Dear Editor:
We appreciate Wang and Chen’s invitation to comment on measurement error in quantifying body-composition changes with MRI (1). The full range of sources of MRI measurement error is too broad to cover here, and so we limit our comment to addressing one human component of measurement error, specifically, the large CVs for readings of the same scan to which Wang and Chen draw attention.
The only certain way to reduce or eliminate analyst variability is with fully automated processing of digital images. Such software is not yet available (to our knowledge) for quantifying the volumes of tissues and organs that we studied where the images are acquired with multi-slice MRI. In the absence of automated digital image processing, analyst variability can be reduced through training and experience and by using a single analyst. Some reduction in variability may be achieved with higher-resolution images, but such improvement would require longer scan times (thereby increasing study participant burden) or stronger magnetic field instruments, both of which add cost to the conduct of a research study. These additional costs need to be considered against an equivalent or superior gain in power that could be achieved by other means (e.g., increasing the number of study cases).
With regard to the specific assertion that significant changes over a 2-y period may be largely explained by the intra- and/or intermeasurement errors, this is highly unlikely if errors are random and unbiased. As we state in the article, our analyst (of many years’ experience) was blind to whether subjects were in the treatment or control condition; hence, a systematic bias is improbable. We agree that a large CV makes a measurement technique unsuitable for clinical applications; however, because the CI of a sample mean is smaller than the CI of an individual measurement, an adequate sample size allows research studies to detect sample mean changes that are orders of magnitude smaller than the errors of individual measurements.
Acknowledgments
Neither of the authors declared a conflict of interest.
REFERENCE
- 1.Gallagher D, Kelley DE, Thornton J, Boxt L, Pi-Sunyer FX, Lipkin E, Nyenwe E, Janumala I, Heshka S; MRI Ancillary Study Group of the Look AHEAD Research Group. Changes in skeletal muscle and organ size after a weight-loss intervention in overweight and obese type 2 diabetic patients. Am J Clin Nutr 2017;105:78–84. [DOI] [PMC free article] [PubMed] [Google Scholar]