Tumor Multiregional Mean Apparent Propagator (MAP) Features in Evaluating Gliomas-A Comparative Study With Diffusion...

Area of Science:

• Neuroimaging
• Radiology
• Oncology

Background:

• Accurate glioma classification is crucial for treatment planning and prognosis.
• Non-invasive preoperative imaging techniques are essential for glioma evaluation.

Purpose of the Study:

• To assess the utility of tumor multiregional mean apparent propagator (MAP) features for glioma diagnosis.
• To compare the diagnostic performance of MAP features against diffusion-kurtosis imaging (DKI).

Main Methods:

• Retrospective analysis of 70 untreated glioma patients using 3-T diffusion-MRI with varied b-values.
• Evaluation of MAP metrics (MSD, QIV, NG, NGAx, NGRad, RTOP, RTAP, RTPP) and DKI metrics (AK, MK, RK).
• Statistical analysis included Mann-Whitney U, Kruskal-Wallis, ROC analysis, and Random Forest modeling.

Main Results:

• MAP and DKI metrics differed significantly between low-grade gliomas (LGGs) and high-grade gliomas (HGGs) in tumor parenchyma (TP) and peritumoral areas (PT).
• MAP and DKI metrics also showed significant differences between isocitrate-dehydrogenase (IDH)-mutated and IDH-wildtype gliomas.
• Random Forest analysis demonstrated high accuracy for grading (82%) and IDH genotyping (79%) using combined MAP and DKI features.

Conclusions:

• Tumor multiregional MAP features are effective for glioma evaluation.
• MAP features show comparable performance to DKI in tumor parenchyma.
• MAP features may offer superior performance to DKI in peritumoral areas for glioma grading.