characteristic curve（AUC）,its calibration was evaluated with calibration curves,and its clinical net benefit was analyzed via decision curve analysis（DCA）. Results The LASSO regression model ultimately identified 11 predictors.The Boruta algorithm confirmed 16 important features,including those mentioned above.The model built based on the Boruta algorithm achieved AUCs for predicting malignancy risk of 0.944 in the training set,0.926 in the internal test set,and 0.897 in the external test set.Both calibration curves and DCA indicated that the model possessed good calibration and clinical utility. Conclusion sThe nomogram model constructed based on dual-modality imaging features combined with clinical factors demonstrates good predictive performance for the malignancy risk of BI-RADS category 4 lesions.Incorporating the Boruta algorithm aids in achieving more comprehensive and robust feature selection,thereby enhancing the reliability of feature selection results and the clinical interpretability of the model.
Objective To evaluate the diagnostic performance of amide proton transfer-weighted (APTw) imaging in distinguishing triple-negative breast cancer (TNBC) from non-TNBC and to determine whether it provides added value to diffusion-weighted imaging (DWI) for TNBC identification. Methods This retrospective study included 106 patients with pathologically confirmed breast cancer,classified into TNBC (n=36) and non-TNBC (n=70) groups.All patients underwent multiparametric MRI,including dynamic contrast-enhanced (DCE),APTw,and DWI sequences.Amide proton transfer (APT) and apparent diffusion coefficient (ADC) values were measured from the corresponding maps.Differences in these parameters between the two groups were analyzed.Receiver operating characteristic (ROC) curve analysis was used to assess the diagnostic performance of APT,ADC,and their combination. Results The mean age of the cohort was （54.3±9.4）years.Significant differences were observed between TNBC and non-TNBC groups.TNBC lesions exhibited significantly higher APT values(2.43±0.37)% vs. (1.95±0.38)% （P<0.05） and significantly higher ADC values (1.23±0.20)×10⁻³ mm²/s vs. (0.97±0.13)×10⁻³mm²/s (P<0.05).The area under the ROC curve (AUC) for APT and ADC in differentiating TNBC from non-TNBC was 0.81 and 0.85,respectively.The combined model demonstrated superior diagnostic performance with an AUC of 0.91,which was significantly higher than that of either parameter alone (P<0.05). Conclusion APTw imaging is a valuable tool for differentiating TNBC from non-TNBC.Furthermore,the combination of APTw and DWI significantly improves diagnostic accuracy for TNBC identification compared to using either technique alone.
Objective To investigate the clinical value of a multi-parametric MRI-Transformer fusion model in differentiating HER2-0 from HER2-low breast cancer. Methods Multi-parametric MRI images of 544 patients from two centers were retrospectively collected.Both radiomics and 2.5D deep learning features were extracted.Following feature dimensionality reduction,machine learning algorithms were utilized to construct a radiomics (Rad) model and an early fusion (DLREarly) model,respectively.A Transformer framework was then introduced to construct a deep fusion (DLRT) model combining the two types of features.Subsequently,a comprehensive clinical-imaging model was developed by integrating independent clinical predictors.Model performance was evaluated using metrics including the area under the receiver operating characteristic curve (AUC). Results The comprehensive model achieved the best performance,yielding AUCs of 0.871,0.829,and 0.814 in the training,internal validation,and external validation sets,respectively,while demonstrating good calibration and clinical benefit.The DLRT model based on the Transformer architecture outperformed the Rad model and the DLREarly model,demonstrating robust performance in the external validation set. Conclusion The fusion model based on the multi-parametric MRI-Transformer architecture effectively integrates radiomics and deep learning features,enabling robust prediction and differentiation of HER2-0 and HER2-low breast cancer.
Objective To develop a Nomogram based on intratumoral,peritumoral,and combined intratumoral-peritumoral radiomics features from DCE-MRI combined with clinical and pathological variables for predicting tumor-infiltrating lymphocytes (TILs) levels in breast cancer. Methods This retrospective study included 257 patients with pathologically confirmed invasive breast cancer,comprising a training set of 205 cases and a validation set of 52 cases.Preoperative DCE-MRI phase 4 images were collected.The intratumoral region of interest (ROI) was manually delineated,and a peritumoral ROI was obtained by expanding 3 mm outward from the intratumoral ROI.Radiomics features were extracted and screened from the intratumoral,peritumoral,and combined intratumoral-peritumoral ROIs to construct three prediction models.The contribution of each feature was visualized using SHAP waterfall plots.After identifying the optimal model,a Nomogram was constructed by integrating it with clinical and pathological variables.Model performance was evaluated using the area under the receiver operating characteristic curve (AUC) and 95% confidence interval (CI).DeLong test was used to compare differences in AUCs.Calibration curves and decision curve analysis (DCA) were employed to assess the clinical utility of the models,with all results presented visually. Results Menopausal status,PR,and K-67 were identified as independent predictors.Sixteen optimal radiomics features were selected from 1688 candidates.The combined intratumoral-peritumoral (P&T) radiomics model achieved the highest AUCs of 0.812 (training) and 0.785 (validation).The integrated Nomogram combining radiomics and clinicopathological features further improved performance (AUC 0.829 and 0.790,respectively),outperforming single models (DeLong test,P<0.05).DCA and calibration curves confirmed its superior clinical utility and reliability. Conclusion The Nomogram developed in this study can effectively predict TILs levels in breast cancer and is expected to serve as a new reference for evaluating patient status and guiding individualized treatment strategies.
Objective To investigate the prognostic value of T1-weighted contrast-enhanced MRI imaging classification in patients with lung adenocarcinoma leptomeningeal(LUAD) metastases(LM) who received intrathecal chemotherapy via an Ommaya reservoir. Methods A retrospective analysis was performed on patients with LUAD LM confirmed by cerebrospinal fluid cytology between December 2021 and April 2024, all of whom received intrathecal chemotherapy via an Ommaya reservoir. We evaluated MRI findings and survival outcomes, including intracranial progression-free survival (iPFS) and overall survival (OS).Patients were first categorized into MRI-positive and MRI-negative groups. For MRI-positive patients, lesions were further classified into three types according to enhancement patterns:Type A (purely linear),Type B (purely nodular),and Type C (mixed).The prognostic impact of these imaging subtypes was analyzed using Kaplan-Meier curves,Log-rank tests,and Cox regression models. Results Among the 120 patients, MRI-positive and MRI-negative cases comprised 69 (57.5%) and 51 (42.5%),respectively.MRI positivity was associated with a poorer prognosis and served as an independent risk factor for both shorter iPFS (6.5 vs. 9.0 months,P=0.009) and OS (11.8 vs. 19.5 months,P=0.005). Within the MRI-positive cohort, patients with Type C lesions had significantly worse OS than those with Type B lesions (7.5 vs. 20.0 months,P=0.027), whereas differences in iPFS among the subtypes were not statistically significant.Cox regression analysis revealed that,compared with Type B patients,Type C patients had a significantly increased risk of death(HR 2.461,95%CI：1.069-5.664,P=0.034),establishing it as an independent unfavorable prognostic factor for OS in this cohort.However,the LM imaging classification did not show a significant association with iPFS. Conclusion A T1-weighted contrast-enhanced MRI-based classification provides prognostic value in patients with lung adenocarcinoma-related leptomeningeal metastases undergoing intrathecal chemotherapy via an Ommaya reservoir.Specifically,the mixed enhancement pattern is an independent predictor of poor overall survival.
Objective This study aimed to investigate the predictive value of lesion-specific pericoronary adipose tissue attenuation index (FAI) combined with CT-derived fractional flow reserve (CT-FFR) for major adverse cardiovascular events (MACE) in patients with coronary artery disease (CAD). Methods We retrospectively analyzed the clinical data of 202 patients with chest discomfort or suspected CAD who underwent coronary computed tomography angiography (CCTA) at the Second Affiliated Hospital of Bengbu Medical College from June 2022 to December 2023.Patients were divided into the MACE group(n=76) and non-MACE group(n=126) according to the occurrence of MACE.Univariate Cox regression analysis was used to screen potential risk factors for MACE.Variables with P<0.05 were included in the multivariate Cox proportional hazards regression model to identify independent risk factors.The predictive efficacy of the combined indicator model for MACE was evaluated using receiver operating characteristic (ROC) curves. Results Univariate and multivariate Cox regression analyses demonstrated that lesion-specific pericoronary FAI,CT-FFR,plaque type,and diabetes mellitus were independent risk factors for MACE in patients with CAD. Kaplan‑Meier survival analysis showed that patients with lesion-specific pericoronary FAI>-81.96 HU and CT-FFR≤0.8 had a significantly higher risk of MACE(P<0.001).Comparison between single‑factor and combined models revealed that the combined model incorporating FAI,CT-FFR,plaque type,and diabetes mellitus achieved the optimal predictive performance,with an area under the ROC curve of 0.884.Time‑dependent ROC analysis further confirmed that the combined model exhibited favorable predictive efficacy for MACE in CAD patients at different time points. Conclusion Lesion-specific pericoronary adipose tissue attenuation index,CT-FFR,plaque type, and diabetes mellitus are independent risk factors for MACE in patients with CAD.The combined model of the four indicators shows superior predictive performance and can provide important references for risk stratification and prognostic evaluation in CAD patients.
Objective This study aims to conduct a systematic review to evaluate the performance of ML models in detecting Macrotrabecular-Massive Hepatocellular Carcinoma (MTM-HCC),thereby providing an evidence-based foundation for developing and refining intelligent diagnostic tools. Methods Electronic searches were conducted in PubMed,Embase,Cochrane Library,Web of Science,CNKI,VIP,CBM, and WanFang Data databases to collect studies on machine learning for detecting MTM-HCC from inception to July 21,2025.The Prediction model Risk Of Bias Assessment Tool (PROBAST) was used to assess the risk of bias. Subgroup analyses were performed based on modeling variables (non-radiomics vs. radiomics features)during meta-analysis. Results A total of 21 studies involving 3925 HCC patients,including 1092 with MTM-HCC,were included.Thirteen studies were based on non-radiomics features and 8 on radiomics features.In the validation sets, the pooled sensitivity (SEN) of all ML models was 0.75[95% CI (0.69,0.81)],specificity (SPE) was 0.78[95% CI (0.69,0.84)], positive likelihood ratio (PLR) was 3.4[95% CI (2.4,4.7)], negative likelihood ratio (NLR) was 0.32[95% CI (0.25,0.40)], diagnostic odds ratio (DOR) was 11 [95% CI (7,17)], and summary receiver operating characteristic curve area (SROC-AUC) was 0.80 [95% CI (0.76, 0.83)]. For ML models based on radiomics features, the pooled SEN was 0.78 [95% CI (0.71, 0.84)],SPE was 0.77 [95% CI (0.66, 0.85)], PLR was 3.4 [95% CI (2.3, 5.2)], NLR was 0.29 [95% CI (0.21, 0.38)], DOR was 12 [95% CI (7, 21)], and SROC-AUC was 0.81 [95% CI (0.78, 0.85)]. Conclusion Machine learning is feasible for detecting MTM-HCC, with radiomics models performing marginally better. It holds promise as a potential adjunctive tool for preoperative identification. Nevertheless, the quality of current evidence requires enhancement, and further optimization of model development and validation is needed to improve predictive accuracy.
Objective To investigate the differences in quantitative parameters of intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI) and T2* values of the quadratus femoris muscle among patients with ischiofemoral impingement (IFI),suspected IFI cases,and healthy individuals, analyze their correlations with IFI morphological parameters,and screen for independent influencing factors of IFI. Methods A retrospective analysis was performed on 128 cases of quadratus femoris muscle examinations in Renhe Hospital Affiliated to China Three Gorges University, which were divided into the IFI group (37 cases), the suspected IFI group (48 cases), and the healthy control group (43 cases).All subjects underwent IVIM-DWI and fat analysis calculation technique (FACT) scanning to measure the parameters D,D*,f,rBF,and T2* values.The differences in multimodal MRI parameters among the three groups were compared,and the correlations between these parameters and the ischiofemoral space (IFS),quadratus femoris space (QFS),and visual analog scale (VAS) score were analyzed. Results Significant differences were observed among the three groups in D,D*,f,and T2* values (P<0.05).Post-hoc pairwise comparisons revealed that the observation group had significantly higher D and D* values than both the control and suspected groups,while the suspected group showed higher D values than the control group.The observation group had significantly lower f and T2* values than the control and suspected groups,and the suspected group had lower T2* values than the control group (P<0.05).D and D* values showed moderate and weak negative correlations with IFS (rs= -0.433,r=-0.202, respectively),while f and T2* values showed weak positive correlations with IFS(r=0.248,r=0.352, respectively).D and D* values also showed moderate and weak negative correlations with QFS (rs=-0.409,r=-0.214,respectively), while f and T2* values showed weak positive correlations with QFS (r=0.244,r=0.299,respectively).The D and D* values were strongly and moderately positively correlated with the VAS score,respectively (rs=0.762,r=0.469),and the f value was moderately negatively correlated with the VAS score (r=-0.570). Conclusion The combined application of IVIM-DWI quantitative parameters and T2* values can reveal microstructural changes in the quadratus femoris muscle,clarify the pathophysiological mechanism of IFI,provide novel imaging biomarkers for the accurate diagnosis of IFI,and offer references for early intervention and individualized treatment.
