Multiphase CT-based deep learning radiomics nomogram models for preoperative WHO/ISUP grading of clear cell renal cell carcinoma: a two-center validation study

Key Points

• Assess the effectiveness of DLRN models for predicting nuclear grade in clear cell renal cell carcinoma using multiphase CT imaging.
• Used multiphase CT imaging data for model training and validation.
• Developed deep learning radiomics models to analyze imaging features.
• Carried out a two-center validation to ensure model reliability.”],
• results
• observed improvements in predictive accuracy for nuclear grade classification using DLRN models compared to conventional methods.
• Observed improvements in predictive accuracy for nuclear grade classification using DLRN models compared to conventional methods.
• The DLRN model showed a strong correlation with actual nuclear grades post-surgery.
• Demonstrated that non-invasive CT-based assessments can effectively replace more invasive grading procedures.