Reproducibility of rapid short echo time CSI at 3 tesla for clinical applications.

Purpose: To validate the reproducibility of a chemical shift imaging (CSI) acquisition protocol with parallel imaging, using automated repositioning software. Materials and Methods: Ten volunteers were imaged three times on two different 3 Tesla (T) MRI scanners, receiving anatomical imaging and two identical CSI measurements, using automated repositioning software for consistent repositioning of the CSI grid. Offcenter parameters of the CSI plane were analyzed. Coefficients of variation (CoV), Cramer-Rao lower bounds (CRLB), intraclass correlation coefficients (ICC), and coefficients of repeatability (CoR) for immediate repetition and between scanners were calculated for N-acetylaspartate, total choline, creatine, myo-inositol (Myo) and glutamine+glutamate (Glx). Proportions of variance reflecting the effect of voxel location, volunteer, repetition, time instance and scanner were calculated from an analysis of variance analysis. Results: The offcenter vector and angulations of the CSI grid differed less than 1 mm and 2 degrees between all measurements. The mean CoV and CRLB were less than 30% for all metabolites, except for Myo. The variance due to voxel location in the volume of interest and the error represent the largest contributions in variability. The ICC is the lowest for Myo and Glx. CoR for immediate repetition and between scanners display values between 22 and 83%. Conclusion: We propose a CSI protocol with acceptable reproducibility, applicable in clinical routine. J. Magn. Reson. Imaging 2013;37:445456. (C) 2012 Wiley Periodicals, Inc.