Multiplexed sensitivity encoding diffusion-weighted imaging in the assessment of meningiomas: improving the image...

Meningiomas often occur near the skull, where tumor imaging is prone to distortion and susceptibility artifacts, resulting in an unsatisfactory image quality of conventional single-shot echo-planar diffusion-weighted imaging (ssEPI-DWI). Our aim in this study was to evaluate the image quality of multiplexed sensitivity encoding diffusion-weighted imaging (MUSE-DWI) in meningiomas and the accuracy of the apparent diffusion coefficient (ADC) histogram parameters for tumor grading and proliferative activity evaluation based on MUSE-DWI and ssEPI-DWI.

A total of 73 consecutive patients with pathologically confirmed meningiomas were included in this prospective study. Three neuroradiologists independently assessed the image quality using a 5-point Likert scale, whereas two other neuroradiologists quantitatively calculated the signal-to-noise ratio of lesion (SNR_lesion), signal-to-noise ratio of the normal white matter (SNR_normal), and lesion to brain contrast-to-noise ratio (CNR). Inter-rater agreement was assessed using the intra-class correlation coefficient (ICC). Image quality scores, SNR values, and CNR values of the two diffusion techniques were compared using the Wilcoxon signed-rank test. Histogram metrics of ADC values were extracted from the whole tumor. Receiver operating characteristic (ROC) curve and logistic regression analyses were performed to evaluate the diagnostic performance of single histogram parameter and combined models for tumor grading. Spearman rank correlation was applied to assess the correlation between histogram parameters and the Ki-67 labelling index (LI).

Sharpness, distortion, artifact, lesion conspicuity, and overall image quality were significantly better in MUSE-DWI than those in ssEPI-DWI, with good agreement among the three raters (P<0.05; ICC: 0.83-0.90), whereas there was no significant difference in diagnostic confidence between the two DWI sequences (P>0.05; ICC: 0.80-0.82). The values of SNR_lesion, SNR_normal, and CNR were all significantly higher in MUSE-DWI than they were in ssEPI-DWI with good agreement between the two raters (P<0.05; ICC: 0.82-0.92). In both diffusion techniques, high-grade meningiomas showed significantly lower ADC 10th percentiles (C10) and ADC median values (P=0.005-0.039), and higher ADC interquartile range (IQR) values (P=0.005-0.025), compared to low-grade meningiomas. After false discovery rate (FDR) correction, only the ADC C10 (q=0.025) and ADC IQR (q=0.025) values derived from MUSE-DWI remained significantly different between high- and low-grade meningiomas. The best diagnostic accuracy was obtained by combining the ADC C10 and ADC IQR from MUSE-DWI [area under the curve (AUC) =0.774]. After nested cross-validation, the model still showed an AUC of 0.769. For MUSE-DWI, ADC IQR (P=0.007) and variance (P=0.040) were significantly positively correlated with Ki-67 LI. For ssEPI-DWI, only ADC IQR (P=0.047) was significantly positively correlated with Ki-67 LI. However, after FDR correction, no ADC histogram parameter remained significantly associated with Ki-67 LI (all q values >0.05).

Compared with ssEPI-DWI, MUSE-DWI improves the diffusion image quality of meningiomas, enhancing the accuracy of meningioma grading and proliferative activity evaluation.