ausblenden:
Schlagwörter:
-
Zusammenfassung:
Introduction
In quantitative MRI measurements, one challenge is to measure prepared longitudinal magnetization fast and precisely, but
without influences of relaxation effects during the readout. For example in Chemical Exchange Saturation Transfer (CEST)
MRI a fast acquisition sequence, a spiral-centric-reordered 3D snapshot, is used [1]. However, the prepared magnetization
changes towards the steady state over the readout leading to T1 Look-Locker-induced image blurring affecting image
intensity and contrast. To reduce this, we optimized a sequence with variable flip angles to achieve an exact estimation of the
prepared magnetization state.
Methods
To create a contrast image similar to an amide CEST contrast the difference of two T1 prepared measurements (TI = 0.5 s and
4 s) with an inversion flip angle of 180° were used. The readout sequence is a spiral-centric-reordered 3D snapshot [1]
(matrix: 64x64x4, spiral elongation=0.5, GRAPPA 2, FA=10°). The target for the optimization process is the difference of two
ideal sequences, where each k-space line is prepared individually. Varying the flip angle should improve the measurement of
the prepared magnetization. Signal simulations and optimization were performed using the Phase Distribution Graph (PDG)
algorithm [2] implemented in the MRzero framework [3], using brain data acquired from the BrainWeb [4] database.
Results
The final training stage of the optimization process is shown in figure 1. (b) to (d) are the differences between the two T1
prepared sequences for the original flip angle choice, the optimized flip angle pattern, shown in (e), and a fully relaxed
acquisition. (a) shows the minimization of the loss. By varying the flip angles over the repetition, the RMS error of the
magnitude images between the original sequence with 10° excitation pulse and the fully relaxed sequence improves from
0.0564 to 0.0272, when using variable flip angles as excitation. The T1LL blurring is clearly reduced and the prepared
magnetization state retrieved.
Conclusion
As has been shown previously, variable flip angles can reduce the signal alterations due to the readout. Here, we
demonstrated the removal of T1LL-blurring by a flip angle optimization for a 3D GRE snapshot sequence as used for CEST
measurements.
