Temperature dependence, accuracy, and repeatability of T-1 and T-2 relaxation times for the ISMRM/NIST system phantom measured using MR fingerprinting

Purpose Before MR fingerprinting (MRF) can be adopted clinically, the derived quantitative values must be proven accurate and repeatable over a range of T-1 and T-2 values and temperatures. Correct assessment of accuracy and precision as well as comparison between measurements can only be performed when temperature is either controlled or corrected for. The purpose of this study was to investigate the temperature dependence of T-1 and T-2 MRF values and evaluate the accuracy and repeatability of temperature-corrected relaxation values derived from a B-1-corrected MRF-fast imaging with steady-state precession implementation using 2 different dictionary sizes. Methods The International Society of MR in Medicine/National Institute of Standards and Technology phantom was scanned using an MRF sequence of 2 different lengths, a variable flip angle T-1, and a multi-echo spin echo T-2 at 14 temperatures ranging from 15 degrees C to 28 degrees C and investigated with a linear regression model. Temperature-corrected accuracy was evaluated by correlating T-1 and T-2 times from each MRF dictionary with reference values. Repeatability was assessed using the coefficient of variation, with measurements taken over 30 separate sessions. Results There was a statistically significant fit of the model for MRF-derived T-1 and T-2 and temperature (p < 0.05) for all the spheres with a T-1 > 500 ms. Both MRF methods showed a strong linear correlation with reference values for T-1 (R-2 = 0.996) and T-2 (R-2 = 0.982). MRF repeatability for T-1 values was <= 1.4% and for T-2 values was <= 3.4%. Conclusion MRF demonstrated relaxation times with a temperature dependence similar to that of conventional mapping methods. Temperature-corrected T-1 and T-2 values from both dictionaries showed adequate accuracy and excellent repeatability in this phantom study.