Non-invasive differentiation of diffuse midline glioma and midline glioblastoma using DCE-MRI perfusion parameters and machine learning classification in pediatric and young adult patients.
In a retrospective cohort of 62 pediatric and young adult patients with histologically confirmed midline gliomas, 95th-percentile DCE-MRI perfusion features distinguished DMG from midline glioblastoma with moderate performance, with a three-feature random forest model reaching 76.67% cross-validation accuracy.
Open original publication →What the AI sees
In a retrospective cohort of 62 pediatric and young adult patients with histologically confirmed midline gliomas, 95th-percentile DCE-MRI perfusion features distinguished DMG from midline glioblastoma with moderate performance, with a three-feature random forest model reaching 76.67% cross-validation accuracy.
Research significance
The study provides evidence that rCBV, rCBF, and Ve differ between DMG and midline glioblastoma; it is reasonable but unproven to hypothesize that a validated imaging classifier could support treatment selection when biopsy is unsafe or unavailable, rather than replace molecular diagnosis.
Source abstract
PURPOSE: Diffuse midline glioma (DMG) and midline glioblastoma (mGBM) are aggressive WHO grade 4 tumors with comparable median survival of 12-18 months but require fundamentally different therapeutic approaches. DMG is molecularly defined by H3 K27-alteration, whereas mGBM lacks this alteration, however, non-invasive differentiation remains challenging due to overlapping conventional MRI features and the difficulty of obtaining tissue diagnosis from eloquent midline locations. METHODS: This retrospective study included 62 patients with histologically confirmed midline gliomas (30 mGBM, 32 DMG) evaluated with 3T MRI. Quantitative DCE-MRI perfusion parameters (rCBV, rCBF, Slope-2, Ktrans, Vp, Ve) were computed and compared between the midline tumor types. Statistical analyses included Shapiro-Wilk test, t-test, and ROC curve analysis using perfusion parameters. Machine learning-based classification was also performed using four classifiers and 5-fold cross-validation, evaluating all possible feature combinations among the best features from the perfusion parameters. RESULTS: Among the statistical features extracted from each parametric map, the 95th percentile values showed the highest discriminative performance for differentiating mGBM from DMG, outperforming the mean, standard deviation, and other percentile measures. DMG exhibited significantly lower 95th percentile perfusion parameter values compared to mGBM (p < 0.05). Individual perfusion parameters, particularly rCBF, rCBV, Ve showed discriminative performance achieving AUC values ranging from 70.62% to 75.31%, for differentiating mGBM vs. DMG. Machine learning classifiers used these features for evaluating 7 combinations. Three parameter combination (rCBV + rCBF + Ve) using a random forest classifier achieved the highest cross-validation accuracy (76.67 ± 7.16%) with consistent sensitivity (80.00%) across all models. CONCLUSIONS: Quantitative DCE-MRI perfusion analysis provides significant diagnostic value for differentiating DMG from mGBM, offering a non-invasive alternative when tissue diagnosis is not obtained. Both individual parameters and optimized multi-parametric approaches demonstrate clinically useful performance for guiding treatment decisions.