Purpose

Presurgical grading is particularly important for selecting the best therapeutic strategy for mningioma patients.1 The purpose of this study is to investigate the value of histogram analysis of diffusion kurtosis imaging (DKI) maps in the differentiation of grades and histological subtypes of meningiomas.

Methods

A total of 172 patients with histopathologically proven meningiomas underwent preoperative magnetic resonance imaging (MRI) and were classified into low-grade and high-grade groups. The imaging protocol included pre-contrast, including axial fluid-attenuated inversion recovery (FLAIR) T1-weighted images, axial T2-weighted FSE images, axial FLAIR T2-weighted images, DKI, and subsequent contrast-enhanced axial/sagittal/coronal FLAIR T1W images. The DKI dataset used a spin-echo diffusion-weighted echo-planar imaging sequence(TR=6,000 ms, TE=94 ms, FOV = 24 cm, matrix = 128 × 128, number of sections = 48, sections thickness = 3 mm, spacing = 0 mm, number of b0 = 3, b values = 1,000 and 2,000 s/mm, gradient directions=30 for each, NEX = 1, acquisition time = 6 minutes 24 seconds). DKI dataset was first checked to ensure no significant image distortion, and the diffusion images were corrected for eddy-current distortion and simple head motion using the FMRIB Software Library (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSL). Then DKI data were processed using the Diffusional Kurtosis Estimator (version 2.5.1, www.nitrc.org/projects/dke).2 Maps for the MD and FA, and MK based on DKI were evaluated in this study. Tumor segmentation was conducted by using the ITK-SNAP program (version 4.6.1, University of Pennsylvania, www. itksnap.org). Based on conventional MR images (pre- and post-contrast) and DK sequence (b=0 s/mm2) images, regions of interest (ROIs) were manually drawn on the solid region of every tumor section on MD maps. Cystic change, necrosis, calcification, and hemorrhage were carefully avoided. Then the whole tumor ROIs on MD maps were transferred to FA and MK maps to get the same regions of the tumor using in-house software written in Python 3.6.1 (https://www.python.org). Finally, DKI measures on a voxel-by-voxel basis for the solid components of the whole tumor were calculated to generate the corresponding histogram. The following metrics of each histogram were obtained: 1) mean, median, maximum, minimum; 2) percentiles（10, 25, 75, 90）; 3) kurtosis, skewness, and variance. ROIs were randomly drawn by two independent neuroradiologists (with 8 and 11 years of experience) without prior knowledge of the histopathological diagnosis. The results of the experienced radiologist were further analyzed. Comparisons of different grades and subtypes were made by Mann-Whitney U test，Kruskal-Wallis tests, ROC curves analysis, and multiple logistic regression.

Results

Significantly higher maximum, Skewness, and variance of MD, mean, median, maximum, variance, 10th, 25th, 75th, and 90th percentile of MK were found in high-grade than low-grade meningiomas (all P<0.05). DKI histogram parameters differentiated 7 of 10 pairs of subtype pairs (atypical versus meningothelial/ fibrous/transitional/angiomatous meningiomas; angiomatous versus fibrous/transitional meningioma; fibrous versus meningothelial meningiomas). The 90th percentile of MK yielded the highest AUC of 0.870 and was the only independent indicator for grading meningiomas.

Discussion and Conclusion

Histogram analysis can depict the intensity distribution of a volume of interest, and provide more quantitative information on tumor features, such as uniformity, heterogeneity, and symmetry.3 DKI is an advanced diffusion modality that can reflect different histopathologic characteristics at the cellular level. The histogram analysis of DKI is valuable in the differentiation between high-grade and low-grade meningiomas. Additionally, DKI histogram parameters help to determinate the subtypes of meningiomas. Moreover, the 90th percentile of MK was the most promising DKI histogram metrics for predicting the grade of meningiomas.