Deep learning algorithm excels in kidney trauma detection

A deep learning model, RenotrNet, demonstrated high accuracy in detecting kidney trauma on CT scans, achieving 0.88 accuracy in internal testing and 0.93 in external validation with RSNA data. The model showed robust performance metrics, including 0.75 sensitivity and 0.95 specificity internally, and 0.73 sensitivity and 0.94 specificity externally. Its high negative predictive value of 0.98 suggests substantial reliability, highlighting its potential for clinical deployment in trauma diagnosis.