Objective To validate the reproducibility and consistency of 3T CSE-MRI-based quantification of marrow proton density fat fraction(PDFF)by different MR producers using single-voxel proton MRS as a reference.Methods Thirty-eight subjects underwent lumbar spine MRS(PDFFMRS),United Imaging CSE-MRI(PDFFFACT),and Siemens CSE-MRI(PDFFDixon).The precision and reproducibility of quantitative PDFF by CSE-MRI were evaluated by root mean square error(RMSE)and root mean square coefficient of variation(RMS-CV),respectively.The intra-group correlation coefficient(ICC)was used to assess the agreements of inter-observers and intra-observers for measuring PDFF.Linear regression and Bland-Altman analysis were used to evaluate the reproducibility among the PDFFMRS,PDFFFACT and PDFFDixon.Results Bone marrow PDFFMRS,PDFFFACT and PDFFDixon values were 58.2％(38.7％,67.0％),56.7％(37.2％,66.6％)and 56.3％(36.8％,65.4％),respectively.The RMSE of PDFFFACT and PDFFDixont was 1.48％and 1.57％.RMS-CV was 2.37％and 2.25％.PDFF measurements for intra-and inter-observers had substantial agreement,all ICC>0.99.Marrow PDFF by MRS,FACT and Dixon VIBE sequence with linear regression analysis were as follows:R2=0.985～0.989,all P<0.001.92.1％-97.4％data points,calculated using the Bland-Altman method,were within the limits of agreement.Conclusions The reproducibility from UNI and Siemens for quantifying spinal marrow PDFF is good and their accuracy is equivalent to MRS.
维普
万方数据