This study aims to assess multimodal Radiomics-based prediction model for prognostic prediction in spinal tumors.

Sensitivity for prediction recurrence and survival of currently available prognostic scores is limited. This study proposes to establish a multimodal radiomics model for identifying tumor recurrence and prognostic prediction of spinal tumors. The study will investigate the relationship between the radiomics and the tumor microenvironment. The study includes the construction of multimodal radiomics-based prediction model and the validation of the prediction model.

Inclusion Criteria:

A diagnosis of spinal tumors for which SBRT is appropriate Karnofsky Performance Status >60 Life expectancy of at least 3 months No contraindication to undergoing MR imaging Age >18 Complete the CT MRI or PET-CT examination before and after treatment Accept stereotactic radiosurgery

Exclusion Criteria:

Subjects unable to undergo MRI (includes non-MRI compatible material or devices and severe claustrophobia) Subjects with prior procedural intervention to the involved vertebral body that would result in artifact (kyphoplasty, screw and/or rod placement); minimally invasive surgery without instrumentation of the involved vertebral body and instrumentation immediately above or below the index lesion is allowed.

Subjects with spinal cord compression; minimally invasive "separation" surgery to first resect the epidural component is allowed Pregnant or breast-feeding women Allergy to standard IV contrast agents used in MRI

Patients who had spinal tumors and completed CT, MRI, PET-CT examination before and after stereotactic radiosurgery.