Nevin Manimala Statistics

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

J Magn Reson Imaging. 2023 Dec 22. doi: 10.1002/jmri.29202. Online ahead of print.


BACKGROUND: Glioma classification affects treatment and prognosis. Reliable imaging methods for preoperatively evaluating gliomas are essential.

PURPOSE: To evaluate tumor multiregional mean apparent propagator (MAP) features in glioma diagnosis and to compare those with diffusion-kurtosis imaging (DKI).

STUDY TYPE: Retrospective study.

SUBJECTS: 70 untreated glioma patients (31 LGGs (low-grade gliomas), 34 women; mean age, 47 ± 12 years, training (60%, n = 42) and testing cohorts (40%, n = 28)).

FIELD STRENGTH/SEQUENCE: 3-T, diffusion-MRI using q-space Cartesian grid sampling with 11 different b-values.

ASSESSMENT: Tumor multiregional MAP (mean squared displacement (MSD); q-space inverse variance (QIV); non-Gaussianity (NG); axial/radial non-Gaussianity (NGAx, NGRad); return-to-origin/axis/plane probability (RTOP, RTAP, and RTPP)); and DKI metrics (axial/mean/radial kurtosis (AK, MK, and RK)) on tumor parenchyma (TP) and peritumoral areas (PT) in histopathologically gliomas grading and genotyping were assessed.

STATISTICAL TESTS: Mann-Whitney U; Kruskal-Wallis; Benjamini-Hochberg; Bonferroni-correction; receiver operating curve (ROC) and area under curve (AUC); DeLong’s test; Random Forest (RF). P value<0.05 was considered statistically significant after multiple comparisons correction.

RESULTS: Compared with LGGs, MSD, and QIV were significantly lower in TP, whereas NG, NGAx, NGRad, RTOP, RTAP, RTPP, and DKI metrics were significantly higher in HGGs (high-grade gliomas) (P ≤ 0.007), as well as in isocitrate-dehydrogenase (IDH)-mutated than IDH-wildtype gliomas (P ≤ 0.039). These trends were reversed for PT (tumor grades, P ≤ 0.011; IDH-mutation status, P ≤ 0.012). ROC analysis showed that, in TP, DKI metrics performed best in TP (AUC 0.83), whereas in PT, RTPP performed best (AUC 0.77) in glioma grading. AK performed best in TP (AUC 0.77), whereas MSD and RTPP performed best in PT (AUC 0.73) in IDH genotyping. Further RF analysis with DKI and MAP demonstrated good performance in grading (AUC 0.91, Accuracy 82%) and IDH genotyping (AUC 0.87, Accuracy 79%).

DATA CONCLUSION: Tumor multiregional MAP features could effectively evaluate gliomas. The performance of MAP may be similar to DKI in TP, while in PT, MAP may outperform DKI.


PMID:38131220 | DOI:10.1002/jmri.29202

By Nevin Manimala

Portfolio Website for Nevin Manimala