The diagnostic value of high-frequency power-based diffusion-weighted imaging in prediction of neuroepithelial tumour grading

Abstract

Objectives

To retrospectively evaluate the diagnostic value of high-frequency power (HFP) compared with the minimum apparent diffusion coefficient (MinADC) in the prediction of neuroepithelial tumour grading.

Methods

Diffusion-weighted imaging (DWI) data were acquired on 115 patients by a 3.0-T MRI system, which included b0 images and b1000 images over the whole brain in each patient. The HFP values and MinADC values were calculated by an in-house script written on the MATLAB platform.

Results

There was a significant difference among each group excluding grade I (G1) vs. grade II (G2) (P = 0.309) for HFP and among each group for MinADC. ROC analysis showed a higher discriminative accuracy between low-grade glioma (LGG) and high-grade glioma (HGG) for HFP with area under the curve (AUC) value 1 compared with that for MinADC with AUC 0.83 ± 0.04 and also demonstrated a higher discriminative ability among the G1-grade IV (G4) group for HFP compared with that for MinADC except G1 vs. G2.

Conclusions

HFP could provide a simple and effective optimal tool for the prediction of neuroepithelial tumour grading based on diffusion-weighted images in routine clinical practice.

Key Points

• HFP shows positive correlation with neuroepithelial tumour grading.

• HFP presents a good diagnostic efficacy for LGG and HGG.

• HFP is helpful in the selection of brain tumour boundary.