Get access

Addressing phase errors in fat-water imaging using a mixed magnitude/complex fitting method

Authors

  • D. Hernando,

    Corresponding author
    1. Department of Radiology, University of Wisconsin, Madison, Wisconsin, USA
    • Department of Radiology, University of Wisconsin, Wisconsin Institutes for Medical Research, 1111 Highland Ave, Madison, WI 53705
    Search for more papers by this author
  • C. D. G. Hines,

    1. Department of Radiology, University of Wisconsin, Madison, Wisconsin, USA
    Search for more papers by this author
  • H. Yu,

    1. Global Applied Science Laboratory, GE Healthcare, Menlo Park, California, USA
    Search for more papers by this author
  • S.B. Reeder

    1. Department of Radiology, University of Wisconsin, Madison, Wisconsin, USA
    2. Department of Medical Physics, University of Wisconsin, Madison, Wisconsin, USA
    3. Department of Biomedical Engineering, University of Wisconsin, Madison, Wisconsin, USA
    4. Department of Medicine, University of Wisconsin, Madison, Wisconsin, USA
    Search for more papers by this author

Abstract

Accurate, noninvasive measurements of liver fat content are needed for the early diagnosis and quantitative staging of nonalcoholic fatty liver disease. Chemical shift-based fat quantification methods acquire images at multiple echo times using a multiecho spoiled gradient echo sequence, and provide fat fraction measurements through postprocessing. However, phase errors, such as those caused by eddy currents, can adversely affect fat quantification. These phase errors are typically most significant at the first echo of the echo train, and introduce bias in complex-based fat quantification techniques. These errors can be overcome using a magnitude-based technique (where the phase of all echoes is discarded), but at the cost of significantly degraded signal-to-noise ratio, particularly for certain choices of echo time combinations. In this work, we develop a reconstruction method that overcomes these phase errors without the signal-to-noise ratio penalty incurred by magnitude fitting. This method discards the phase of the first echo (which is often corrupted) while maintaining the phase of the remaining echoes (where phase is unaltered). We test the proposed method on 104 patient liver datasets (from 52 patients, each scanned twice), where the fat fraction measurements are compared to coregistered spectroscopy measurements. We demonstrate that mixed fitting is able to provide accurate fat fraction measurements with high signal-to-noise ratio and low bias over a wide choice of echo combinations. Magn Reson Med, 2012. © 2011 Wiley Periodicals, Inc.

Ancillary