Background: The assessment of the International Society of Urological Pathology (ISUP) nuclear grade is crucial for the management and treatment of clear cell renal cell carcinoma (ccRCC). This study aimed to explore the value of using integrated multimodal information for ISUP grading and prognostic stratification in ccRCC patients, to guide postoperative adjuvant therapy. Methods: This retrospective study analyzed a total of 729 patients from three cohorts, utilizing whole slide i-mages and computed tomography (CT) images. Artificial intelligence algorithms were used to extract morphological and textural features from whole slide images and CT images separately, creating single-modality predictive models for ISUP grading. By combining the CT and pathology single-modality predictive features, a multimodal predictive signature (MPS) was developed. The prognostic performance of the MPS model was further validated in two independent cohorts. Results: The single-modality predictive models for CT and pathology performed well in predicting ISUP grade for ccRCC. The MPS model achieved higher area under the curve values of 0.95, 0.93, and 0.95 across three independent patient cohorts. Additionally, the MPS model was able to distinguish patients with poorer overall survival. In the external validation cohort, uni- and multivariate analyses showed hazard ratios of 2.542 (95% confidence interval [CI]: 1.363–4.741, P < 0.0001) and 1.723 (95% CI: 0.888–3.357, P = 0.003), respectively. The C-index values for the two cohorts were 0.75 and 0.71. Furthermore, the MPS outperformed single-modality models, providing a complementary tool for current risk stratification in ccRCC adjuvant therapy. Conclusion: Our novel MPS model demonstrated high accuracy in ISUP grading for ccRCC patients. With further validation across multiple centers, the MPS model could be used for precise detection of nuclear grading in ccRCC, serving as an effective tool for assisting clinical decision-making.

Objectives: To automatically identify and diagnose bladder outflow obstruction (BOO) and detrusor underactivity (DUA) in male patients with lower urinary tract symptoms through urodynamics exam.Patients and Methods: We performed a retrospective review of 1949 male patients who underwent a urodynamic study at two institutions. Deep Convolutional Neural Networks scheme combined with a short‐time Fourier transform algorithm was trained to perform an accurate diagnosis of BOO and DUA, utilizing five‐channel urodynamic data (consisting of uroflowmetry, urine volume, intravesical pressure, abdominal pressure, and detrusor pressure). We used fivefold cross‐validation, constructing training and internal test sets from 1725 patients from Renmin Hospital of Wuhan University (RHWU) at a 4:1 ratio, and used an independent external validation set consisting of 224 patients from The Central Hospital of Wuhan (TCHO) to build and evaluate the DI model. We further conducted subgroup analyses to provide a more detailed description of the AI model's interpretability regarding urodynamics.Results: The AUC scores of BOO and DUA, which were measured through the STFT‐based deep learning method, were 0.945 ± 0.020 and 0.929 ± 0.039 in RHWU and 0.881 and 0.850 in TCHO, respectively. The diagnostic efficiency of other subgroup analyses and indicators was also effective.Conclusion: In this study, the proposed deep neural network combined with the short‐time Fourier transform method is robust and feasible for interpreting the results of urodynamics in men and has the potential for application to assist clinicians in real clinical settings.

To explore new metrics for assessing radical prostatectomy difficulty through a two-stage deep learning method from preoperative magnetic resonance imaging. The procedure and metrics were validated through 290 patients consisting of laparoscopic and robot-assisted radical prostatectomy procedures from two real cohorts. The nnUNet_v2 adaptive model was trained to perform accurate segmentation of the prostate and pelvis. A modified network PointNet was used for indirectly regressing 15 anatomical landmarks based on Gaussian heatmaps. Novel metrics proposed in this study that characterized the spatial relationship between the prostate and pelvis were included to evaluate the surgical difficulty. The two-stage process achieved decent segmentation and landmark localization results with the Mean Validation Dice of 0.8641 and millimeter-level accuracy. We found the coefficients of PV, ρ, PT, PAP, AG, PSD₁, PSD₂, πρ²/ISTA, AG+PG, AG×PG, PSD₂×ρ, PAP/(AG+PG) with Estimated Blood Loss and PSD₂, PSD₂×ρ with Operation Time, respectively with statistic significant, which provides possibilities for assessing surgical difficulty evaluation. The entire pipeline had been validated on the external dataset, and the results were consistent. The two-stage anatomical landmark localization approach is feasible. Indicators describing pelvic-prostate spatial constraints significantly impact surgical difficulty in radical prostatectomy, leading to increased blood loss and longer operation times, while isolated pelvic measurements have minimal effect on surgical outcomes.

Background: The purpose of this study was to investigate the value of machine learning models for preoperative non-invasive prediction of International Society of Urological Pathology (ISUP) grading in clear cell renal cell carcinoma (ccRCC) based on CT urography (CTU)-related peritumoral area (PAT) radiomics features. Methods: We retrospectively analysed 328 ccRCC patients from our institution, along with an external validation cohort of 175 patients from The Cancer Genome Atlas. A total of 1218 radiomics features were extracted from contrast-enhanced CT images, with LASSO regression used to select the most predictive features. We employed four machine learning models, namely, Logistic Regression (LR), Multilayer Perceptron (MLP), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost), for training and evaluation using Receiver Operating Characteristic (ROC) analysis. The model performance was assessed in training, internal validation, and external validation sets. Results: The XGBoost model demonstrated consistently superior discriminative ability across all datasets, achieving AUCs of 0.95 (95% CI: 0.92–0.98) in the training set, 0.93 (95% CI: 0.89–0.96) in the internal validation set, and 0.92 (95% CI: 0.87–0.95) in the external validation set. The model significantly outperformed LR, MLP, and SVM (p < 0.001) and demonstrated prognostic value (Log-rank p = 0.018). Transcriptomic analysis of model-stratified groups revealed distinct biological signatures, with high-grade predictions showing significant enrichment in metabolic pathways (DPEP3/THRSP) and immune-related processes (lymphocyte-mediated immunity, MHC complex activity). These findings suggest that peritumoral imaging characteristics provide valuable biological insights into tumor aggressiveness. Conclusions: The machine learning models based on PAT radiomics features of CTU demonstrated significant value in the non-invasive preoperative prediction of ISUP grading for ccRCC, and the XGBoost modeling had the best predictive ability. This non-invasive approach may enhance preoperative risk stratification and guide clinical decision-making, reducing reliance on invasive biopsy procedures.

Chronic kidney disease and its progression to end-stage renal disease represent a major global health burden, driven by renal interstitial fibrosis. The fu