DCE-MRI and IVIM-based habitat imaging for differentiating undifferentiated pleomorphic sarcoma and myxofibrosarcoma
Objectives Since diagnostic difficulty and distinct prognoses between undifferentiated pleomorphic sarcoma (UPS) and myxofibrosarcoma (MFS), accurate preoperative differentiation is essential to guide clinical decisions. The aim of this study was to evaluate the potential of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and intravoxel incoherent motion (IVIM)-based habitat imaging to differentiate between UPS and MFS. Materials and methods This retrospective study included 31 UPS and 28 MFS patients with pathologically confirmed who underwent 3.0-T MRI, including conventional MRI, IVIM, and DCE-MRI sequences. The volume transfer constant (Ktrans) and true diffusion coefficient (D) maps were generated from DCE-MRI and IVIM data, respectively. After whole-tumor segmentation, Ktrans and D maps were clustered using K-means algorithm to define habitats. Optimal number of habitats was determined using elbow method and silhouette score. Differences in clinical characteristics, conventional MRI features, and habitat features were compared between UPS and MFS groups. Logistic regression was used to construct diagnostic models. Receiver operating characteristic (ROC) curve and the DeLong test were performed to assess diagnostic performance, and decision curve analysis (DCA) and calibration curve were used to evaluate the clinical benefit and calibration performance of models respectively. Results Three habitats were identified, including Habitat 1 (hypervasopermeability), Habitat 2 (hypo-vasopermeability and hyper-cellularity), and Habitat 3 (hypo-vasopermeability and hypo-cellularity). Several habitat features derived from DCE-MRI and IVIM-based habitat imaging differed significantly between UPS and MFS ( p < 0.05). Both the Combined_Model (AUC = 0.957, 95%CI: 0.870-0.993) and Habitat_Model (AUC = 0.932, 95% CI: 0.835-0.981) achieved significantly better diagnostic performance than Conventional_MRI_Model (AUC = 0.757, 95% CI: 0.628-0.860). All the models constructed on the basis of habitats features had good clinical benefit and calibration performance (all Brier<0.15). Conclusion DCE-MRI and IVIM-based habitat imaging can help differentiate UPS from MFS.
Authors
- Yuejun Liu
- Zesheng Zhang
- Xu Wang
- Shaowu Wang
- Wenjia Wang
- Kai Zhang
- Jie Zhou
- Chunlian Chen
Institutions
- Dalian University of Technology (CN)
- Second Affiliated Hospital of Dalian Medical University (CN)
- Jinzhou Medical University (CN)
Publication Details
- Journal
- Frontiers in Oncology
- Published
- 2026-11-16
- DOI
- https://doi.org/10.3389/fonc.2026.1937873
- Primary Topic
- MRI in cancer diagnosis
- Type
- article
- Field-Weighted Citation Impact
- 0.00