AI-based imaging tools are increasingly being validated for preoperative risk stratification and surgical decision-making. One study in breast cancer reported a machine learning model that predicts which patients may skip unnecessary surgery by estimating pathological complete response after neoadjuvant chemotherapy, combining ultrasound, mammography, MRI, and clinical inputs. Separately, an approach for liver surgery planning tested geometric, perfusion-based, and deep learning models against each other, finding each excels in different surgical scenarios. The result suggests modular decision-support tools may be more realistic than one-size-fits-all models for complex resection planning. These efforts converge on a key clinical goal: reducing overtreatment while preserving oncologic outcomes, and shifting AI from detection toward workflow-integrated planning.