Prognostic Impact of Pelvic Bone Magnetic Resonance Imaging Features in Patients with Cervical Cancer Undergoing Definitive Concurrent Chemoradiotherapy

This dataset contains the imaging-derived resources and supplementary results used in the study "Pelvic Bone T1-weighted MRI for Generalizable Cervical Cancer Survival Prediction". Contents Detailed Results Per-run and aggregated survival analysis results (CSV). Figures (PNG/PDF) of time-dependent C-index, AUC, and feature importance. Feature Tables Radiomics features (IBSI-compliant, CSV). Deep learning feature embeddings from BEiT and other backbones. Clinical features aligned by PatientID. Segmentation Masks Pelvic bone segmentation masks (NIfTI, de-identified). Correspond to preprocessed T1-weighted MRI slices. Dataset Metadata Train/validation splits for 30 runs. PatientID mapping (de-identified). Preprocessing workflow description. Notes All imaging data are de-identified, with no PHI included. For analysis code and computational environment, see the companion code repository (DOI: [insert link]). KeywordsPelvic bone, Cervical cancer, Survival prediction, MRI, Radiomics, Deep learning LicenseCC-BY-NC 4.0