Reproducibility of Metabolite Concentration Evaluated by Intraclass Correlation Coefficient Using Clinical MR Apparatus

OBJECTIVE: The purpose of this study was to evaluate the reproducibility of in vivo quantitative proton MR spectroscopy (MRS) by repeated-measure analysis of variance and intraclass correlation coefficient (ICC) including assessment of both inter- and intrasubject variation. METHODS: The concentration of metabolites was quantified using long-TR and short-TE stimulated-echo acquisition mode sequences with a quadrature head coil. All preparation procedures for MRS measurement except volume of interest determination were done by an automatic preparation scanning system equivalent to a clinical setting. Analyses were conducted by an LCModel running based on our original basis set by two different reference methods. RESULTS: The ICC showed a good correlation (r = 0.61-0.99) in any metabolite and almost the same value between the two different reference methods, but the concentration of myoinositol had a large intersubject variation. CONCLUSION: Our results suggest that the reproducibility of quantified metabolite concentration is acceptable in the clinical setting regardless of the reference method, although quantified values may have different intersubject variations.