AUTOMATED NEPHROMETRY SCORES THROUGH DIRECT PREDICTION OF EACH COMPONENT

The RENAL and PADUA nephrometry scores are tools often used in managing kidney tumors that offer quantifiable measures of tumor complexity to guide surgical planning and predict outcomes. Though clinically useful, the manual calculation of these scores is labor-intensive and prone to significant interobserver variability, resulting in inconsistencies in diagnostic interpretation that could result in suboptimal outcomes. This study explores using artificial intelligence (AI) to automate the generation of RENAL and PADUA scores. By utilizing machine learning models trained on a comprehensive dataset of abdominal CT scans, the study aims to enhance the reliability and efficiency of nephrometry scoring, potentially improving the precision of clinical decision-making in kidney cancer management. Whereas prior approaches automate nephrometry scores through segmentation followed by geometric measurements, this study employs a deep neural network to predict each component directly, streamlining the process and potentially enhancing accuracy.