Radiomics uses a large number of medical imaging features and can demonstrate voxel-wise intratumor heterogeneity. We calculated the radiomic signature for each patient using a weighted linear combination of the radiomic features selected by machine learning methods. The study endpoint was DFS, defined as the interval between TME surgery and disease progression, which included tumor local recurrence, distant metastasis, or death, or the date of the last follow-up visit (censored). The association between the radiomic signature and DFS was explored. Then, the three models were built to estimate the DFS in patients.