Meeting Banner
Abstract #4268

Proton MR Spectroscopic Imaging (MRSI) Metabolite Ratios in Human Brain: Calculating Repeatability When Measurement Error Is Proportional to the Mean

David John Manton1, Gary Paul Liney1, Roberto Garcia-Alvarez2, Lindsay W. Turnbull1

1YCR Centre for MR Investigations, Hull, East Yorkshire, UK; 2GE Healthcare, Madrid, Spain

NAA:choline peak area ratios were measured in the brains of eight healthy volunteers at 3T using PRESS MR spectroscopic imaging with the acquisitions carried out twice during the same examinations in order to allow repeatability to be quantified. Measurement error was found to be proportional to the mean which meant that simple repeatability limits (e.g. mean +/- 1.96 x standard deviation) would be inappropriate given that precision varied with mean. Measurement error was not proportional to the mean after the use of the square root of the inverse data transformation and this allowed appropriate (non-constant) repeatability limits to be calculated.