Multimodal MRI-based clinical radiomics model for predicting postoperative anemia in patients with spinal metastases

Motivation: Anemia is a common complication after spinal metastasis surgery. Goal(s): To explore the role of MRI-based radiomics models in predicting postoperative anemia. Approach: The patients diagnosed with spinal metastases and underwent surgery from December 2012 to December 2023 were divided into postoperative anemia and non-anemia groups. Radiomics features were extracted from ROI on sagittal traditional sequences of preoperative MRI scans. Radiomics models were developed using logistic regression analysis, supported by a five-fold cross-validation technique. After stepwise logistic regression analyses, three models —clinical, Radscore, and combined models —were developed. Results: The combined clinical and Radscore based on T1WI and FS-T2WI sequences model had the best performance. Impact: Accurate prediction of postoperative anemia in patients with symptomatic spinal metastases is crucial for effective management, and the model incorporating clinical and Radscore factors is capable of enhancing treatment planning and outcomes.