Gopal Varma1, Patricia Coutinho de Souza1, Leo Tsai1, Rupal Bhatt2, and Aaron Grant1
1Radiology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States, 2Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, United States
Synopsis
Balanced steady-state
free-precession (bSSFP) offers high sensitivity and good temporal resolution,
and makes efficient use of hyperpolarized magnetization. Several strategies for spectroscopically
selective imaging with bSSFP have been proposed [1-5]. Here we investigate the use of simple
binomial excitation pulses to selectively null the signals from either pyruvate
or lactate, the two dominant metabolites in tumors, thereby obtaining images
that are dominated by either lactate or pyruvate, respectively. The method is robust to off-resonance
effects, and can be used to augment existing spectroscopic bSSFP techniques.Purpose
To investigate the use of
binomial pulses in balanced steady-state free-precession (bSSFP) to aid in
acquisition of spectroscopically selective images of hyperpolarized pyruvate
and its metabolites.
Introduction
bSSFP is an attractive technique
for hyperpolarized imaging because it provides good sensitivity and makes
efficient use of the magnetization. Strategies
for spectroscopically selective bSSFP imaging include techniques that acquire multiple
echoes per TR [1,2], selectively excite a single metabolite [3,4], or make use
of a variable RF phase advance to separate signals from different metabolites [5]. These
techniques generally have a high sensitivity to off-resonance effects that
requires careful shimming to reduce so-called ‘bright band’ bSSFP artifacts. Here we investigate the use of a simple
binomial pulse to selectively null the signal from either pyruvate or lactate
in alternate frames of a dynamic acquisition. In applications where the alanine
and pyruvate hydrate signals are small, the resulting images provide an
approximation of spectrally selective images of pyruvate and lactate.
Methods:
A bSSFP sequence was modified to
include two successive slice-selective RF excitation pulses separated by a
delay τ as shown in
Fig. 1. The pulse sequence was
implemented for dynamic imaging where the relative phase between the two RF
pulses was alternated between 0 and 180 degrees in successive images. This
binomial pulse modulates the signal of a metabolite at chemical shift Δ by an overall factor of either cos(πΔτ) or sin(πΔτ) for a phase difference of 0 or 180 degrees, respectively. For imaging of pyruvate and its metabolites
at 9.4T, τ was set to
0.404ms and the center frequency of the scanner was placed at the chemical
shift of pyruvate. With these
parameters, the signals from pyruvate and lactate (approximately 1.24kHz
off-resonance) are nulled in alternating frames. TR was set to 2.20ms, a value that minimizes
the likelihood of bSSFP ‘bright band’ artifacts for pyruvate and all of its
metabolites. To ensure smooth entry
into the steady state, the RF pulse amplitude was modulated by a Fermi function
during each frame.
In
vivo imaging was performed using
a 9.4T horizontal bore scanner (Biospec 94/20, Bruker, Billerica MA) with IACUC
approval. A mouse bearing a subcutaneous A498 tumor was anesthetized and
situated for imaging as described previously [5] with a 28mm 13C
coil situated around the tumor.
Ultrasound gel was applied over the tumor to reduce the susceptibility
mismatch at the air/tissue interface.
Hyperpolarized pyruvate solution was prepared using a DNP hyperpolarizer
(Hypersense, Oxford Instruments, Oxfordshire UK). T2 weighted anatomical images were used to
select a 5mm slice for subsequent imaging.
Local shimming was performed in the tumor and surrounding vessels to
obtain approximately 100Hz proton linewidth.
To confirm the scanner center frequency setting, a small 50 microliter
bolus of 100mM pyruvate solution was administered by tail vein during
acquisition of spectra from the imaging slice.
Subsequently, a 150 microliter bolus was administered, and bSSFP images
were acquired with 1.5mm in-plane resolution, (net) tip angle 15 degrees, and
1.1s per frame. Slice-selective spectra
were analyzed using AMARES in jMRUI [6]. bSSFP images were reconstructed using standard
Fourier methods and sorted into even and odd frames.
Results:
Fig. 2 shows the bSSFP images overlaid on the
proton reference images at several time points.
Images on the left have relative RF phase 0, while those on the right
have relative phase 180, values
are chosen to null lactate and pyruvate respectively. Fig. 3 shows time courses from region of
interest placed over the tumor in the bSSFP images, together data from the
slice-selective spectra.
Discussion:
The images on the left and right in Fig.
2 are expected to be dominated by pyruvate and lactate, respectively. The pyruvate-dominated
images on the left show the expected features, including large intravascular signal (arrow in top frame) followed
by rapid arrival in the tumor. Likewise,
the lactate-dominated images on the right show low intravascular signal, late
buildup in the tumor, and long signal persistence. Both sets of images include contributions
alanine and pyruvate hydrate, which are expected to be small. Previous studies in A498 tumors have shown
that signals from these metabolites are approximately an order of magnitude
smaller than those from pyruvate and lactate [5].
Conclusions:
Binomial
pulses provide a simple method for nulling the signal from a chosen metabolite
in bSSFP. This method is relatively
robust against off-resonance errors, and can be used to obtain approximate
spectrally selective images of pyruvate and lactate in applications where other
metabolites are small. This technique
can be readily combined with existing methods to improve their performance.
Acknowledgements
This work was supported in
part by the NIH through grants R21 EB014471 and R01 CA169470.References
[1]
Leupold et al MAGMA 22:251-6 (2009) [2]
Perman et al MRI 28:458-465. [3] Von
Morze et al MRI 31:163-170 (2013)
[4] Månsson et al MRM 68:1894-9 (2012) [5] Varma et al Proc ISMRM 2013 [6] Naressi et al. Magn Reson Mater Phys 12:141-152
(2001)