Grading of clear cell renal cell carcinoma using diffusion MRI with a multimodal apparent diffusion model

Frontiers in Oncology, Journal Year: 2025, Volume and Issue: 15

Published: March 20, 2025

Objective To assess the feasibility of utilizing parameters derived from a multimodal apparent diffusion (MAD) model to distinguish between low- and high-grade clear cell renal carcinoma (ccRCC). Method Diffusion-weighted imaging (DWI) scans with 12 b-values (0 - 3000 s/mm²) were conducted on 54 patients diagnosed ccRCC (30 low-grade 24 high-grade). The MAD parameters, including coefficients (D r, D h , ui f ) representing restricted diffusion, hindered unimpeded flow, respectively, computed. Proportions corresponding these types (f r heterogeneous nature (α also obtained. Parameters compared groups. Receiver operating characteristic (ROC) curves used evaluate diagnostic performance coefficient (ADC) mono-exponential model. Result Significant differences observed in (low-grade: 1.360 ± 0.11 μm 2 /ms; group, 1.254 0.13 P = 0.0327), 0.06 0.005; high-grade: 0.08 0.009; 0.0233), α 0.872 0.22; 0.896 0.39; 0.0294). Additionally, ADC values 0.924 0.854 0.04 0.0323) showed statistical significance. combination provided highest accuracy 0.667, sensitivity 0.750, specificity 0.734, area under curve 0.796, outperforming individual ADC. Conclusion shows promise as non-invasive tool for distinguishing ccRCC, which may aid preoperative planning personalized treatment strategies.

Language: Английский

---

Preoperative prediction of WHO/ISUP grade of ccRCC using intratumoral and peritumoral habitat imaging: multicenter study

Cancer Imaging, Journal Year: 2025, Volume and Issue: 25(1)

Published: May 3, 2025

Abstract Objectives The World Health Organization/International Society of Urological Pathology (WHO/ISUP) grading clear cell renal carcinoma (ccRCC) is crucial for prognosis and treatment planning. This study aims to predict the grade using intratumoral peritumoral subregional CT radiomics analysis better clinical interventions. Methods Data from two hospitals included 513 ccRCC patients, who were divided into training (70%), validation (30%), an external set (testing) 67 patients. Using ITK-SNAP, radiologists annotated tumor regions interest (ROI) extended surrounding areas by 1 mm, 3 5 mm. K-means clustering algorithm region three sub-regions, Least Absolute Shrinkage Selection Operator (LASSO) regression identified most predictive features. Various machine learning models established, including models, based on heterogeneity (ITH) score, comprehensive models. Predictive ability was evaluated receiver operating characteristic (ROC) curves, area under curve (AUC) values, DeLong tests, calibration decision curves. Results combined model showed strong power with AUC 0.852 (95% CI: 0.725–0.979) test data, outperforming individual ITH score highly precise, AUCs 0.891 0.854–0.927) in training, 0.877 0.814–0.941) validation, 0.847 0.725–0.969) testing, proving its superior across datasets. Conclusion A combining Habitat, Peri1mm, salient features significantly more accurate predicting pathologic grading. Key points Question : Characterize non-invasively WHO/ISUP pathological preoperatively. Findings An integrated subregion characterization, characteristics, can Clinical relevance Subregion characterization outperforms single-entity approach. model, compared boosts prognostic accuracy targeted actions.

Language: Английский

---

Machine Learning-Based Prediction of Consistency and Histological Characteristics in Renal Cell Carcinoma Venous Tumor Thrombus Through Volumetric Radiomics

Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: May 7, 2025

Background. Renal cell carcinoma (RCC) possesses a distinctive inclination to infiltrate the inferior vena cava, resulting in formation of venous tumour thrombus (VTT). Accurately assessing consistency VTT prior surgery is essential for effective treatment strategizing and favourable results. The study aimed investigate performance volumetric radiomic MRI analysis prediction histomorphological vascular patterns RCC (VTT) with assistance machine learning. Methods. Twenty-four patients underwent nephrectomy thrombectomy this study. Preoperatively abdominal DW-MRI was conducted, followed by creation 3D model thrombus. First-order features were computed from complete volume utilizing ADC maps. The immunohistochemical staining performed using CD34, SMA VEGFR. learning employed develop predictive models histologic features. Patients grouped based on into either solid or friable categories. Results. thrombi detected 13 (54.2%) 11 (45.8%) cases, respectively. Large vessels predominantly observed VTTs (73.3%; p=0.015). Rich vascularization main pattern at 51.5%, contrasting 9.1% (p=0.008). There strong association between vessel size following features: entropy (r=0.722), skewness (r=0.635), mean (r=0.610). outperformed distinguishing large small vessels, achieving highest (AUC 0.930; p<0.001). In rich poor vascularization, median showed best = 0.881; p < 0.001). Using analysis, we've developed two predicting crucial traits prognostic accuracies 89% 75% size. Conclusions. Leveraging data MR-DWI, along models, we identified unique vascular patterns among patients. These predict DWI.

Language: Английский

---

Diagnostic value of an interpretable machine learning model based on clinical ultrasound features for follicular thyroid carcinoma

Quantitative Imaging in Medicine and Surgery, Journal Year: 2024, Volume and Issue: 14(9), P. 6311 - 6324

Published: Aug. 22, 2024

Follicular thyroid carcinoma (FTC) and follicular adenoma (FTA) present diagnostic challenges due to overlapping clinical ultrasound features. Improving the diagnosis of FTC can enhance patient prognosis effectiveness in management. This study seeks develop a predictive model for based on features using machine learning (ML) algorithms assess its effectiveness.

Language: Английский

---

Multi‐instance learning for identifying high‐risk subregions associated with synchronous distant metastasis in clear cell renal cell carcinoma

Medical Physics, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 1, 2024

Abstract Background Clear cell renal carcinoma (ccRCC) is one of the most common histological subtypes tumors. Purpose To identify high‐risk subregions associated with synchronous distant metastasis. Methods This study enrolled a total 277 patients ccRCC. Voxel intensity and local entropy values were compiled within region interest for all patients. Unsupervised k ‐means clustering yielded three per tumor. Radiomic features extracted, random forest‐based feature selection was conducted. The selected used in multi‐instance support vector machine (mi‐SVM) model training, predictions made on validation cohort. Model performance evaluated using five‐fold cross‐validation. subregion highest score metastasis identified across cohorts. Results mi‐SVM an average area under curve (AUC) 0.812 training cohort 0.805 In entire metastasis, 2, characterized by tumor periphery intratumoral transitional components, accounted proportion (48.57%, 30.6/63) among subregions. It represents clear carcinoma. Conclusion peripheral transition zones are

Language: Английский

---