An analysis of the accuracy of magnetic resonance flip angle measurement methods.

Several methods of flip angle mapping for magnetic resonance imaging have been proposed. We evaluated the accuracy of five methods of flip angle measurement in the presence of measurement noise. Our analysis was performed in a closed form by propagation of probability density functions (PDFs). The flip angle mapping methods compared were (1) the phase-sensitive method, (2) the dual-angle method using gradient recalled echoes (GRE), (3) an extended version of the GRE dual-angle method incorporating phase information, (4) the AFI method and (5) an extended version of the AFI method incorporating phase information. Our analysis took into account differences in required imaging time for these methods in the comparison of noise efficiency. PDFs of the flip angle estimate for each method for each value of true flip angle were calculated. These PDFs completely characterize the performance of each method. Mean bias and standard deviation were computed from these PDFs to more simply quantify the relative accuracy of each method over its range of measurable flip angles. We demonstrate that the phase-sensitive method provides the lowest mean bias and standard deviation of flip angle estimate of the five methods evaluated over a wide range of flip angles.