
Accurate preoperative assessment of tumor-internal os distance (TID) is essential for selecting candidates for fertility-sparing surgery in early-stage cervical cancer. This study evaluated the interobserver agreement of DKI-assisted MRI-based TID measurements and their diagnostic performance for pathological safe-distance status and fertility-sparing feasibility in FIGO stage IB cervical cancer. Patients with stage IB cervical cancer who underwent preoperative pelvic MRI with sagittal DKI between September 2016 and June 2026 were prospectively enrolled. TID was measured using sagittal T2-weighted imaging alone (T2WI_TID) and DKI-assisted MRI (MR_TID). Interobserver agreement, diagnostic performance for pathological safe-distance status and final fertility-sparing feasibility, and model performance were evaluated. A prediction model was developed in a training cohort and tested in an internal testing cohort. A total of 170 patients were included. MR_TID showed higher interobserver agreement than T2WI_TID (intraclass correlation coefficient, 0.894 vs. 0.802). For binary pathological safe-distance status, MR_TID showed better discrimination than T2WI_TID (area under the receiver operating characteristic curve, 0.946 vs. 0.840; P < 0.001). For final fertility-sparing feasibility, MR_TID also outperformed T2WI_TID (0.877 vs. 0.806, P = 0.021), with an exploratory cutoff of 10.5 mm. A model incorporating MR_TID, MR-based tumor diameter, deep stromal invasion, parametrial invasion, and lymph node metastasis showed good discrimination in the training and testing cohorts (area under the receiver operating characteristic curve, 0.921 and 0.900, respectively). DKI-assisted MRI improved TID measurement reproducibility and binary pathological safe-distance discrimination, although its role in fertility-sparing selection requires external validation.
CT-derived body composition is associated with outcomes in esophageal cancer, but its added value beyond routine pretreatment risk assessment remains uncertain. We evaluated whether pretreatment L3 body composition improves survival prediction beyond a prespecified clinical-inflammatory model in ESCC treated with definitive concurrent chemoradiotherapy. This retrospective two-center study included 296 patients with nonmetastatic ESCC (development, n = 201; external validation, n = 95). Skeletal muscle index (SMI), skeletal muscle radiation attenuation (SMRA), visceral-to-subcutaneous adipose tissue ratio (VAT/SAT), and intermuscular adipose tissue index (IMATI) were measured on portal venous phase CT at L3. A clinical-inflammatory reference model (Model A) was compared with Model B, which added the four CT measures. Cox models underwent 1000-resample bootstrap internal validation and locked external validation without refitting or recalibration. In development, Model B increased the optimism-corrected C-index from 0.606 to 0.659 (difference, 0.053; 95
This study aimed to develop and validate an interpretable machine learning (ML) model that integrates clinical data, normalized distance from the metabolic hotspot to the tumor centroid (NHOC) / normalized distance from the metabolic hotspot to the tumor perimeter (NHOP) metrics, and PET/CT habitat radiomics to predict recurrence in lung adenocarcinoma (LUAD) and perform postoperative risk stratification. In this retrospective study, we analyzed patients with resectable stage Ⅰ LUAD from two hospitals and a public dataset, comprising a training cohort (n = 253) and an internal validation cohort (n = 106), including a subset with radiogenomics data (n = 26). NHOC and NHOP were extracted using LIFEx software. Nine machine learning algorithms were evaluated to identify the optimal radiomics features. The best-performing algorithm was used to construct predictive models incorporating clinical variable, NHOC/NHOP, and habitat-based radiomics features. Shapley Additive Explanations (SHAP) were employed to interpret model contributions, while paired RNA-sequencing explored potential biological pathways. Smoking history and NHOCpeak were identified as independent predictors constituting the clinical model. The combined model demonstrated moderate discrimination performance for recurrence prediction. achieved an AUC of 0.876 (95
This study aimed to develop and evaluate a multimodal predictive system (MPS) for assessment of renal sinus fat invasion (RSFI) in clear cell renal cell carcinoma (ccRCC) patients. A total of 875 ccRCC patients from four independent institutions were retrospectively analyzed. The cohort was divided into a training set (n = 445), validation set (n = 111), and two multicentre evaluation sets (n = 181, n = 138). Radiomics features were extracted from the intratumoral ROI and tumor-plus-peritumoral ROIs expanded by 1, 3, and 5 mm. Multiple machine learning classifiers were constructed and evaluated. In parallel, 2D, 2·5D, and 3D deep learning (DL) models based on ResNet34 were trained. A clinical model was developed using significant variables. The representative models from each modality were integrated via stacking ensemble learning to construct the MPS. Diagnostic performance was evaluated by AUC, sensitivity, specificity, and accuracy, while model interpretability was analyzed using SHAP. The peri-3 mm ExtraTrees radiomics model achieved the highest AUC values of 0.753 (95
To evaluate the potential of mpMRI and ADC values to predict unfavorable disease (UFD) and biochemical recurrence (BCR)/PSA persistence in patients with Grade Group 1 (GG1) or GG2 prostate cancer (PCa) undergoing surgery, with potential implications for active surveillance (AS) risk stratification. We retrospectively selected 689 patients with GG1 or GG2 PCa in the peripheral zone operated between 2012 and 2015. The outcomes of interest were UFD (GG ≥ 3 and/or pT ≥ 3a and/or pN1) and biochemical recurrence (BCR) /PSA persistence. Logistic regression models assessed predictors including age, PSA density, GG at biopsy and percentage of positive cores (Model 1). Additional models incorporated PI-RADS category, extraprostatic extension (EPE) score, and ADC values. 366 patients (53.1