Protocol for CT-Based Renal Stiffness Mapping: Multimodal Imaging and Machine Learning to Predict Elastography Values

Protocol for a prospective 50-patient study aiming to build a machine learning model that predicts MR elastography-derived kidney stiffness directly from routine CT attenuation values, potentially obviating dedicated elastography.

• Design: Monocentric, prospective, non-randomized diagnostic study (n=50 adults) undergoing nephrectomy, with a multimodal imaging protocol including CT, MRE, and US elastography.
• Primary Outcome: Development of a regression model using machine learning to predict MRE-derived elasticity from CT density, validated by repeated nested cross-validation.
• Key Limitation: Monocentric design and small sample size; external validation will not be performed, and the model's utility beyond the study population is unknown (study protocol description only).

RadPigeon summaries are original and for information only. They are not clinical advice.