A spatial mapping of the transmitted radiofrequency (RF) field is necessary in a number of scenarios, especially at high field strengths, where the nominal flip-angle (FA) may be significantly different than the applied angle. For cardiac MRI, having B₁ maps that are cardiac phase-resolved may be beneficial in various applications, such as parallel transmission [1], characterizing effects of implantable objects [2] and myocardial quantification at various cardiac phases [3, 4]. Actual flip angle imaging (AFI) has been proposed [5] as an efficient 3D B₁ mapping technique, which utilizes two interleaved acquisitions with the same nominal FA but different repetition times. Its 2D extensions have also been studied [6]. In this work, we propose a retrospectively ECG-gated sequence based on the AFI technique for 2D cardiac phase-resolved B₁ mapping.

Sequence: An AFI sequence consisting of two interleaved FLASH acquisitions with different repetition times (TR₁, TR₂=nTR₁), was implemented as a cardiac cine sequence, enabling image acquisition during steady-state. Retrospective ECG gating was used to calculate images for various cardiac phases. To achieve the rapid TRs required for cardiac imaging and to have sufficient spoiling for AFI quantification, FLASH dephasing gradients were applied at 80 mT/m maximum amplitude and the maximum slew rate. For the kth cardiac phase, the ratio between the two interleave images (I₁^k, I₂^k), r_k = I₁^k/I₂^k is used to calculate the FA for that phase using the formula [5], α_k = arccos[(r_k·n - 1)/(n - r_k)].

Imaging: All imaging was performed on a Siemens 3T Prisma scanner. Typical sequence parameters were 340×220mm² matrix=144/108, TE/TR1/TR2 =1.5ms/3.9ms/19.9ms, slice thickness= 14mm, duration = 18-20 heart-beats.

Phantom Imaging: A homogenous water/NaCl phantom was imaged with the proposed sequence using a symmetric sinc pulse with bandwidth-time-product (BWTP)=2, and nominal FA between 8 and 80 degrees. Since 2D AFI estimates the total FA experienced by the slice, the slice-profile was simulated offline to calculate the slice-corrected expected FA as reference value. The effect of improved slice-profiles on the effective FA was studied using symmetric sinc pulses with BWTP between 1 and 16.

In-vivo Imaging: Imaging was performed on 2 healthy adult subjects, using a symmetric sinc pulse with BWTP=4. A mid-ventricular short-axis slice was acquired in a breath-hold using the proposed sequence. B₁ homogeneity within the myocardium was studied throughout the cardiac cycle. Furthermore, in-vivo B₁ maps were acquired with varying nominal FA, and compared to the expected FA from the specific slice-profile simulations. To avoid partial voluming effects from the blood pool, myocardium was manually segmented using the high acquisition resolution. Subsequently, the myocardium signal was smoothed using a 3×3 moving average filter prior to fitting.

The effect of the slice-profiles on the nominal FA is depicted in Fig 1. The imperfect slice-profiles lead to a lower effective FA. These are well estimated by the proposed sequence, with a maximum deviation <13% with the simulations. Fig. 2 shows the effect of different RF pulses, where the pulses with increased BWTP lead to visually-improved slice-profiles, which in turn increases the effective FA for the slice. An in-vivo example is depicted in Fig 3 for the proposed acquisition with nominal FA=35°, and slice-corrected expected FA=22°. The estimated average B₁ values were 18°, with standard deviation 4.5° throughout the myocardium. 10% (1.9°) B₁ variation was observed across the cardiac cycle. Fig. 4 shows another in-vivo experiment for different nominal FA, showing example B₁ maps from the end-diastolic phase. There is good agreement between the estimated FA using the proposed sequence and the slice-corrected expected FA.Unlike other B₁ mapping methods, 2D AFI allows estimation of the actual gross total FA seen by the slice, including the slice-profile, rather than the B₁ scaling factor. This allows the in-vivo evaluation of different pulses and resulting slice-profiles. However, for highly non-rectangular slice-profiles, signal averaging across the slice in AFI steady-state may deviate from that in the target application, potentially leading to a mismatch in FA estimation. Due to in-flow effects, blood signal can hardly be driven to steady state in 2D cardiac applications. However, assuming a sufficiently low amount of blood regurgitation and a sufficiently long nTR, the blood-signal can be assumed to be almost fully-relaxed, allowing for B₁ quantification, although this was beyond the scope of this study. Instead to avoid the impact of in-flow on the myocardial signal, B₁ maps were acquired with relatively high in-plane resolution and the myocardium was manually segmented before further processing. The proposed sequence allows for cardiac phase-resolved B₁ mapping, with applications in phase-specific shimming or quantification.