Two separate imaging AI studies point toward noninvasive risk stratification before surgery. In gastric cancer, researchers in Guangzhou built a machine-learning model combining MRI radiomics and clinical variables to predict lymphovascular invasion and stratify disease-free survival risk preoperatively. Separately, researchers in Nanjing developed an interpretable model that analyzes MRI tumor subregions to predict prostate cancer aggressiveness noninvasively before biopsy, aiming to improve how patients are selected for invasive diagnostic pathways. Together, the studies emphasize interpretable radiomics approaches designed to guide clinical decision-making at key pre-procedure junctures, where avoiding unnecessary interventions and focusing on likely aggressive biology can affect both outcomes and resource use.