Kylian Haliot1,2,3, Andreea Hertanu4, Olivier Girard4, Lucas Soustelle4, Guillaume Duhamel4, Julie Magat1,2,3, and Bruno Quesson1,2,3
1IHU Liryc, Fondation Bordeaux Université, Pessac, France, 2U1045 CRCTB, Université de Bordeaux, Bordeaux, France, 3INSERM, CRCTB, U1045, Bordeaux, France, 4Aix-Marseille Univ, CNRS, CRMBM, Marseille, France
Synopsis
The
Purkinje network (PN) of the heart plays a major role in cardiac electrical
diseases such as in sudden cardiac death. Imaging of this fine network is of
central interest to better characterize the links between electrical disorders
and structural modifications of the heart. PN is composed of fibers containing
cardiomyocytes surrounded by a collagen sheath (PF). In this study, we implemented
inhomogeneous magnetization transfer (ihMT) sequence at 9.4T and evaluate this
technique on ex vivo samples of pig
heart fixed into formalin. We show that ihMT is a suitable candidate to
increase the contrast between PF and cardiac muscle.
Introduction
The Purkinje conduction network (composed of
Purkinje Fibers (PF)) plays a crucial role in normal cardiac function but it
has also been implicated in arrhythmogenesis and sudden cardiac death1.
PF have the particularity of being composed of cardiomyocytes surrounded by a
collagen sheath2,3. Thus, magnetization transfer (MT) techniques are
good candidates to increase contrast between cardiac tissue and PF4.
Beside conventional MT, inhomogeneous MT (ihMT) was developed to explore a
different source of contrasts related to dipolar relaxation times T1D
of tissues. Recently Varma et al.5 demonstrated successful applications of ihMT
on ex vivo tissues with short T1D (1 ms) components outside
the brain, such as in the rat heart. Here, we hypothesize that ihMT can provide
images with improved contrast between cardiac muscle and PF based on their
different T1D values. Thus, a rapid sequence incorporating ihMT
module was implemented at 9.4T and evaluated on an ex vivo cardiac sample from pig.Materials & Methods
Sample
A pig’s
heart (~40 kg) was surgically collected after a sternal thoracotomy followed by
pentobarbital euthanasia in accordance with institutional animal care. The
heart was rapidly rinsed with saline solution. A sample (3.5×4×5.5 cm3)
was selected and then cut in the longitudinal direction of the heart in the right
ventricle including the moderator band (Fig.1), which is a well-known
conductive fiber. The sample was fixed and agitated in formalin at 10% for 48
hours.
Prior to MRI acquisitions,
the sample was removed from formalin and immersed into a perfluoropolyether (3M™
Fluorinert™ FC-770) with a temperature probe (SA Instruments, Stony Brook, NY) positioned
in contact with the sample. Hot water was circulated in a tubing system to warm
the sample and maintain its temperature at 37°C during MRI acquisition.
Acquisition
MRI
acquisitions were performed at 9.4T/30 cm (Bruker BioSpin MRI, Ettlingen
Germany) using a Tx/Rx volume antenna (87mm inner diameter). Images were
acquired with an in-plane resolution of (250 μm)2 and a slice thickness of 2 mm. 2D
ihMT-RARE sensitivity-boosted7 sequences were implemented (Fig.2) with
TR/TE/flip angle/RARE factor/Matrix= 56 ms/6 ms/90°/56/160×100. Frequency alternated
pulses (ALT) and four configurations of frequency offsets Δf (7/10/14/20 kHz) from the water peak were tested.
For
each frequency configuration, the M0 image was
accumulated 4 times
and each of the MT+, MT+-, MT- and MT-+
images were produced after 60 accumulations. IhMT images with two different T1D filtering strengths8
obtained by tuning the repetition time of saturation pulses to 1.1-ms and 3-ms were evaluated. The saturation
period (total duration t=1600 ms) was composed of 12 bursts of Np=12
pulses repeated every BTR=130ms, and deposited a root-mean squared saturation
power B1rms=9 μT.
Post-processing
The ihMT datasets were composed of 5 different images: M0
image (without any MT preparation), MT+ image (with a MT preparation
pulse at +Δf), MT-
image at -Δf and MT+- and MT-+ with dual-frequency MT
preparations. The images were post-processed
with a homemade program using Matlab (MATLAB 9.5, The Mathworks, Inc). A binary
mask was computed based on M0 image to remove background. The ihMT ratio
maps were calculated as $$$\small ihMTR = \left[\left(MT^{+}+MT^{-}\right)-\left(MT^{+-}+MT^{-+}\right)\right]/M_{0}\times100$$$ and conventional MT ratio maps were derived from the same
dataset as $$$\small MTR=\left(1-MT^{+}/M_{0}\right)\times100$$$ .
Two regions of interest were
selected in the cardiac muscle (ROImu) and in the moderator band
(ROIfib) on both ihMTR and MTR maps to plot the respective ratio as
a function of the frequency offset. Finally, the relative contrast RC defined as
$$$\small ROI_{fib}/ROI_{mu}$$$ was calculated.Results
Cardiac sample exhibits different
ihMTR contrast for different frequencies (see ihMTR maps on Fig.3A). The ihMTR
graph for fiber and muscle (Fig.3A) shows a maximum signal for frequencies
above 15 kHz. With relatively weak filtering of the short T1D signal
(Fig.3B black curve), the ihMTR is 8% for fiber and 11.8% for the tissue with a
maximized RC of 0.68. For stronger T1D filtering (Fig.3B red curve),
ihMTR is 3.5% for fiber and 4.5% for the tissue and the RC is 0.78. The MTR graph
(Fig.3C) shows an extremum for frequencies below 7 kHz. The MT relative
contrast between the tissue and the fiber diminishes to 0.85 with increasing
frequency offset at 20 kHz for the 1.1-ms filtering and 0.95 for the 3-ms
filtering. For
frequencies above 15 kHz, we observe an ihMT relative contrast between the
collagen of the PF and the myocardium up to 40% higher compared to the MT
relative contrast.Conclusion
IhMT technique was successfully
implemented and tested at 9.4T on heart samples from pig containing conductive
fibers. The observed contrast demonstrates that cardiac muscle and PF have
different T1D. Our results show that ihMT provides better contrast
between PF and myocardium than conventional MTAcknowledgements
This
work received
financial support from the French National Investments for the
Future Programs:
ANR-10-IAHU-04 (IHU Liryc) and from the Nouvelle-Aquitaine council.References
[1] Samanta R, Pouliopoulos J et al. Heart Rhythm 2016;13(1). [2] Eliska O, Cas Lek Cesk, 2006;145(4). [3] Morita T, Shimada T et al. Arch Histol Cytol, 1991;54(5). [4] Magat J, Fouillet A et al. ISMRM 2016 [5] Varma G, De Souza PC et al. ISMRM 2018 [6] Varma G, Duhamel G,
de Bazelaire C et al. Magn Reson Med. 2015;73(2). [7] Mchinda S, Varma G, Prevost VH et
al. Magn Reson Med. 2018;79(5). [8]
Prevost VH, Girard O et al. NMR Biomed.
2017;30(6).