Integrated MR-PET imaging is a versatile tool for the non-invasive characterization of cardiovascular disease. In this work, we developed an MS-CAIPIRINHA-based imaging technique to extend the anatomical coverage of the MRI perfusion assessment to six slices per RR. As a proof of principle, the described approach was combined with simultaneous 18F-FDG viability imaging and Late Gadolinium Enhancement (LGE) imaging investigating improvements in anatomical coverage for a relevant patient cohort.
An ECG-gated TurboFLASH prototype similar to [2] was implemented
on a whole-body MR-PET system (Biograph mMR, Siemens Healthcare, Erlangen,
Germany). The pulse sequence featured a dual-band pulse to excite two slices simultaneously,
while the RF phase of the second slice was toggled between 0° and 180°. Within each RR-interval, three saturation
recovery (SR)-prepared acquisition blocks consisting of 52 radial projections
were acquired. A model-based algorithm, which is both enforcing the sensitivity
profiles of the utilized coil array and the sparsity in the spatial wavelet-domain
was then applied to reconstruct the undersampled datasets:
$$ \min_{I_{sl}} \; \biggl\Vert \biggl(\sum_{sl=1}^2 \Phi_{sl} \, E_{sl} \ I_{sl} \biggr)-y \, \biggr\Vert_2^2 + \lambda \sum_{sl=1}^2 \Vert \Psi I_{sl} \Vert_1 $$
$$$y$$$ represents the measured k-t-space multi-coil data
for one of the three dual-slice acquisition.
$$$I_{sl}$$$ corresponds to the temporal image series for
one of the two slices sl.
$$$E_{sl}$$$ is the encoding operator, which incorporates an
inverse Fourier transform, the re-gridding back to the initial radial projections,
and the superposition of the coil sensitivities to obtain multi-coil data. The
coil sensitivities for each slice were
determined by a fully sampled pre-scan under breathhold prior to contrast agent
injection.
$$$\Phi_{sl}$$$ performs the applied CAIPIRINHA phase
modulation, and
$$$\Psi$$$ transforms each frame of the image series into
the spatial wavelet domain.
$$$\lambda$$$ balances the weighting of the sparsity and the
data consistency term and was chosen empirically. A projection onto convex sets
(POCS) implementation in MATLAB (MathWorks, Natick, MA, USA) was used to
perform the optimization on an Intel Core i7-3820 CPU @ 3.60 GHz.
Typical imaging parameters were: Slice distance
between adjacent slices = 2 mm, slice distance between simultaneously acquired
slices = 30 mm, TE = 1.03 ms, TR = 1.33 ms, flip angle = 10°, slice thickness =
8 mm, in-plane resolution = 2.04 × 2.04 mm², 52 radial projections per image
pair, image matrix = 160 × 160, Rtotal ≈ 5.
The described MS-CAIPIRINHA approach was tested
against a conventional 3-slice SR-FLASH perfusion sequence with Cartesian
k-space ordering and similar sequence parameters as described above [3]. Conventional
and MS-CAIPIRINHA perfusion images were acquired separately for 60-90 RR
intervals during the same patient scan using two identical bolus injections 0.05 mM/kg Gd-DTPA.
LGE images were acquired across
the whole LV myocardium after a cumulative dose of 0.2 mM/kg Gd-DTPA and 15 min
equilibration time.
MR imaging was performed in five patients with
collateralized coronary total occlusions (CTO) in combination with a
simultaneous 18F-FDG viability scan after metabolic preparation by
insulin-clamping. List-mode PET data were acquired for 45 min starting 60 min
after injection of 18F-FDG.