Deep learning-based quantification of T2-FLAIR mismatch sign: extending IDH mutation prediction in adult-type diffuse lower-grade glioma

European Radiology, Journal Year: 2025, Volume and Issue: unknown

Published: March 7, 2025

To investigate the predictive value of quantitative T2-FLAIR mismatch ratio (qT2FM) with fully automated tumor segmentation in adult-type diffuse lower-grade gliomas (LGGs). This retrospective study included 218 consecutive patients (mean age, 47 years ± 15 [SD]; 125 males) diagnosed LGG. The cohort was classified into IDH wild-type (IDHwt), IDH-mutant 1p/19q-codeletion (IDHmut-Codel), and without (IDHmut-Noncodel) subtypes. Tumor masks were obtained using deep learning-based segmentation, qT2FM calculated from differences signal intensity ratios on T2 FLAIR images. Multivariable logistic regression identified predictors for identifying IDHmut-Noncodel mutation status. Point-biserial correlations analyzed between qualitative T2FM, median apparent diffusion coefficient (ADC) value. Diagnostic performance evaluated a receiver operating characteristic curve. group had higher (0.37 0.38, p = 0.004) than IDHmut-Codel (0.24 0.39) IDHwt groups (0.07 0.62). only independent imaging predictor (OR 3.43, 95% CI: 1.30-9.05, 0.01). Independent younger age (p < 0.001), frontal lobe location 0.007), cortical involvement 0.034). significantly correlated visual T2FM (vT2FM) ADC Adding to vT2FM improved (AUC 0.77, 0.70-0.82) status 0.71-0.83) each parameter alone. ratio, derived is valuable subtype Question Does deep-learning-based quantification sign provide accurate prediction IDH-mutant, 1p/19q non-codeleted astrocytomas enhance identification status? Findings Quantifying tool achieved high accuracy astrocytomas, enhanced prediction. Clinical relevance Integrating clinical protocols enhances diagnostic precision guides treatment strategies, underscoring role advanced neuro-oncology.

Language: Английский

Clinical and imaging characteristics of supratentorial glioma with IDH2 mutation

Neuroradiology, Journal Year: 2024, Volume and Issue: 66(6), P. 973 - 981

Published: April 24, 2024

Language: Английский