Preoperative imaging annotations for cancer surgery were used to train AI for automatic image annotation, feature extraction, and analysis. This was further utilized to develop liver cancer prognosis prediction models. We provided preoperative CT images of liver cancer patients, including non-contrast phase (N), arterial phase (A), portal venous phase (P), and delayed phase (D) original images, along with corresponding tumor annotations (RTSS). Each phase consisted of approximately 40-50 images (depending on the actual phases executed during the examination, not all phases may be present).