DECISION TREE FOR DIFFERENTIATING HIGH-GRADE GLIOMA, SOLITARY BRAIN METASTASIS, AND PRIMARY CENTRAL NERVOUS SYSTEM LYMPHOMA USING MULTIPARAMETRIC MAGNETIC RESONANCE IMAGING
Main Article Content
Abstract
Objectives: To develop a decision tree model based on diffusion tensor imaging (DTI), magnetic resonance perfusion (MRP), and magnetic resonance spectroscopy (MRS) to differentiate high-grade glioma (HGG), solitary brain metastasis (SBM), and primary central nervous system lymphoma (PCNSL). Methods: A cross-sectional descriptive study was conducted on 64 patients (23 PCNSL, 17 SBM, 24 HGG) who underwent magnetic resonance imaging prior to surgery/biopsy at Viet Duc University Hospital from August 2025 to March 2026. FA, MD, rCBV, rCBF values, and Cho/NAA, Cho/Cr ratios were quantified in the solid tumor (t) and peritumoral (p) regions. The model was constructed using the CART (classification and regression tree) algorithm. Results: The CART algorithm utilised rCBVp, MDt, and Cho/NAAp as classification nodes. The branch with rCBVp > 1.043 predicted a high probability of HGG (94.7%). For the branch with rCBVp £ 1.043, the thresholds of MDt (0.785) and Cho/NAAp (0.726) facilitated the differentiation between SBM and PCNSL. The overall accuracy of the model reached 76.6% (PCNSL: 84.4%; HGG: 79.7%; SBM: 89.1%), with a cross-validation error rate of 35.9%. Conclusion: The CART model integrating rCBVp, MDt, and the Cho/NAAp ratio is a potential tool, with high accuracy in the differential diagnosis of HGG, SBM, and PCNSL.
Keywords
Decision tree, CART algorithm, Multiparametric magnetic resonance imaging, High-grade glioma, Solitary brain metastasis, Primary central nervous system lymphoma
Article Details
References
2. Vallée R, Vallée JN, Guillevin C, et al. Machine learning decision tree models for multiclass classification of common malignant brain tumors using perfusion and spectroscopy MRI data. Frontiers in Oncology. 2023;13:1089998.
3. Panyaping T, Wimolsiri N, Sananmuang T. Diagnostic accuracy of differentiation between primary central nervous system lymphoma and high-grade glioma using dynamic susceptibility contrast perfusion curve analysis. Journal of The Medical Association Of Thailand. 2020;103(9):920-25.
4. Wang S, Kim S, Chawla S, et al. Differentiation between glioblastomas, solitary brain metastases, and primary cerebral lymphomas using diffusion tensor and dynamic susceptibility contrast-enhanced MR imaging. American Journal of Neuroradiology. 2011;32(3):507-14.
5. Liang R, Wang X, Li M, et al. Meta-analysis of peritumoural rCBV values derived from dynamic susceptibility contrast imaging in differentiating high-grade gliomas from intracranial metastases. International Journal of Clinical and Experimental Medicine. 2014;7(9):2724-9.
6. Würtemberger U, Rau A, Reisert M, et al. Microstructural differentiation of cerebral metastases, glioblastoma, meningioma, and primary CNS lymphoma using advanced diffusion imaging techniques. Neuro-Oncology Advances. 2025;7(1):vdaf076.
7. Neska-Matuszewska M, Bladowska J, Sąsiadek M, Zimny A. Differentiation of glioblastoma multiforme, metastases and primary central nervous system lymphomas using multiparametric perfusion and diffusion MR imaging of a tumor core and a peritumoral zone-searching for a practical approach. PLOS One. 2018;13(1):e0191341.
8. Song YY, Lu Y. Decision tree methods: applications for classification and prediction. Shanghai Archives of Psychiatry. 2015;27(2):130-5.