MP2RAGE for deep gray matter measurement of the brain: A comparative study with MPRAGE.

Abstract

To compare magnetization-prepared two rapid acquisition gradient echoes (MP2RAGE) imaging with conventional MPRAGE imaging for deep gray matter (GM) segmentation, reproducibility, contrast ratio (CR) and contrast-to-noise ratio (CNR), and to evaluate reproducibility of T1 maps derived from MP2RAGE. MP2RAGE and MPRAGE imaging were conducted twice for each of 20 volunteers on a 3 Tesla MRI scanner. Images were normalized and segmented using SPM12 with the DARTEL algorithm. Reproducibility of segmentation was evaluated using coefficients of variation (COVs) of deep GM probability maps between first and second scans, which was compared between MP2RAGE and MPRAGE. Differences in deep GM probability were compared voxel-wise. CR and CNR analyses were conducted using regions of interest. COVs of T1 maps were also evaluated. Comparison of GM probability maps demonstrated that putamen, caudate nucleus and thalamus were segmented significantly larger in MP2RAGE than in MPRAGE, and MP2RAGE was inferior only at some areas of globus pallidus and lateral thalamus (P < 0.05; false discovery rate, FDR). CRs of deep GM structures were significantly better in MP2RAGE (P < 0.0001). COVs of deep GM probability maps were significantly higher at large areas of the deep GM in MPRAGE (P < 0.05, FDR). COVs ranged from 0.50 to 3.31% in MP2RAGE and from 0.62 to 4.12% in MPRAGE. COVs of the T1 map were around 2%. MP2RAGE yields greater reproducibility and better tissue contrast than MPRAGE in deep GM. T1 maps derived from MP2RAGE were highly reliable. MP2RAGE is useful for measurement and analysis of deep GM.