A 31P-MR inversion transfer (IT) method with a short adiabatic inversion pulse is proposed and its test–retest reliability was evaluated for two spectral fitting strategies.Methods:
Assessment in a test–retest design (3 Tesla, vastus muscles, 12 healthy volunteers, 14 inversion times, 22 ms asymmetric adiabatic inversion pulse, adiabatic excitation); spectral fitting in Fitting Tool for Interrelated Arrays of Datasets (FitAID) and Java Magnetic Resonance User Interface (jMRUI); least squares solution of the Bloch-McConnell-Solomon matrix formalism including all 14 measured time-points with equal weighting.Results:
The cohort averages of k[PCr→γ-ATP] (phosphocreatine, PCr; adenosine triphosphate, ATP) are 0.246 ± 0.050s−1 versus 0.254 ± 0.050s−1, and k[Pi→γ-ATP] 0.086 ± 0.033s−1 versus 0.066 ± 0.034s−1 (average ± standard deviation, jMRUI versus FitAID). Coefficients of variation of the differences between test and retest are lowest (9.5%) for k[PCr→γ-ATP] fitted in FitAID, larger (15.2%) for the fit in jMRUI, and considerably larger for k[Pi→γ-ATP] fitted in FitAID (43.4%) or jMRUI (47.9%). The beginning of the IT effect can be observed with magnetizations above 92% for noninverted lines while inversion of the ATP resonances is better than −72%.Conclusion:
The performance of the asymmetric adiabatic pulse allows an accurate observation of IT effects even in the early phase; the least squares fit of the Bloch-McConnell-Solomon matrix formalism is robust; and the type of spectral fitting can influence the results significantly.