Reproducibility of 31P cardiac magnetic resonance spectroscopy at 3 T

The purpose of this work was to take advantage of the new clinical field strength of 3 T to implement and optimize a chemical shift imaging (CSI) acquisition protocol to produce spectra of high quality with high specificity to the myocardium within a clinically feasible scan time. Further, an analysis method was implemented dependent purely on anatomical location of spectra, and as such free from any potential user bias caused by inference from spectral information. Twenty healthy male subjects were scanned on two separate occasions using the optimized CSI protocol at 3 T. Data were analyzed for intra- and inter-subject variability, as well as intra- and inter-observer variability. The average phosphocreatine (PCr)/adenosine triphosphate (ATP) value for scan 1 was 2.07 +/- 0.38 and for scan 2 was 2.14 +/- 0.46, showing no significant difference between scans. Intro-subject variability was 0.43 +/- 0.35 (percentage difference 20%) and the inter-subject coefficient of variation was 18%. The intro-observer variability, assessed as the absolute difference between analyses of the data by a single observer, was 0.14 +/- 0.24 with no significant difference between analyses. The inter-observer variability showed no significant differences between the PCr/ATP value measured by four different observers as demonstrated by an intro-class correlation coefficient of 0.763. The increased signal available at 3 T has improved spatial resolution and thereby increased myocardial specificity without any significant decrease in reproducibility over previous studies at 1.5 T. We present an acquisition protocol that routinely provides high quality spectra and a robust analysis method that is free from potential user bias. Copyright 2008 John Wiley & Sons, Ltd.