Nested Habitat Analysis Based on MRI for Prediction of Progression-Free Survival in Aggressive Spinal Tumors

Motivation: The accurate prognosis of aggressive spinal tumors is crucial for treatment planning. Goal(s): This study introduces a novel MRI-based nested habitat radiomics method, designed to identify the most aggressive regions within heterogeneous spinal tumors. Approach: By analyzing a cohort of 259 patients with primary spinal tumors who underwent surgical resection, this method uses MRI features in a layered approach to predict progression-free survival (PFS). Results: The model's performance was validated with AUC of 0.88 with the habitat analysis outperforming traditional whole-tumor models. The nested habitat model could serve as a noninvasive prognostic tool, providing valuable insights for individualized surgical and post-surgical treatment plans. Impact: This study's MRI-based nested habitat radiomics model enhances prediction accuracy of progression-free survival (PFS) in aggressive spinal tumors by focusing on the most aggressive regions, outperforming traditional models and providing valuable insights for personalized surgical and post-surgical treatment strategies.