Maximilian Fuetterer1, Julia Busch1, Constantin von Deuster1,2, Christian Binter1, Nikola Cesarovic3, Miriam Lipiski3, Christian Torben Stoeck1,2, and Sebastian Kozerke1,2
1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland, 2Division of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom, 3Division of Surgical Research, University Hospital Zurich, Zurich, Switzerland
Synopsis
A velocity-selective
excitation scheme with bipolar slice-select gradients for hyperpolarized
cardiac perfusion imaging is presented. Using the approach, an excitation ratio
of >5 of myocardial signal to left-ventricular blood pool signal can be
achieved based on differences in blood and tissue velocities. Thereby increased
myocardial signal and reduced left-ventricular signal spilling is obtained.
Dynamic perfusion images acquired with hyperpolarized [13C]urea in pigs show
higher SNR and less signal leakage in the myocardium relative to a conventional
excitation approach. Introduction
Metabolically inert
hyperpolarized [
13C]-labeled substrates such as [13C]urea or HP001 have been used to investigate tumor, renal and
cardiac perfusion (1–3). Due to its nontoxicity, [13C]urea is a promising compound for cardiac perfusion imaging in
patients with impaired renal function or other contra-indications for
Gadolinium. Achieving sufficiently high spatial and spectral resolution in
large animals and humans, however, is challenging due to the limited [
13C]urea polarization and the relatively short relaxation time in
combination with low heart rates. Excitation of the large blood compartments of
the heart strongly saturates the hyperpolarized signal before it arrives in the
myocardium and causes significant signal leakage into the myocardium. To
address these issues, a velocity-selective binomial excitation scheme is
presented to improve image quality in hyperpolarized cardiac perfusion imaging.
Theory
For a given
velocity $$$v_{enc}$$$
the accumulated phase of moving spins subject to a
bipolar gradient pair can be set to π by adjusting the
duration $$$T$$$ of a single
gradient lobe and its area $$$A$$$ according to:
$$\gamma A T v_{enc} = π$$
When combined
with a binomial 1-1 excitation, this effect can be exploited to tip back the
magnetization of spins moving with $$$v_{enc}$$$ with the second
RF pulse, whilst exciting stationary spins twice. For through-plane motion the
bipolar gradients can be efficiently integrated into the slice selection scheme
as illustrated in Fig-1. This excitation can be combined with any single-shot
readout.
Methods
The
velocity-sensitive excitation scheme was implemented into a dynamic,
cardiac-triggered single-shot EPI sequence on a 3T Philips Ingenia system
equipped with a four-channel 13C transmit/receive coil. Three healthy
female pigs (30-35kg) were anesthetized (isoflorane 2-3%), ventilated (4-5L/min) and
venous catheters for medication and injection of hyperpolarized substances were
introduced. Pharmacological stress was induced by administration of dobutamine
with a target heart rate of 120bpm. All experiments were performed in
accordance with the Swiss Animal Protection Law and Ordinance.
Prior to
in-vivo imaging, the proposed sequence was validated on a 2-compartment water flow
phantom with a constant velocity of 48.5 cm/s and a series of different $$$v_{enc}$$$ encodings from
25cm/s to 80cm/s. Measurements were compared to simulated signal intensities
obtained from Bloch simulations (Fig-2).
In-vivo, 2D
anatomical and flow cine scans were acquired to determine through-plane blood
velocities and appropriate trigger delays (Fig-3). An early diastolic heart
phase was chosen to ensure both high blood velocities and mostly stationary
myocardium. Two $$$v_{enc}$$$ encodings
(100cm/s and 25cm/s) were used to measure left-ventricular (LV) and myocardial
velocities and the timing of the velocity-sensitive excitation pulse for
hyperpolarized perfusion imaging was adjusted accordingly.
For perfusion
imaging, a 6.7M [13C]urea glycerol solution was doped with 18.5mM trityl
radical and polarized in a commercial GE SpinLab Hyperpolarizer. 20ml of 200mM
[13C]urea solution were injected over 4s into the femoral vein after
dissolution and transport in a magnetic carrier device (4). Image acquisition was
started 8 heart beats after the beginning of injection to avoid signal
saturation inside the right ventricle. Sequence parameters were chosen as: FOV
110x110mm2, in-plane resolution 3x3mm2 (reconstructed to
1x1mm2), slice thickness 15mm, partial Fourier factor 0.65,
flip-angle 60° resulting in an effective TE of 10.2/12.5ms (conventional/velocity-selective
excitation). One image per heart beat was acquired for 60s during breath-hold.
Perfusion
curves were extracted for septal and lateral regions of the myocardium and peak
myocardial SNR was calculated.
Results
Sequence
validation in the flow phantom showed good agreement of the excitation pattern with
simulated signal intensities for different $$$v_{enc}$$$ encodings
(Fig-2).
In-vivo, measured
velocities in the early diastolic phase were 38-47 cm/s (LV) and 2.5-7.0 cm/s
(myocardium), resulting in effective relative excitation of 0-20% (LV) and
97.7-99.7% (myocardium). Accordingly, an excitation ratio of myocardium-to-LV
of >5 for a $$$v_{enc}$$$ of 42.5 cm/s was achieved (Fig-3).
Peak SNR in the myocardium with the
velocity-selective vs. the conventional scheme was 47.4±6 (lateral), 41.3±16.3 (septal) vs. 23.9±6.2
(lateral), 13.8±6.8 (septal), respectively. Signal-time curves obtained with
the velocity-selective scheme showed the expected physiological delay between
LV and myocardial bolus arrival while the conventional scheme results in
contaminated myocardial signal (Fig-5).
Discussion
The proposed
velocity-selective tip-back excitation significantly increased myocardial SNR
during bolus passage by preserving magnetization and reducing the effective
excitation in the blood pool, thereby also minimizing signal spillage and
contamination of the myocardial perfusion curves.
Due to the
sharp excitation profile around $$$v_{enc}$$$, deviations from the nominal mean blood velocity
result in residual blood pool signal at the boundary between LV and myocardium
and the RV where velocities are less homogeneous (Fig-4).
Conclusion
The presented
velocity-selective excitation scheme improves image quality and SNR in
hyperpolarized cardiac perfusion imaging.
Acknowledgements
This work was supported by the Molecular
Imaging Network Zurich (MINZ) and UK EPSRC (EP/I018700/1).References
1. Von Morze C, Larson PEZ, Hu S,
Yoshihara H a I, Bok R a., Goga A, Ardenkjaer-Larsen JH, Vigneron DB. Investigating
tumor perfusion and metabolism using multiple hyperpolarized 13C compounds:
HP001, pyruvate and urea. Magn. Reson.
Imaging [Internet] 2012;30:305–311. doi: 10.1016/j.mri.2011.09.026.
2. Durst M, Koellisch U, Gringeri C, Janich M a., Rancan G,
Frank A, Wiesinger F, Menzel MI, Haase A, Schulte RF. Bolus tracking for improved metabolic
imaging of hyperpolarised compounds. J. Magn. Reson. [Internet] 2014;243:40–46.
doi: 10.1016/j.jmr.2014.02.011.
3. Lau AZ, Miller JJ, Robson MD, Tyler DJ. Cardiac perfusion
imaging using hyperpolarized 13 c urea using flow sensitizing
gradients. Magn. Reson. Med. [Internet] 2015;00:n/a–n/a. doi:
10.1002/mrm.25713.
4. Shang H, Skloss T, von Morze C, Carvajal L, Van Criekinge
M, Milshteyn E, Larson PEZ, Hurd RE, Vigneron DB. Handheld electromagnet
carrier for transfer of hyperpolarized carbon-13 samples. Magn. Reson. Med.
[Internet] 2015;00:n/a–n/a. doi: 10.1002/mrm.25657.