Adiabatic excitation for 31P spectroscopy in the human heart at 7T
Ladislav Valkovič1,2, William T Clarke1, Benoit Schaller1, Lucian A B Purvis1, Stefan Neubauer1, Ivan Frollo2, Matthew D Robson1, and Christopher T Rodgers1

1Oxford Centre for Clinical Magnetic Resonance Research, University of Oxford, Oxford, United Kingdom, 2Department of Imaging Methods, Institute of Measurement Science, Slovak Academy of Sciences, Bratislava, Slovakia


31P-MRS is of particular interest in cardiovascular medicine, as the PCr/ATP ratio can serve as a predictor of mortality. However, due to inherently low signal-to-noise ratio (SNR), cardiac 31P-MRS is not yet practical in the clinic. To increase SNR, the use of 7T and dedicated receive arrays has been proposed. However, the peak B1+ was inadequate for the use of B1 insensitive pulses, thus far. In this study, we demonstrate the feasibility of homogeneous adiabatic excitation for cardiac 31P-MRS using a novel quadrature 31P transceiver at 7T. This constitutes an important step towards absolute quantification of cardiac metabolites at 7T.


Phosphorus MR spectroscopy (31P-MRS) is a valuable technique for non-invasive measurement of high-energy metabolites, e.g. adenosine triphosphate (ATP) and phosphocreatine (PCr) that allows assessment of tissue energy metabolism in vivo. 31P-MRS is of particular interest in cardiovascular medicine, as the PCr/ATP ratio in the heart changes in most major disease states and can serve as a predictor of mortality1,2. However, due to inherently low signal-to-noise ratio (SNR), cardiac 31P-MRS is not yet practical in the clinic. To increase SNR, the use of ultra-high fields (e.g., 7T) has been proposed3. Further improvement in SNR, as well as in heart coverage was recently demonstrated using a dedicated receive array4. However, the peak B1+ of the array coil was deemed inadequate for the use of B1 insensitive excitation pulses and homogeneous 90° excitation of the whole heart would constitute a significant improvement and provide an important step towards absolute quantification.

Therefore, our aim was to test the feasibility of adiabatic excitation for cardiac 31P-MRS using a novel quadrature 31P transceiver at 7T.

Materials and Methods

All measurements were performed on a 7T MR system (Siemens) equipped with a dual-tuned Tx/Rx quadrature RF-coil, comprising a purpose-built quadrature 31P coil (two 15cm loops, with overlap decoupling) and a single 1H loop (10cm in diameter). The coil design and safety tests are described in detail in a different abstract.

Coil performance was tested using a two-compartment phantom, which consists of an 18L chamber filled with 73mM NaCl(aq) and a 2cm cube, filled with KH2PO4(aq), fixed at a 10cm depth, simulating a position within the heart. Fully-relaxed non-localized FIDs were acquired using a 10ms long adiabatic half passage (AHP) excitation pulse, optimized for low B1+ and SAR requirements using Bloch simulations (Figure 1). The transmitter voltage was increased from 50V to 450V in 25V steps, with additional repetitions between 350V and 450V.

Five healthy volunteers (3M/2F) were recruited, in compliance with local regulations, and were examined lying supine with the quadrature coil positioned over their heart. To test the adiabatic excitation in vivo, two cardiac 3D 31P UTE-CSI5 scans with AHP excitation were acquired in each volunteer. First, at a transmit voltage giving 100% predicted SAR (i.e., ~445V) and then at 50V less (to test whether we had exceeded the adiabatic threshold). Because of SAR restrictions, relatively long TR (4000ms) was mandatory. Therefore, to keep the acquisition time below 32min, and to keep skeletal muscle contamination to minimum, the spatial resolution in sagittal direction was sacrificed, yielding a matrix size of 8x16x8 (RLxAPxHF) with a FOV of 240x240x200mm3.

The fitted phantom spectral amplitudes were compared to identify a range of voltages giving 90° excitation at the depth of the heart. The in vivo data were fitted using a Matlab implementation of AMARES6 and corrected for T1 relaxation3. Twelve voxels with SNR>7 were selected from the heart in each volunteer for PCr/ATP quantification and comparison of the two acquisitions by a Bland-Altman analysis of agreement7 to demonstrate that the adiabatic onset has been achieved in vivo.

Results and Discussion

Figure 2 depicts the signal amplitudes of KH2PO4(aq) acquired in the phantom at different transmit voltages, normalized to the amplitude at the maximum voltage. The amplitude increases steadily and then plateaus, indicating that an adiabatic 90° excitation is achieved from 350V upwards at a 10cm depth from the coil.

In vivo, a sample spectrum and PCr/ATP map (Figure 3) demonstrate low contamination of the heart voxels by the signal from the chest muscles and the consistent ratio across the heart is further evidence for a homogeneous excitation flip angle throughout the sensitive volume of the coil. The PCr/ATP ratios calculated for every volunteer from each UTE-CSI acquisition are given in Table 1. All individualized ratios, from both measurements, as well as the mean value of 1.89±0.36 are in good agreement with literature values on healthy volunteers1-3. Comparing the two UTE-CSI acquisitions for all 60 selected voxels by a Bland-Altman analysis of agreement (Figure 4), a mean bias of -0.03 could be identified. This bias represents higher repeatability than the previously reported inter-scan repeatability at 3T (-0.07)8, which supports our phantom data and indicates that we have exceeded the adiabatic onset in the heart in vivo.


This proof-of-concept study demonstrates that a 15cm quadrature pair can provide sufficient B1+ across the heart for adiabatic excitation in vivo. This work paves the way for the first human studies with absolute quantification of cardiac metabolites at 7T.


Funded by a Sir Henry Dale Fellowship from the Royal Society and the Wellcome Trust [098436/Z/12/Z], and by the Slovak grant agency VEGA [2/0013/14].


1. Bottomley P, et al. Radiology 1987;165(3):703-7

2. Neubauer S, et al. Circulation 1992;86(6):1810-8

3. Rodgers CT, et al. Magn Reson Med 2014;72(2):304-15

4. Rodgers CT, et al. Proc. ISMRM 2014;22:2896

5. Robson MD, et al. Magn Reson Med 2005;53(2):267-74

6. Purvis LA, et al. Proc. ISMRM 2014;22:2885

7. Bland JM & Altman DG. The Statistician 1983;32(3):307-17

8. Tyler DJ, et al. NMR Biomedicine 2009;22(4):405-13


Figure 1: Amplitude and phase of the 10ms long tanh/tan AHP pulse used here (a). The shape of the pulse is given by lambda=3 and beta=atan(88). A Bloch-equation simulation of the dependence of Flip angle on the B1+ field (b).

Figure 2: Test for adiabaticity in a phantom at 10cm depth. Normalized signal amplitudes acquired at different voltages are overlaid with a Bloch-equation simulation of the pulse shown in Figure 1 (dotted line). A 95% confidence interval delimits the plateau region where the amplitude is stable, indicating adiabatic 90° excitation.

Figure 3: Typical acquired cardiac 31P-MR spectrum (a) and PCr/ATP map (b). Position of the coils is marked by the signals of fiducials and the approximate position of the heart and chest wall is delineated. Note the consistent PCr/ATP ratio across the heart throughout the sensitive volume of the coil.

Table 1: Mean PCr/ATP ratios of each volunteer (throughout the selected voxels of the heart) measured by UTE-CSI at the two used voltages (MaxV and 50V less).

Figure 4: Bland-Altman plot of the in vivo comparison between the scan at maximum voltage and at 50V less for the PCr/ATP ratio. The mean difference between the two scans is -0.03, showing extremely low bias. Note that both, PCr and ATP were corrected for saturation assuming homogeneous 90° excitation.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)