This study was conducted by Blondeau et al. where they examined whether AI-based CT imaging can reliably measure sarcopenia and whether sarcopenia predicts survival outcomes within patients who have muscle-invasive bladder cancer. The research was conducted as a retrospective study analyzing clinical, biological, and imaging data from bladder cancer patients who were being treated with neoadjuvant chemotherapy and then a cystectomy. Pre-treatment and pre-surgery CT scans were analyzed using both manual measurements and an AI-based deep learning tool with a focus on skeletal muscle area at the L3 level to determine sarcopenia through specific criteria. The study showed that AI-based skeletal muscle area measurements were highly consistent with manual measurements, having strong correlations, minimal measurement bias, and notable inter-observer agreement. The prevalence of sarcopenia varied widely depending on the definition used, ranging from 32.2% to 54%, and undernutrition before surgery was present in 12.6% of patients. Sarcopenia, defined by Prado’s criteria before chemotherapy and before surgery, as well as by Caan’s criteria before surgery, was associated with worse overall survival, while only Prado-defined sarcopenia before chemotherapy was associated with poorer progression-free survival. Furthermore, undernutrition and elevated neutrophil-to-lymphocyte ratio before surgery were strongly associated with reduced overall survival. The discussion shows that AI-based assessment of body composition is a reliable, rapid, and reproducible alternative to manual segmentation and has the potential to be integrated into routine evaluation of patients with muscle-invasive bladder cancer. It also emphasizes that sarcopenia is associated with poorer survival, but should be interpreted alongside other clinical and biological factors rather than as a standalone prognostic marker.