Ultra short echo time (UTE) MRI of mouse myocardial infarction

Full text

Biomedical NMR
Department of Biomedical Engineering
Eindhoven University of Technology
Ultra short echo time (UTE) MRI
of mouse myocardial infarction
Jules Laurent Nelissen
MSc thesis
Eindhoven, March 23, 2012
Committee:
ir. Bastiaan van Nierop
dr. ir. Gustav Strijkers
dr. Marc Kouwenhoven
dr. Noortje Bax
prof. dr. Klaas Nicolay

2
Abstract
Fibrosis is an important hallmark of various cardiac pathologies. The excessive accumulation of
extracellular matrix proteins, such as collagen, plays a pivotal role in the progression of cardiac
pathology towards heart failure. Contrast enhanced MRI can be used to detect indirectly fibrosis
associated with ischemic and non-ischemic heart disease. However, the accumulation of gadolinium
based contrast agents is rather nonspecific. Ultra short echo time (TE) MRI has the potential to
detect protons with very high transverse relaxation rates (low T2 relaxation time), including those
associated with fibrotic tissue. Therefore, ultra short TE (UTE) MRI may be exploited as a means to
provide a more specific detection of myocardial fibrosis. Therefore, the aim of this study was to
demonstrate in vivo and ex vivo visualization of fibrotic tissue in infarcted mouse myocardium by UTE
MRI.
The UTE MRI sequence was available on a 9.4T small animal MRI scanner. Sequence optimization
was performed by measuring the actual k-space trajectories and gradient timing using phantoms.
Myocardial infarction was induced in C57BL/6 mice (♂, age 11 weeks, n=20) by means of permanent
occlusion of the left descending coronary artery. In vivo UTE MRI was performed 6 and 14 days after
surgery. Immediately after the last in vivo MRI session mice were euthanized, their hearts excised
and used for overnight ex vivo UTE measurements. Finally, hearts were embedded in paraffin, cut in
5-µm-thick slices and stained with Picrosirius Red to confirm the excessive presence of collagen in
the infarct area.
Both in vivo and ex vivo UTE showed good general image quality. The ΔUTE images, obtained by
subtracting a long-TE from a short-TE UTE image, showed clear contrast between remote and infarct
myocardium. Histology confirmed that the infarct area contained significant amounts of collagen and
a strong correlation was observed between the contrast enhanced infarct volume determined in the
UTE difference images and the red stained collagen. Infarct volume was determined in the Picrosirius
Red histology images. These findings showed that the enhanced area on the UTE difference images
indeed corresponded to areas with large amounts of collagen. In conclusion, this study showed that
in vivo and ex vivo contrast of mouse myocardial infarction can be obtained using UTE MRI.

3
Contents
Abstract ............................................................................................................................................. 2
1 Introduction ............................................................................................................................. 5
1.1 Myocardial fibrosis .............................................................................................................. 5
1.2 Imaging of myocardial fibrosis ............................................................................................ 8
2 Ultra short echo time MRI ..................................................................................................... 10
2.1 Short history and UTE sequence ....................................................................................... 10
2.2 Technical considerations of UTE........................................................................................ 12
3 Materials and Methods ......................................................................................................... 13
3.1 Sequence optimalization ................................................................................................... 13
3.1.1 Phantom test ............................................................................................................. 13
3.1.2 Gradient delay ........................................................................................................... 14
3.1.3 Trajectory measurements ......................................................................................... 14
3.1.4 Acceleration of the in-vivo data acquisition .............................................................. 15
3.2 Mouse model ..................................................................................................................... 16
3.3 Animal study design .......................................................................................................... 17
3.4 Ex vivo MRI ........................................................................................................................ 18
3.5 In vivo MRI ......................................................................................................................... 18
3.6 Histology ............................................................................................................................ 20
4 Results ................................................................................................................................... 22
4.1 Sequence optimalization ................................................................................................... 22
4.1.1 Phantom test ............................................................................................................. 22
4.1.2 Gradient delay ........................................................................................................... 24
4.1.3 Trajectory .................................................................................................................. 24
4.1.4 Acceleration of the in-vivo data acquisition .............................................................. 25
4.2 Ex vivo MRI ........................................................................................................................ 27
4.3 In vivo MRI ......................................................................................................................... 31
4.4 Histology ............................................................................................................................ 34
5 Discussion .............................................................................................................................. 36
5.1 Sequence optimalization ................................................................................................... 36
5.1.1 Phantom test ............................................................................................................. 36
5.1.2 Gradient delay ........................................................................................................... 36
5.1.3 Trajectory .................................................................................................................. 36
5.1.4 Acceleration of the in-vivo data acquisition .............................................................. 37
5.2 Ex vivo MRI ........................................................................................................................ 37
5.3 In vivo MRI ......................................................................................................................... 38
5.4 Histology ............................................................................................................................ 39

4
6 Conclusion ............................................................................................................................. 40
7 Dankwoord ............................................................................................................................ 41
8 Literature ............................................................................................................................... 42
9 Appendix ................................................................................................................................ 45

5
1 Introduction
Cardiovascular disease is one of the leading causes of death in the western world. Each year, 17.3
million people die from cardiovascular diseases, accounting for 31 % of all deaths globally (Figure 1)1-
2. Cardiovascular disease is caused by disorders of the heart and blood vessels, and includes diseases
such as coronary heart disease, cerebrovascular disease, peripheral artery disease, rheumatic heart
disease, congenital heart disease, hypertension and heart failure3. Heart failure is a condition of the
heart in which it cannot longer provide adequate blood flow and/or pressure to meet the body’s
demands4.
Figure 1 Distribution of major causes of death in 2011 calculated by the World Health Organization. Cardiovascular diseases
are responsible for 31% of all deaths globally5.
Multiple causes, including myocardial infarction, genetical disorders, hypertension and diabetes
can ultimately lead to heart failure, which forms a burden to patients in terms of suffering, reduced
quality of life and poor survival rate; 30 – 40 % of patients die within one year following diagnosis
and 60 – 70 % die within 5 years6-7. Regardless of the precise cause of heart failure several common
processes are identified which play an important role in the failing heart; these include myocardial
hypertrophy, vascular adaptation, loss of metabolic flexibility, and myocardial fibrosis 8-13.
1.1 Myocardial fibrosis
Myocardial fibrosis is characterized by an increase in the amount of various forms of collagen and
other extracellular matrix (ECM) components in the interstitium and perivascular regions of the
myocardium. Fibrosis results from the disruption of the equilibrium between synthesis and
degradation of ECM proteins, in particular collagen type I and III, which results in an excessive
increase in the amount of collagens in the myocardium (Figure 2) 14-16. Myocardial fibrosis is found
throughout the heart in the left ventricle (LV), septum and also in the right ventricle (RV). Fibrosis is
increasingly cited as one of the most important hallmarks of heart failure17-21.

6
Figure 2 Schematic overview of the pathway leading to cardiac fibrosis. Fibroblasts are stimulated by the growth factors
angiotensin II (ANGII), transforming growth factor β1 (TGF- β1) and endothelin-1 (ET-1) to transform to myofibroblasts,
which produce a more fibrotic ECM. Together with an imbalance in the activity of matrix metalloproteinases (MMPs) and
their inhibitors (TIMPs) this is one of the prime factors leading to systolic and diastolic dysfunction of the heart 16.
Changes in levels of hormones, cytokines, chemokines and growth factors in the myocardium
induce the transition from fibroblast into myofibroblast and are key to the fibrotic process.
Myofibroblast produce a different type of ECM as compared to fibroblasts and are regulated by
growth factors of the renin-angiotensin-aldosterone system, endothelin-1, and transforming growth
factor β1. The altered balance of matrix metalloproteinases (MMPs) and their inhibitors, tissue
inhibitors of metalloproteinases, might also play a crucial role in the changes of the ECM. MMPs
normally facilitate the degradation of the ECM and an imbalance could lead to too much collagen
degradation, resulting in dilatation of the LV or conversely to insufficient collagen degradation
resulting in a stiffer myocardium15-16, 22-23.

7
Figure 3 Different types of myocardial fibrosis. Approximately 75% of the healthy myocardium consists of cardiomyocytes
and the remainder of the tissue is occupied by blood vessels, macrophages, mast cells, myofibroblasts, and collagen type I
and III. During various disease processes the balance between collagen production and degradation can result in an increase
of myocardial fibrosis inbetween the cardiomyocytes. Three different types are known; reactive interstitial fibrosis (A),
replacement/scarring fibrosis (B) and infiltrative interstitial fibrosis (C) 24.
Three subtypes of myocardial fibrosis can occur: reactive interstitial fibrosis,
replacement/scarring fibrosis, and infiltrative interstitial fibrosis (Figure 3). Reactive interstitial
fibrosis (Figure 3A) is characterized by a diffuse distribution of collagen in the interstitium and
formation of perivascular fibrosis. This type is often caused by pressure overload of the LV due to
hypertension or aortic valve stenosis, but can also occur due to ageing or metabolic disorders such as
diabetes. Replacement or scarring fibrosis (Figure 3 B) is characterized by more extensive formation
of fibrosis and is most often caused by myocardial infarction or transient ischemia. Infiltrative
interstitial fibrosis (Figure 3 C) can occur due to various genetic disorders such as amyloidosis or
Anderson-Fabry disease 24.

8
The formation of fibrosis leads to changes in composition of the myocardium and has functional
consequences for the heart in terms of increased stiffness, contractile uncoupling, altered perfusion,
and electrical uncoupling. The consequences are compromised function, ischemia and arrhythmias.
Fibrosis thus plays an important role in heart failure and adverse cardiac events14, 25.
1.2 Imaging of myocardial fibrosis
A variety of (imaging) techniques such as echocardiography, Single Photon Emission Computed
Tomography (SPECT), Positron Emission Tomography (PET), collagen biomarkers, endomyocardial
biopsy and Magnetic Resonance Imaging (MRI) have been used to assess the presence, extent, and
turnover of myocardial fibrosis25-26. All these techniques are mainly used for the detection of
replacement fibrosis. Imaging techniques for the assessment of diffuse, reactive interstitial fibrosis
are still not available, but could make a significant contribution to the improvement of the diagnosis
of this type of myocardial fibrosis.
New non-invasive MRI techniques for myocardial fibrosis are increasingly recognized as a
promising approach in diagnosis of heart failure17, 20, 24-29. Late gadolinium enhancement (LGE) MRI is
currently the non invasive gold standard for quantification and detection of replacement myocardial
fibrosis in patients with ischemic heart disease30. LGE MRI is based on a bolus injection of a
paramagnetic contrast agent (CA) such as gadolinium (Gd) diethylene triamide penta aceticacid
(DTPA), which gives a bright signal enhancement. After a bolus injection is given, Gd-DTPA enters in
normal perfused myocardial tissue resulting in hyperenhancement in the first 60 seconds. In the
following 1 - 3 minutes the CA washes out of the healthy myocardium, while the Gd-DTPA fails to
penetrate areas of microvascular obstruction. After 10 – 15 minutes delayed enhancement can be
seen in areas with replacement fibrosis, due to the larger distribution volume and slower washout
kinetics as compared to healthy tissue31-37. This assessment of myocardial fibrosis by LGE MRI works
well if it is present in a (macroscopic) bulky form, but in case of more (microscopic) diffuse fibrosis
the technique is less suitable26, 38.
Figure 4 Key results of a T1 mapping (A) and equilibrium contrast MRI (B) study performed by Iles et al. and Flett et al,
respectively. Both studies show the correlation between the used technique and collagen content in the myocardium. In the
graph of the T1 mapping study (A) the mean post-contrast T1 time, of the myocardium after a bolus injection of a Gd-based
CA is plotted against the myocardial collagen content obtained by biopsy. In the graph of the equilibrium contrast study (B)
the calculated volume of distribution in the myocardium, based on the mean T1 determined by T1 mapping under
continuously infusion of a Gd-based CA during the whole experiment, is plotted against the myocardial collagen volume
fraction obtained from biopsy samples 39-40.

9
Recent studies show that contrast enhanced T1 mapping and equilibrium contrast cardiovascular
MR may be promising techniques to improve the diagnostic accuracy of diffuse reactive interstitial
fibrosis. Key graphs of both studies are shown in Figure 4 24-26, 41-42. In Figure 4 A the correlation
between post-contrast T1 in the myocardium, acquired with a contrast enhanced T1 mapping
technique, and myocardial collagen content obtained by biopsy from Iles et al. study are shown. The
results of the equilibrium contrast MR study from Flett et al., based on a T1 calculation of the
myocardial volume of distribution during continuous infusion of Gd-DTPA, are shown in Figure 4 B,
where the correlation of the calculated volume of distribution in the myocardium and the collagen
volume fraction determined by biopsy are plotted. Both techniques are promising and indeed show a
modest, but significant correlation between post contrast T1 and collagen content. Both techniques
are performed in human studies in which it is relatively simple to obtain T1 measurements in the
heart. This is more challenging in pre-clinical research with mouse models, often used to test new
imaging techniques, because of the higher heart rate compared to humans. Another disadvantage of
both T1 based studies and LGE MRI is the use of a Gd based CA, which does not provide a direct
readout for myocardial fibrosis24. Recently, a study by De Jong et al. provided evidence suggesting
that it is possible to detect myocardial fibrosis based on collagen content with ultra short echo time
(TE) MRI in ex vivo infarcted rat hearts with a myocardial infarction43.
The aim of this study was to obtain a direct readout of myocardial fibrosis based on endogenous
contrast of collagen rich tissue in the infarcted mouse myocardium in vivo using an ultra short TE
(UTE) MRI sequence. First, phantom experiments were performed for sequence optimalization.
Second, proof-of-principle experiments were performed in ex vivo mouse hearts with a permanent
occlusion of the left descending coronary artery. Finally, the technique is tested in vivo in mouse with
infarcted myocardium and healthy littermates as control.

10
2 Ultra short echo time MRI
Healthy tissue contains little or no short T2 components. However, during disease the amount of
short T2 components is increased due to iron deposition, hemorrhage, calcification, amyloidosis or
fibrosis44. Myocardial fibrosis, which is in increased amount present in a diseased heart, results in an
increase of short T2 components. This is caused by an excessive increase of the ECM molecule
collagen, which is highly ordered and induces strong dipolar interactions between protons in water
and collagen. This results in extremely short tissue T2s ranging from 10-500 µs, depending on the
order of organization of collagen in the tissue 45-46. Signal from these short T2 components cannot be
detected with conventional MR pulse sequences with echo times in the order of milliseconds. This is
the result of competition between the RF pulse tending to rotate the magnetization towards the
transverse plane and relaxation processes tending to decrease the magnetization in the transverse
plane. The decay of the MR signal of these short T2* components is so fast that there is no
magnetization left when readout starts. Several techniques such as magic angle imaging,
magnetization transfer imaging, short echo time imaging, ultra short echo time imaging, but also
techniques such as magnetic sandwich echo and double quantum filter imaging are developed in the
last decade to generate endogenous contrast based on short T2 components. UTE MRI pulse
sequences have TEs in the range of about 0.01 to 0.5 ms and enable the detection of the very short
T2 components present in myocardial infarction46-48. The main reason UTE MRI is used in this project
and none of the other techniques summed up above is that it is more difficult to use these MRI
techniques in the in vivo mouse heart than UTE MRI.
2.1 Short history and UTE sequence
Various 2D and 3D UTE pulse sequences have been developed since Bergin et al. first introduced
an MRI technique in 1991 with unconventional half pulse slice selection and radially-acquired
projections to produce images with very short TE to image lung parenchyma49. Since then UTE was
extensively used for bone, cartilage, fibro-cartilage and the musculoskeletal system. A typical result is
shown in Figure 5 A50.
Figure 5 UTE examples from literature, imaging of cortical bone (A), visualization of calcified plaque (B), direct detection of
collagen in ex vivo rat hearts with myocardial infarction (C). Adapted from Du et al., Chan et al., De Jong et al. 43, 50-51.
Recently, UTE is used for cardiovascular applications such as calcified plaque imaging (Figure 5 B)
and the already mentioned detection of myocardial fibrosis in ex vivo rat hearts with a myocardial
infarction (Figure 5 C)43, 51.

11
Figure 6 Sequence diagram of the 3D UTE sequence available on the 9.4T Bruker small animal MRI scanner.
The pulse sequence diagram of the 3D UTE sequence used in this cardiovascular UTE project to
generate endogenous contrast of myocardial fibrosis based on collagen is shown in Figure 6. It is a 3D
radial sequence with a non slice-selective short RF excitation block pulse of typically 20 µs, followed
by radial sampling of k-space. Important to notice is that the sampling of k-space starts already
during ramp-up of the gradient to minimize signal loss due to relaxation of the short T2 components.
Radial sampling of k-space has the advantage over other k-space filling strategies, that sampling
starts in the center of k-space where the coarse image details are stored and the signal-to-noise ratio
is determined. A short TE of approximately 21 µs is typically used, and an additional gradient delay in
the order of microseconds is used to synchronize the start of the read-out with the actual start of the
gradient upslope. At the end of the readout a spoiler gradient is used to spoil all transversal
magnetization before the next RF excitation. The contrast in this UTE sequence depends on T2*
rather than T2. Reconstruction of the spatial signal from the radial sampled k-space can be done by
projection reconstruction or by regridding the radial spokes onto a Cartesian grid followed by an
inverse Fast Fourier Transformation. In this study the regridding method is used to reconstruct the
3D UTE datasets. To assist the regridding algorithm often the actual measured k-space trajectory is
used. In the 3D UTE sequence the trajectory measurement can be performed in several ways. Pre-
measured in a phantom and stored, pre-measured as preparation module, and the theoretical build-
in trajectory can be used. The trajectory measurement method in the UTE sequence is based on the
measurement of the phase evolution in each spatial encoding direction in response to a gradient.

12
2.2 Technical considerations of UTE
Apart from smart radial filling of k-space, the choice of radiofrequency (RF) pulse shape is crucial
to be able to detect protons with very high transverse relaxation rates. During a RF pulse there is
competition between the pulse tending to rotate the net magnetization vector and relaxation
processes tending to decrease the magnetization in the transverse plane. In case of short T2
components, where T2 is of the same order of magnitude as the duration of the RF pulse or even
shorter, relaxation processes dominate and the magnetization actually tipped in the transverse plane
is less than expected based on the specified flip angle. For tissue with a relatively long T2 the optimal
flip angle is defined by the Ernst angle equation, but this is only valid in case the T2 is long as
compared to the duration of the excitation process44, 52. Simultaneously, the longitudinal
magnetization becomes saturated leading to a reduction of the magnetization in the transverse plane
with a decrease in the longitudinal direction. Therefore, RF pulses should have a short duration and
low flip angle (<20 degree) in UTE sequences to be able to detect components with low T2 relaxation
times45-47, 53-56.
To be able to detect short T2 components, which have a low abundance, such as myocardial
fibrosis, long T2 reduction techniques are required, whereas in tissues with mainly long T2
components no long T2 reduction techniques are needed to generate contrast of T2 components
with the UTE sequence. The first long T2 reduction technique is a long 90 degree RF block pulse
followed by a dephasing gradient. The long components are rotated into the transverse plane and
dephased by the crusher gradients, while leaving the short T2 components mostly unaffected. The
second method is to use long inversion preparation pulses to invert the long T2 components
selectively followed by an inversion time to null them. The third method, used in this study, is the so-
called delta (Δ) or subtraction UTE method. In this method an UTE image with a long TE is subtracted
from a short TE UTE image to produce a difference image with signal only from the short T2
components44-47.
Figure 7 Magnetization decay of short, long, and total magnetization to illustrate the ΔUTE method. The short TE UTE image
is acquired at the right dc (data collection) indicator in this figure and contains information of long and short T2
components. The long TE image acquired at the left dc (data collection) indicator mainly consists of information of the long
T2 components. Subtraction of long and short TE image give a ΔUTE image with only short T2 components. Adapted from
Robson et al.47.
The principle of the ΔUTE method can be described using Figure 7 where the decay of total
magnetization, short T2 components and the long T2 components are shown as a function of time.
This graph illustrates that the magnetization of the short T2 components decays faster than that of
the long T2 components. To detect these short components the first readout (indicated with dc in
Figure 7) is done with ultra short TE (µs) and signals of both short and long components are obtained.
The second acquisition is done with a longer TE (ms). Since the short T2 components have already
decayed, only signal from the long T2 component is recorded. After subtraction of the second from
the first echo a difference-image is produced in which most of the signal comes from the short T2
components47.

13
3 Materials and Methods
All MRI measurements were performed with a 9.4T small animal MRI scanner (Bruker, Ettlingen,
Germany) equipped with a 72-mm-diameter quadrature transmit coil and a 4 element phased-array
receive coil. First, simple tests were done to see if the 3D UTE sequence is able to detect short T2
components. Second, the gradient delay time was optimized and k-space trajectories were measured
to optimize the UTE sequence. Third, proof of principle ex vivo UTE MRI experiments of healthy and
infarcted mouse hearts were performed to test if the optimized 3D UTE sequence can be used to
detect myocardial fibrosis. Fourth, in vivo 3D UTE, Cinematographic (Cine) MRI and LGE MRI were
performed on mice with myocardial infarction and healthy littermates. Finally, histology was done to
correlate infarct volume on ex vivo ΔUTE MR images with infarct volume determined with Picrosirius
Red staining.
3.1 Sequence optimalization
3.1.1 Phantom test
First, 3D UTE MRI measurements were performed using a 50 mL Falcon Tube, see Figure 8 A,
filled with 0.1 mM Prohance (Gadoteridol, Gd-HP-DO3A) in water and alginate gel (Johannes Weithag
KG, Germany) a natural polysaccharide from brown algae, which forms an irreversible hydrocolloid
gel when mixed with water. Alginate forms a highly organized cross-linked structure that was
expected to have a very short T2. Therefore, it was considered a suitable phantom for the UTE
sequence.
Figure 8 Phantoms used for sequence optimization. Short T2 phantom made from alginate and water containing 0.1 mM
Prohance (A). Glass sphere calibration phantom filled with 1g/L CuSO4 used for gradient delay and k-space trajectory
optimalization (B).
First, an experiment was performed using the alginate phantom to compare a conventional 3D
FLASH MR sequence (α = 15o; TR =10 ms; FOV = 3 x 3 x 1.5 cm2; matrix = 192 x 192 x 128) with
TE=2.85 ms to the 3D UTE sequence (FOV = 3 x 3 x 3 cm2; matrix = 128 x 128 x 128; TR = 8 ms;
undersampling = 2) with short TE = 0.02 ms and long TE = 4 ms. Second, measurements were done to
test the effect of undersampling, the gradient delay, and the deviation between expected and actual
k-space trajectory on the image quality of this phantom.

14
3.1.2 Gradient delay
The first tests using the alginate sample made clear that the gradient delay of the 3D UTE
sequence had to be optimized to account for a hardware gradient delay, i.e. a difference in the
intended and the actual start of the upslope of the gradient. The optimal gradient delay was
determined using a glass sphere phantom (Figure 8 B) filled with an aqueous solution of 1 g/l CuSO4
positioned in the isocenter of the scanner, by varying the gradient delay from 1 to 100 µs in steps of
1 µs for the matrix size 128 x 128 x 128 and a field of view (FOV) of 3 x 3 x 3 cm3.
Figure 9 From left to right: correct gradient delay, gradient delay to short, gradient delay to long. Adapted from the Bruker
Paravision manual version 5.1. The correct gradient delay example is the most homogeneous and has the least artefacts.
The optimal gradient delay was determined by selecting the gradient delay resulting in an MR
image with a homogeneous appearance according to the correct gradient delay example, shown in
Figure 9.
3.1.3 Trajectory measurements
Trajectory measurements were done for matrix size 128 x 128 x 128 and FOV (3 x 3 x 3 cm3) in
the previously used CuSO4 containing glass sphere, a glass sphere filled with oil, a small oil cylinder,
and a large oil cylinder. Protocols were stored and the Cartesian coordinates of the different
measured trajectories were compared to the ideal trajectories. This was done by plotting the
different measured trajectories with Matlab (MathWorks, Natick, Massachusetts, USA), and
determine if the trajectories follow a straight smooth line, having little curvature at the origin, with a
slope of approximately one in x, y, and z dimensions.

15
3.1.4 Acceleration of the in-vivo data acquisition
The acquisition of the 3D UTE sequence can be done using ECG triggering, which is necessary for
in vivo application, resulting in an acquisition time of approximately the number of radial spokes
multiplied by the R-R interval. This results in acquisition times of approximately 25.554 x 100 ms ≈ 43
minutes (FOV = 3 x 3 x 3 cm3; matrix 128 x 128 x 128; undersampling = 2) depending on heart and
respiratory rate. To shorten the acquisition time a logic device (TUeDACS, Experiment Automation
Group TU/e, Eindhoven, The Netherlands) was used (Figure 10 C).
Figure 10 A: Logic device (BioNMRTriggerbox) interface: 1) number of trigger pulses; 2) initial delay before 1st trigger pulse;
3) time between 2 trigger pulses; 4) pulse width of trigger pulse; 5) blanking time; 6) On/Off; 7) Reset. B: schematic overview
of the generated output. C: Photograph of the TUeDACS BioNMRTriggerbox.
Upon the detection of a trigger, from the SAII monitoring system (Small Animal Instruments Inc.,
New York, USA), the logic device sends a trigger to the scanner, possibly delayed by an initial delay
time, which can be followed by n triggers each separated by a repeat time and the possibility to use a
blanking time (Figure 10 A&C). The logic device enabled a reduction of the acquisition of the ECG
triggered 3D UTE sequence by sending multiple triggers to the scanner after detection of a single ECG
R-wave.

16
Figure 11 Schematic overview of scanner setup and integration of BioNMRtriggerbox (LOGIC Device).
The logic device was connected to the SAII small animal monitoring system, control computer,
and the AVANCE III hardware unit of the Bruker MR scanner as depicted in Figure 11 and was tested
with an oscilloscope and SAII before connecting and testing with the scanner AVANCE III hardware
unit and 3D UTE sequence.
3.2 Mouse model
32 C57BL/6 mice (♂, age 11 weeks, 20-30 gram) were included in this study. Food and water
were given ad libitum. Mice were separated in a control group (n=12) and a group that underwent a
permanent occlusion of the coronary artery (n=20) to induce myocardial infarction. Mice were
anesthetized with 4% isoflurane for induction and for maintenance with 2.5% isoflurane in medical
air (0.4 l/min). Next, mice were intubated and the lungs were ventilated with a MidiVent mechanical
ventilator (Harvard Apparatus, Massachusetts, USA). A left anterolateral thoracotomy was performed
by incision of the chest to gain access to the thoratic organs, a tissue retractor was placed to keep
the chest open, the lungs were deflated and the left descending coronary artery of the heart was
permanently occluded by tightly lacing a fine suture (6-0, silk), as is depicted in Figure 12.
Figure 12 Left: Photograph of a permanent occlusion of the left descending coronary artery in the mouse heart (LV = left
ventricle, O = place of occlude, A = atrium). Right: schematic overview of the mouse heart indicating the position at which
the coronary artery is permanently occluded. Figure adapted from Michael et al. and Patten et al. 57-58.

17
Immediately after ligation of the left descending coronary artery, a part of apical side of the LV
changed color from bright red to pale red indicating successful ligation. The lung was reinflated, skin
layers closed, and the animals were allowed to recover overnight at 30o Celsius. Temgesic (opioid
analgesia, 0.1 mg.kg) was given pre operative for analgesia, as well at the end of the day of surgery
and one day later.
3.3 Animal study design
Figure 13 Timeline of in vivo, ex vivo, and histology experiments.
The first experiment (Figure 13 I) was designed to test if it was possible to detect myocardial
fibrosis with the 3D UTE, and for further optimalization of the sequence. 19 C57BL/6 were included,
12 with a permanent occlusion of a coronary artery and 7 healthy littermates who served as controls.
Six days after the permanent occlusion of the coronary artery MRI measurements were performed.
The protocol consisted of Cine MRI to determine myocardial function and LGE for infarct size
determination. Immediately after the last MRI measurements, mice were euthanized by
exsanguination and perfusion of the LV with PBS (pH 7.4) to prevent formation of blood clots.
Immediately afterwards the hearts were excised. Ex vivo 3D UTE MRI measurements were done
overnight and afterwards hearts were gradually frozen at a rate of -1 Co/min in Advanced DMEM
(Gibco) containing 10% fetal bovine serum and 10% Dimethylsulfoxide (DMSO) and stored in the
freezer (-80 Co) for histology.
In the second experiment (Figure 13 II) the 3D UTE sequence was applied in vivo. 8 C57BL/6 mice
with a permanent occlusion and 5 healthy controls were included. In vivo 3D UTE and Cine MRI
measurements were done at days 6 and 14, whereas LGE MRI was only performed on day 6. After
the last in vivo measurements mice were euthanized as described above and their hearts were
scanned ex vivo overnight with the 3D UTE sequence. Hearts were stored in the freezer for histology,
as described above.

18
3.4 Ex vivo MRI
After excision of the hearts, they were flushed and filled with Fomblin (Solvay, Paris, France) a
perfluoropolyether used for susceptibility matching and gently put in a CryoTube (Nunc A/S,
Roskilde, Denmark). The tube was fixed on top of the 4-element phased array RF coil and centered in
the isocenter of the magnet to minimize artifacts, which may occur due to gradient imperfections.
After tuning and matching of the 72-mm-diameter volume coil, shimming and flip angle
optimalization the ex vivo hearts were scanned overnight with the 3D UTE sequence (α = 5o; TR = 8,4
ms; FOV = 128 x 128 x 128; matrix = 3 x 3 x 3 cm3; undersampling = 2; trajectory = premeasured) with
TE = 21, 30, 40, 50, 60, 70, 80, 90µs, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5,
5, 6, 7, 8, 9, 10, 20, 30, 40ms.
The ex vivo 3D UTE datasets were analyzed using Matlab with the following method. First, ΔUTE
3D datasets were obtained by subtracting the various long TE datasets from the short TE (21µs)
dataset. Second, the enhanced volume in the ΔUTE images was quantified and normalized to whole
heart volume. Third, an optimal TE for the long TE was determined by looking at the exponential
decay of both infarcted and remote areas. Fourth, T2* was quantified in an ROI in the infarct and
remote regions by means of a fit of the various UTE images to a mono exponential function,
in which Mxy(t) is the signal at time t, Mxy(0) the signal intensity at time t=0.
3.5 In vivo MRI
Mice were anesthetized with 4% isoflurane in medical air (0.4 l/min) in an induction cage.
Anesthesia (2.5% isoflurane in medical air (0.4 l/min)) was maintained during the whole experiment.
An infusion line filled with Gd-DTPA (0.5 mmol/kg) in 0.9% NaCl was placed in the tail of the mouse.
Mice were transferred to the MR scanner and positioned on top of the phased array coil. Care was
taken to position the heart of the mouse in the middle of the phased array coil. A heat pad was used
to maintain the mouse body temperature at 37oC. The front paws of the mouse were fixed to
electrodes to measure the electrocardiogram (ECG), and respiration was measured with a pressure
balloon placed on the abdomen of the mouse. Temperature and anesthesia levels were well
controlled, monitored to keep a constant heart rate (460 - 550 min-1) and respiration rate (50 - 70
min-1) and registered. Finally, the animal cradle was positioned in the bore of the magnet with the
phased array coil in the isocenter.
Correct positioning was verified using a ruler and running of a tri-pilot MRI survey scan, in which
x, y, and z direction saturation bands in the MR image indicate the magnet isocenter. After global
shimming and RF pulse calibration, Cine MR imaging was performed using a 2D ECG triggered and
respiratory-gated FLASH bright blood sequence (α = 15o; TR/TE = 7/1.8 ms; FOV = 3 x 3 cm2; slice
thickness = 1mm; acquisition matrix = 192 x 96, zero filled to 192 x 192 (long axis (LA)) & 128 x 64,
zero filled to 128 x 128 (short axis (SA)), cardiac frames = 12-15 depending on heart rate; acquisition
time approximately 4 min per slice)59. For each mouse a 2-chamber and 4-chamber LA slice and 3-5
SA slices were acquired, with interslice distance optimized for heart size.
Next, six 3D UTE scans (α = 5o; TR = 6.092 ms; FOV = 3 x 3 x 3 cm3; acquisition matrix = 128 x 128
x 128; undersampling = 2; radial spokes = 25554) were acquired with varying TE (0.021, 0.1, 0.3,
0.714, 1.429, 4 ms). To prevent artifacts from cardiac and respiratory motion the 3D UTE sequence
was ECG triggered and respiratory gated.

19
Upon the detection of an R-wave three k-lines were acquired triggered by 3 pulses from the logic
device with a repeat time of 6.4 ms. The measurements immediately following R-wave detection
resulted in k-lines being measured in a part of the cardiac cycle where cardiac motion is limited. To
correct for timing and gradient errors, calibrated k-space trajectories and a calibrated gradient timing
were used. To further reduce the acquisition time to approximately 20 minutes per dataset the
acquisition matrix was undersampled by a factor 2.
Figure 14 Left: 4 chamber LA cross section of a Cine MR image illustrating the orientation of the saturation slice used to
saturate the blood flowing into the heart Right: Corresponding LA cross section of corresponding 3D dataset. The dark band
in the image at the base of the heart results from the saturation pulse.
To improve contrast between the blood and the myocardium a 3 mm thick saturation slice was
used, planned in SA orientation just above the LV base (Figure 14). Excitation of the saturation slice
was done with a 1.5ms Gauss RF pulse with a flip 90o angle and followed by a spoiler gradient. An off-
line reconstruction was used to prevent long reconstruction delays during scanning.
Finally, LGE measurements were performed to quantify the infarct size. First a baseline scan was
made with a retrospectively gated Cine FLASH sequence with navigator echo (α = 15o; TR/TE =
8/2.7ms; FOV 3x3 cm2; matrix = 256 x 256; slice thickness = 1.5 mm; number of cardiac frames = 15;
acquisition time = 4 min) oriented SA in the infarct region60. Next a Gd-DTPA bolus was injected in the
tail vein of the mouse with an infusion pump (Chemyx Inc., Fusion 100, Stafford, UK, flow rate: 2
ml/min). Immediately after injection a second scan was started to observe the wash-in of the CA. The
scan was repeated until 20 min post injection, as LGE in the infarcted area is expected to last
approximately 15 minutes after bolus injection33-34.
Cardiac function parameters end-diastolic volume (EDV), end-systolic volume (ESV), stroke
volume (SV) and ejection fraction (EF) were quantified from the acquired Cine MRI data using a
commercially available MRI analysis software package (CAAS MRV FARM 2.0, Pie Medical Imaging,
Maastricht, The Netherlands). Cine movies were reconstructed for all mice to confirm akinesia of the
infarct region. For each mouse mean heart and respiratory rate were calculated per scan session.
For the 3D UTE datasets a custom-built rotation and interpolation algorithm was made in Matlab.
The orientation of the datasets was changed to correspond to the acquired Cine MR LA and SA
images. ΔUTE images were obtained by subtraction of the long TE datasets from the short TE dataset
in Matlab, highlighting myocardial tissue with high transversal relaxation rates while signal from
myocardial tissue with low transversal relaxation rates is suppressed.
LGE MR images were all scaled based on the scaling of the baseline scan, made before bolus
injection. LGE images were compared to in vivo ΔUTE images.

20
3.6 Histology
On a selection of all hearts obtained from experiment I and II histology was performed (2 x
control; 1 x PO 6 days; 1 x PO 14 days). For embedding of the hearts, hearts were thawed quickly at
37oC and washed twice with PBS. Afterwards, hearts were fixed in 3.7% formaldehyde for 24 hrs and
further processed for paraffin immunohistochemical investigation. Subsequently, 5-µm-thick serial
frontal sections were mounted on polysine-coated slides (Menzel-Gläser). Serial sections (10 series)
were subjected to standard immunohistochemical procedures61. Briefly, after rehydration of the
slides, collagen was visualized by a Picrosirius Red staining for 20 minutes and washed with demi
water. Finally, all slides were dehydrated and mounted with Entellan (Merck).
Figure 15 Example of heart (healthy) cut in 5µm thick slices from anterior to posterior side.
A Zeiss Axio Observer Z-1 bright field microscope equipped with an Axiocam MrC5 was used to
digitalize the microscope images (Figure 15). To compare the volume of collagen rich areas as
quantified from the 3D Δ UTE MR images with histology, 1.6x/2.5x magnification Mosaix images
(consisting of 12 separate images) of the stained slices were made with an interval of 100µm from
anterior to posterior side of the heart to capture the whole heart. To check for iron content caused
by the formation of blood clots in the infarct area, which could lead to contrast enhancement on the
ΔUTE images, a Prussian Blue staining was performed. The Prussian Blue stained slices were
compared to the Picrosirius Red stained infarct and control slices to determine if there was heme
containing blood accumulation in the infarct region.

21
The ex vivo 3D UTE datasets were compared to histology by study the relation between the
enhanced volume quantified in the ΔUTE images and the collagen rich volume quantified from the
Picrosirius Red histology staining. Both volumes were normalized to total heart volume.
Figure 16 Custom build color detection algorithm based on HSV and RGB color spaces.
Histological images were analyzed with Matlab using a custom-built color detection algorithm
based on the hue-saturation-value (HSV) and red-green-blue (RGB) color spaces (Figure 16) to
quantify whole heart volume and infarct volume. Total heart volume was determined using a HSV
based threshold and the collagen-rich volume was calculated using a RGB threshold for red in the
apical region of the heart where normally the infarct, and thus formation of collagen, was expected.
The total heart volume and enhanced volume in the ex vivo 3D delta UTE images were also
analyzed with Matlab. Total heart volume was calculated based on the short TE UTE images
(TE=21µs) by nulling background with a signal intensity threshold excluding the majority of
background signal while leaving the myocardial tissue unaffected. The enhanced volume in the ΔUTE
images was determined by considering the signal enhanced when signal intensity was larger than the
mean signal intensity in the remote area plus 2x standard deviation. This remote area was chosen in
a substantial portion of the heart were normally no infarct was expected.

22
4 Results
4.1 Sequence optimalization
4.1.1 Phantom test
In Figure 17 the results are shown of the phantom containing alginate and water with 0.1 mM
Prohance scanned with a conventional FLASH 3D sequence (A) and the 3D UTE sequence (B, C, D),
note that the k-space trajectory was pre-measured in the phantom in these experiments. In a cross
section of the 3D FLASH dataset (A) the water appears bright where the alginate gives no signal due
to its short transversal relaxation rate. In the short-TE 3D UTE image (B) both alginate and water
could be clearly observed whereas the signal of alginate in the long-TE 3D UTE image (C) was absent.
In the ΔUTE image (D) only signal from the alginate in the sample remained. These phantom
experiments prove that the signal from the long T2 components can be effectively suppressed in the
ΔUTE images.
Figure 17 FLASH and 3D UTE scans from the alginate – 0.1mM Prohance water phantom with axial cross section of the
FLASH 3D with TE = 2.85 ms (A), corresponding slice of 3D UTE with TE = 21µs (B) and TE = 4 ms (C) and ΔUTE (D).
A selection of images with different undersampling factors is depicted in Figure 18 to show the
effect on image quality. Noise, ringing and ghosting artefacts increase with higher undersampling
factors, which becomes apparent in Figure 18 D and E. The image quality of images obtained with
undersampling factors 1 and 2 are rather similar (Figure 18 A, B), whereas some more artefacts are
observed in Figure 18 C with undersampling = 3.
Figure 18 3D UTE axial cross sectional images from the alginate – 0.1 mM Prohance water phantom with increasing
undersampling. Undersampling factors are 1, 2, 3, 5, 10 in A, B, C, D, E, respectively.
In Figure 19 3D UTE MR images are shown from the alginate sample with different gradient
delays (0, 5, 10 and 50 µs). The alginate appears darker for higher gradient delays, and in Figure 19 B,
C and D a white rim is seen which becomes more prominent with higher gradient delay. For the
image obtained with a gradient delay of 50 µs the alginate and water could not be distinguished
(Figure 19 D) and multiple artefacts appear. The images with the shortest gradient delay results
overall in the best image quality.

23
Figure 19 3D UTE MR images of the alginate – 0.1mM Prohance water phantom with increasing gradient delay. Images
obtained with a 0, 5, 10, 50 µs gradient delay are shown in A, B, C, D.
The importance of k-space trajectory correction for reconstructions of the 3D UTE images is
illustrated in Figure 20. Both 3D UTE images from the same alginate sample (Figure 20 A&B) are
markedly different due to the choice of k-space trajectory used for image reconstruction. The left
image (Figure 20 B) contains numerous artefacts and the two compartments (alginate and water) of
the alginate test sample are hard to distinguish, whereas the right image (Figure 20 A) is almost
artifact free, except for a somewhat brighter rim (indicated with an arrow).
Figure 20 3D UTE images (A&B) and corresponding k-space trajectory spokes (C&D) from the alginate phantom measured
with different trajectories.
The trajectories are depicted below the 3D UTE images, in Figure 20 C&D. Whereas the trajectory
in Figure 20 D is meandering radially, the trajectory in Figure 20 C is a smooth straight line. This
corresponds to the poor image quality in Figure 20 B and good image quality Figure 20 A.

24
4.1.2 Gradient delay
A selection of images of a glass sphere filled with an aqueous solution of CuSO4 obtained with
different gradient delays is displayed in Figure 21. When we compare these images with the example
image in the Bruker manual, the image with the correct gradient delay should be ideally
homogeneous and artefact free. By visual evaluation of the various images it was found that the
image with a gradient delay of 1 µs (Figure 21 A) seems to be the most homogeneous with the least
ghosting artefacts. All other images (Figure 21 B,C,D) show a black or white inner rim inside the
sphere-shaped phantom and more ghosting artefacts. For the images with a gradient delay larger
than 10 µs the center of the sphere appears dark.
Figure 21 3D UTE images of the glass sphere phantom filled with 1g/l CuSO4 with gradient delays 1, 5, 10, 20 µs in A, B, C,
and D, respectively.
4.1.3 Trajectory
The results of the k-space trajectory measurements in different phantoms are pictured in Figure
22. The trajectories measured for the glass-sphere phantoms filled with oil and CuSO4 are similar and
radially smooth. Whereas the trajectories measured for the filled cylinders show lots of irregularities.
Also, the trajectories measured in both cylinders traverse in a different direction as compared to the
trajectories measured in the spherical phantoms.
Figure 22 Radial k-space trajectories for a matrix of 128 x 128 x 128 and FOV of 3 x 3 x 3 cm3 in 4 different phantoms. In red
the trajectory measured in the sphere filled with oil, in green the sphere filled with aqueous solution of CuSO4, in blue the
small cylinder filled with oil, and in black the large cylinder filled with oil. Note that the trajectories obtained in both spheres
are almost equal.

25
The small curvature in the beginning of the trajectory (near the origin), which can be clearly seen
in the trajectory of both spheres (red & green lines) in Figure 23, corresponds to the start of data
acquisition during the ramp-up of the read-out gradient. These first sample points of the trajectory
are the most important for image quality, because the low frequency components, containing the
coarse details of the image, are stored in center of k-space and could end up at a wrong position in
case of an ill-measured trajectory.
Figure 23 Curvature in the first sample points of the measured k-space trajectories (same data as in Figure 22),
corresponding to the ramp-up of the readout gradient. In red the oil-filled sphere, in green the glass sphere filled with
aqueous solution of CuSO4, in blue the small oil-filled cylinder, and in black the large oil-filled cylinder.
4.1.4 Acceleration of the in-vivo data acquisition
The output signal of the logic device was validated with an oscilloscope. Figure 24 shows the
input signal obtained from SAII monitoring system (yellow line) and the output signal of the
BioNMRtriggerbox (purple line). Whereas the output signal of the SAII animal monitoring system
returns in approximately 30 ms back towards zero, the generated output of the BioNMRtriggerbox
returns in only 50 ns towards zero. Typically, three triggers are send to the AVANCE III hardware unit
controlling the 9.4 T Bruker MRI small animal scanner for each trigger received from the SAII
monitoring system.
Figure 24 Screenshot of the oscilloscope showing the input (yellow with black rim) and output (purple) of BioNMRtriggerbox.
The output of the SAII animal monitoring system served as input for the BioNMRtriggerbox. Settings of the
BioNMRtriggerbox are: trigger repeat pulse = 3; trigger pulse width = 1 ms; trigger repeat time = 8.5 ms.

26
In Figure 25 two typical mouse ECG signals are shown. The ECG signal which is used for the
monitoring of the mouse and for prospective triggering of MRI sequences, is often distorted by the
MR imaging gradients. An example is shown in Figure 25 B where the gradients used for acquisition
of three k-lines following the detection of one R-wave as described above, resulted in three negative
peaks after the R peak in the ECG. In Figure 25 A an ECG is shown without gradient induced peaks.
Figure 25 ECG signals of the same mouse with (B) and without (A) influence of the gradient set to acquire UTE k-space
spokes. 3 radial k-space spokes are acquired, triggered by 3 pulse of the BioNMRtriggerbox with trigger repeat time 8.5 ms.

27
4.2 Ex vivo MRI
Figure 26 shows representative examples of 3D UTE short TE (21 µs), long TE (4 ms) and ΔUTE
images of two mouse hearts in the long-axis orientation 6 days (top) and 2 days (bottom) after
permanent occlusion of the coronary artery. Although the infarct area 2 days after surgery showed
marked wall thinning, hardly any collagen was present as was later confirmed by histology.
Figure 26 LA cross-section images of a 3D UTE dataset of a mouse heart with PO of 6 days (top row) and PO of 2 days
(bottom row). For both hearts short TE (21 µs), long TE (4 ms) and ΔUTE images are shown. Contrast enhancement is seen in
the ΔUTE image from the PO heart whereas no contrast enhancement is seen in the PO heart 2 days after myocardial
infarction.
In the long TE image from the 6 days PO mouse heart focal spots of missing signal were observed
in the infarct area, whereas in the long TE image from the 2 days PO heart no such voids were found.
The presence of signal voids points to the presence of short T2* values resulting in rapid dephasing.
The ΔUTE image of the 6 days PO heart shows significant contrast enhancement in the infarct region.
The ΔUTE image of the 2 days PO heart showed no such enhancement.

28
Figure 27 Anterior to posterior LA images of 3D datasets of a PO (A) and a healthy control (B) mouse heart with for both
hearts the short TE (left, 21 µs), long TE (middle, 4 ms) and ΔUTE (right) images. Contrast enhancement is seen in infarct
region of the ΔUTE image from the PO heart and in the aortic arch and basal area of the healthy control heart.
In Figure 27, short TE, long TE 3D UTE and ΔUTE images are shown of an ex vivo heart with a 14
day-old permanent occlusion infarct (right) and a healthy control heart (left). For both hearts LA cross
sections from posterior to anterior were made to see where contrast enhancement was observed.
On the ΔUTE images of the PO heart (Figure 27 A) contrast enhancement was observed in the thin
apical region of the heart in all slices. The control heart (Figure 27 B) showed no contrast
enhancement in the apical side of the heart. However, some enhancement was observed in the basal
side of the heart from the intact aorta arch. Very little signal voids were observed in the infarcted
heart 14 days after permanent occlusion (Figure 27 A) as compared to the infarcted heart 6 days
after permanent occlusion (Figure 26).

29
Figure 28 shows two examples of slices stained with Picrosirius Red (top row) together with the
corresponding LA ΔUTE images from a the PO heart of 2 days and the PO heart of 6 days. In this
image collagen is stained red and remote tissue yellow. In the infarct heart a red area is seen in the
apical infarcted region of the heart confirming the presence of collagen. In the Picrosirius Red image
of the control heart, note that this was an infarct heart with a 2 days old permanent occlusion, no red
enhancement and thus no collagen was observed in the apical region, which matched with the
absence of enhancement on the ΔUTE images.
Figure 28 Picrosirius Red histology staining for collagen and ΔUTE LA images from a PO heart of 2 days (control) and 14 days
(infarct) after myocardial infarction. Contrast enhancement is seen in both histology and ΔUTE image in the apical region of
infarct heart (indicated with arrows) whereas no contrast enhancement is seen in the apical region of the control heart.
When comparing histology images of both 2 days PO and 6 days PO hearts to the corresponding
ΔUTE images of these hearts (Figure 28), it can be clearly seen that the red colored region on the
Picrosirius Red stained histology slide of the 6 days PO heart corresponds to the enhanced region in
the ΔUTE image of this heart. Both the histology and ΔUTE image of the 2 days PO heart show no
contrast enhancement in the apical region.
Table 1 Total heart and enhanced volume determined from the ex vivo ΔUTE 3D datasets for control, 6 days and 14 days
post permanent occlusion of the coronary artery in C57BL/6 mouse hearts and percentage ratio between both determined
volumes.
Table 1 shows the analyzed total heart volume, enhanced volume and the calculated fraction of both
the ex vivo ΔUTE 3D datasets of 6 and 14 days old infarct hearts and control hearts. The total heart

30
volume is the lowest for the control heart, whereas the PO hearts 6 days after infarction have the
largest total heart volume. The total enhanced volume and fraction is the lowest in the control hearts
and increase with infarct duration in the PO hearts.
Figure 29 T2* fits (A&C) of two ROIs, infarct (B, green asterisks) and normal remote (B, red asterisks), in a SA 3D UTE slice of
a 6 days PO mouse heart. Fits are done on the whole TE range of 21 µs to 40 ms for both ROIs (A) and on a selection of
shorter TEs, 21 µs to 500 µs, in the infarcted ROI (C).
T2* fits from both infarct and remote areas, indicated with respectively green and red asterisks
on the SA short TE, long TE and ΔUTE images in Figure 29 B, from a PO heart 6 days after infarction
are shown in Figure 29 A. The signal decay of the infarcted tissue is faster than the remote tissue and
the computed T2*were 18.8 ms for normal remote tissue and 7.2 ms for the infarcted region. For TEs
from 21 µs to 500 µs an extra T2* fit is done in the infarct area to determine the T2* of the short
components such as collagen in the infarct, the found T2* was 110 µs.
To determine the best long TE to obtain optimal contrast from the short T2* components in the
ΔUTE images, the exponential decay in both infarct and remote region was quantified in Figure 29 A.
Optimal contrast in the ΔUTE image is obtained when the signal intensity of the infarcted region has
nearly decayed to zero and as much signal as possible from the normal remote area has remained.
This criterion is best met for TEs between 10 – 20 ms, as can be seen in Figure 29 A, where
approximately 10% of the signal remains from the infarct region, while for the normal region about
40 % of the signal remains.

31
4.3 In vivo MRI
In Figure 30 Cine MR images are shown from a PO and control heart. For each heart Cine frame
1, 4 and 8 are shown to illustrate the contraction of the heart. Differences between the PO heart
(Figure 30 A) and control heart (Figure 30 B) are clearly visible. First, the PO heart is dilated, in
particular in the apical side of the LV, which is markedly thinned as compared to the apical wall
thickness of the control heart. Second, little cardiac contraction is observed. In contrary, the control
heart shows more contractile motion and no thinning of the myocardium is observed.
Figure 30 FLASH Cine MR images (frame 1,4, and 8, corresponding to end-diastole, mid-systole and end-systole) of the heart
from a PO (A) and control (B) C57BL/6 mouse.
The left ventricular function parameters as well as the heart rate and respiration rate during the
MRI measurements, are shown in Table 2 for the control mice and the mice with a PO, both 6 and 14
days after infarction.
Table 2 General animal characteristics of the mice included in the 3D UTE experiment. Indicated are the
heart rate (HR) [min-1] and respiratory rate (Resp) [min-1] during the MR examination, end diastolic (EDV) [µl] and end
systolic volume (ESV) [µl], and stroke volume (SV) [µl] in control, PO 6 and 14 days.

32
For both control and PO mice the mean HR during MR scanning was comparable. Note that the
variation in heart and respiration rate was larger in the PO mice as compared to the controls. A large
difference in EF can be observed when comparing the EF from control to PO mice, in control mice the
mean EF is 56.3 % while the mean EF of both PO mice groups is around 33 %. Both EDV and ESV are
higher for both PO mice groups, indicative for the dilation of the LV. The EF, EDV, and ESV of the
healthy control group and both PO groups showed a significant difference tested with a unpaired
student t-test (all p-values < 0.006). The SV is about the same for all groups.
Figure 31 Long axis and short axis images from the same 3D UTE short TE (TE = 21 µs) dataset. Anatomical details such as
the LV, RV, papillary muscles (PM) are indicated. Also note that the gall bladder, lung tissue and saturation slice in the LA
images could be clearly observed.
In Figure 31 long axis and short axis images of a short TE (TE = 21 µs) 3D UTE dataset of a PO
mouse 6 days after infarction are shown. In both LA and SA images several anatomical details can be
distinguished. The RV, LV, and papillary muscles (PM) are indicated by these abbreviations in the
images. Overall image quality of these in vivo images acquired with the optimized 3D UTE is good.

33
Figure 32 Short axis cross-sections of an UTE short TE (21 µs), long TE (4 ms), and ΔUTE dataset of a control heart (top row)
and an infarct heart (bottom row) together with the corresponding LGE DE MR images. Contrast enhancement can be
observed in the ΔUTE and DE image of the infarct heart, whereas no enhancement is observed in the control heart.
In Figure 32 short and long TE 3D UTE, the corresponding ΔUTE and LGE SA images are shown
from a control and infarct mouse heart. When the ΔUTE image of the infarct heart is compared to
the control heart, clearly contrast enhancement can be observed in the myocardial wall of the infarct
heart, whereas no contrast enhancement is seen in the control heart. When both LGE images are
compared delayed enhancement can be observed in the myocardial wall of the infarct heart
(indicated with black arrows), whereas no delayed enhancement is observed in the control heart. The
delayed enhanced areas in the myocardial wall of the infarct heart correspond nicely to the contrast
enhancement observed in the myocardial wall, indicated with white arrows, of the corresponding
ΔUTE image of the infarct heart. Also, note that enhancement in the ΔUTE image is absent in some
parts of the infarct area.

34
4.4 Histology
A comparison of the Picrosirius Red stained slices of the different groups is shown in Figure 33. In
the control heart (Figure 33 A) only red staining is observed near the base of the heart, in particular
in the valves and aortic arch. In the PO heart 6 days after myocardial infarction, depicted in Figure 33
B, red staining is also observed in the aortic arch and valves. Moreover, red staining is observed in
the infarct area. The heart of 14 days (Figure 33 C) after myocardial infarction also shows red staining
in the infarct area. Note that the red staining is more intense 14 days after infarction as compared to
6 days after infarction, indicating the presence of a larger amount of collagen.
Figure 33 Picrosirius Red stained slices of control (A), PO 6 days (B) and 14 days (C) after myocardial infarction. Red color
represents collagen and yellow remote tissue in this histological staining. The arrow indicates the position of the suture used
to induce the permanent occlusion of the left descending coronary artery.
Hearts were also stained with Prussian Blue (Figure 34) to detect the presence of iron originating
from hemoglobin in blood. In the free wall of the control heart no brown color was detected (Figure
34 D). Note that no red staining is observed (Figure 34 C) in the corresponding Picrosirius Red stained
slice. In the infarct area of the PO mouse some small brown spots were observed, indicated by the
arrow (Figure 34 B). In the corresponding slice stained with Picrosirius Red a lot of red enhanced
collagen was observed (Figure 34 A).
Figure 34 Picrosirius red (A&C) and corresponding Prussian blue (B&D) stained histology slices for a PO (A&B) and control
heart (C&D). Picrosirius red colors collagen red, Prussian blue colors iron brown.

35
In Figure 35 an example is shown of the result of the algorithm used to detect the red enhanced
pixels in the Picrosirius Red stained slices representing collagen. Figure 35 A and B show the raw
microscope image and the detected enhanced pixels image in a PO heart, respectively. Note the
good correspondence between the red staining in the Picrosirius Red slice (Figure 35 A) and the
detected region of interest as obtained from the algorithm used to detect the red enhanced pixels
(Figure 35 B). For the example of a healthy mouse heart it becomes apparent that almost no red
stained pixels were detected. Figure 35 C and D show the raw image (C) and the detected enhanced
pixels (D) from a control heart.
Figure 35 Picrosirius red stained histology slices of a PO heart (A) and healthy control heart (C). Results of red enhanced
color detection algorithm of PO (B) and control heart (D).
Figure 36 shows the correlation between the ratio enhanced volume normalized from the ex vivo
ΔUTE 3D datasets and the ratio enhanced volume normalized to the total volume from the Picrosirius
red histology staining of control (n=2), and PO six days (n=1) and 14 days (n=1) after myocardial
infarction hearts. A linear relationship (r2 = 0.99 and p-value < 0.001) was observed, indicating a
significant correlation between both ratios. The largest fraction was found for the PO heart 14 days
after infarction.
Figure 36 Relationship between the ratio of the red stained volume as normalized to the total heart volume and the
enhanced volume as determined from the ΔUTE dataset normalized to the total heart volume.

36
5 Discussion
5.1 Sequence optimalization
5.1.1 Phantom test
Initial experience with the 3D UTE sequence was obtained with the use of an alginate – 0.1mM
Prohance containing water phantom. This phantom formed a non-expensive, easy-to-make, and
suitable phantom to test the detection of short T2 components with the 3D UTE sequence. The
results of the comparison of the conventional FLASH and 3D UTE sequence with this phantom
showed that the 3D UTE is suitable to detect short T2 components, whereas FLASH images only
showed signal from the water in the phantom. The strength of the ΔUTE method was first shown
using this phantom since the ΔUTE images only showed the alginate, containing mainly short T2
components. It was found that an undersampling up to factor 2 still led to good image quality, while
for higher undersampling factors artefacts appeared. This provided evidence that the acquisition
time for further measurements could be reduced by 50% without compromising image quality.
Exploratory tests with the alginate phantom also showed that relatively short gradient delays (<5µs)
resulted in the most homogeneous images. Most likely only a fairly short delay was necessary to
synchronize the read-out with the ramp-up of the gradients. Finally, initial test showed the
importance of accurate measurements of the actual k-space trajectories for optimal regridding of the
radial k-space spokes to a Cartesian grid, to achieve optimal image quality. Therefore, both the
correct gradient delay and accurate measurements of the actual k-space trajectories were obtained
using oil and CuSO4 containing phantoms.
5.1.2 Gradient delay
The measurements with the CuSO4 containing sphere phantom also showed that most
homogeneous images were obtained with a relative short gradient delay. Test with increasing
gradient delay confirmed that a gradient delay of 1µs led to optimal synchronization of the read-out
with the ramp-up of the gradients. The quality of the images obtained with varying gradient delays
was visually assessed. Optimal gradient delay was discussed with Bruker specialists and determined
in a homogeneous MR image of the CuSO4 containing sphere and in the absence of ringing artefacts
surrounding the sphere. The determined gradient delay provided both good in vivo and ex vivo image
quality, which can be explained by that the determined gradient delay is a property of the gradient
set, and in principle independent of the object in the RF coil.
5.1.3 Trajectory
Actual k-space trajectories were measured in both cylindrical and spherical phantoms. The k-
space trajectories measured using spherical phantoms followed a smooth curve with some curvature
at the start in the origin, due to the ramp up of the gradient. Since the trajectories obtained with the
spherical oil and CuSO4 phantom were very similar, for subsequent measurements the trajectory as
determined using the spherical CuSO4 phantom was used. Attempts were made to also measure the
k-space trajectories using cylindrical oil containing phantoms. However, the obtained trajectories
showed large distortions. These distortions were most likely caused by eddy currents induced in the
phantoms, resulting in large inhomogeneities of the magnetic field, which could disturb the precision
of the trajectory measurements.

37
5.1.4 Acceleration of the in-vivo data acquisition
Acquisition times of the 3D UTE datasets were reduced from 85 minutes to 43 minutes by means
of undersampling with a factor 2. However, for successful in vivo applications the acquisition time
needed to be reduced further, which is especially important in diseased mice not able to cope with
prolonged anesthetic duration. Therefore, the development and implementation of the logic device
enabled further acceleration of the 3D UTE scans from approximately 43 minutes to 15 minutes,
depending on the heart rate. This was accomplished by sending three trigger pulses generated by the
logic device to the scanner immediately after detection of a R-wave to acquire three radial k-space
spokes. The trigger pulses generated by the logic device are able to switch faster, because of the
temporal resolution of 50ns which is an order of magnitude better than the SAI monitoring system,
which might be useful in the future because the AVANCE III hardware unit of the scanner is not able
to make real time fast trigger decisions. Moreover, the logic device contains a second output channel
with the same functionality as the first channel. The second channel allows to send a different trigger
to the scanner which can be used in future research to trigger a second pulse programmed in a MR
sequence to give for example dummy pulses or to trigger another device to start an operation.
5.2 Ex vivo MRI
The 3D UTE ex vivo experiments on mouse hearts 6 days and 14 days after infarction and control
hearts revealed contrast enhancement in the ΔUTE images of the PO hearts, whereas no contrast
enhancement was observed in the control hearts. The contrast enhancement in the ΔUTE images of
the infarcted mouse hearts is very likely due to the presence of myocardial fibrosis, which mainly
consists of collagen having short T2 components. This finding was positively validated with Picrosirius
Red histology staining for collagen content. Signal voids were present in the PO heart 6 days after
infarction, but not 14 days after PO. This was probably due to different composition of the tissue in
the infarct region. This seems a logical explanation, because it is known that the remodeling process
in not completed 6 days after myocardial infarction. This result is in agreement with the histology
findings, where more extensive collagen formation was observed in the PO hearts of 14 days after
myocardial infarction in comparison to the PO hearts 6 days after myocardial infarction.
Large haemorrhages may be still present in the 6 days old myocardial infarct hearts, which could
lead to faster T2* decay causing the observed signal voids due to the presence of blood cloths
containing hemoglobin with iron molecules62. Another possibility is that there is formation of edema
in the infarcted area at this time point, which appears black on the 3D UTE sequence which also has a
T1 weighting due to the short TE and TR besides a strong short-T2* weighting47.
The total heart volume, total enhanced volume and the ratio of both volumes determined from
the 3D UTE datasets of control and PO hearts show that there is an increase in total enhanced
volume and ratio as the infarct ages, and a small decrease in the total heart volume from 6 to 14 days
post myocardial infarction. Important to note is that these determined volumes and ratio were done
on only two control hearts, one PO heart of 6 days after myocardial infarction and one PO heart of 14
days after myocardial infarction. So, quantification of total heart and enhanced volumes in more
hearts is necessary to obtain more conclusive results.
The results of the T2* fits from both infarct (7.2 ms) and remote (18.8 ms) regions performed in
an PO heart, with a T2* = 18.8 ms for remote tissue and a T2* = 7.2 ms for infarct tissue, are
comparable to previously reported values of 7.9 ms and 14.2 ms for ischemia/reperfusion infarct and
remote, respectively63. In particular the T2* in infarct area is long as compared to what is expected
from the reported T2 values of collagen in literature. This may be explained from the multi
exponentional relaxation behavior of the signal in the infarct area. When performing a T2* fit using
TE from 21µs to 500µs a T2* of 110 µs was found, which is a more expected T2* for collagen in the

38
fibrotic infarct area47. Both calculated T2*’s are based on fits performed in the infarcted heart of only
one mouse, so more hearts have to be analyzed before a real statement can be made.
The determined best long-TEs of 10-20ms to obtain optimal contrast in the ΔUTE image,
appeared to be too long for in vivo application for two reasons. First, longer TE results in prolonged
acquisition due to increased TR. Secondly, image quality was reduced. Therefore, the maximal long-
TE used for in vivo application was chosen as TE = 4 ms.
5.3 In vivo MRI
To obtain high quality MR images of the in vivo mouse heart, stability of the mouse during a scan
session is very important. This is mainly due to the susceptibility of UTE MRI for motion and magnetic
field inhomogeneities. Placement of the mouse heart in the isocenter of the magnet is very
important for cardiac MRI in general but especially for UTE application to minimize artefacts from
gradient imperfections and measured actual k-space trajectories. For these reasons the center of the
UTE imaging volume is fixed to the magnets isocenter and cannot be moved, the mouse heart is thus
positioned in the FOV instead of that the FOV is chosen such that it encloses the mouse heart.
Therefore, in this study a FOV of 3 x 3 x 3 cm3 was used to ensure that the whole heart was inside the
imaging volume. The size of the used matrix was also affected by this selected FOV, because a spatial
resolution of approximately 300 µm is needed to observe the thinned myocardial wall in the mouse
hearts with a myocardial infarction in the UTE images. The maximal matrix size was constrained to a
minimal of 128 x 128 x 128 for reasons of memory restrictions, resulting in a spatial resolution of 234
µm3 per pixel.
The acquired Cine MR images clearly showed differences in contraction of the healthy and
infarcted hearts. The stronger observed contraction of the heart in healthy mice in the Cine MR
images led to the decision to acquire less radial spokes each R-wave in healthy controls to reduce
motion artefacts in the 3D UTE images as compared to the mice with a myocardial infarction.
The quantified cardiac function parameters showed that the permanent occlusion of the left
descending coronary artery was successfully induced and led to myocardial infarction. Which was
clearly observed on the basis of reduced EF in the PO mice and increases in EDV and ESV.
The acquired in vivo 3D UTE datasets showed good general image quality. The saturation of the
inflowing blood into the LV with the saturation slice successfully generated contrast between the
lumen and myocardium. The influence of the saturation of the blood to the magnetization in the
myocardium was assumed negligible because each R-R interval the saturation was applied only three
times in rapid succession immediately after R-wave detection and from this point for approximately
80 ms no saturation was applied until a new R-wave was detected. In order to induce magnetization
transfer effects, typically flip angles > 20o are used together with continuous pulsing. Therefore,
these effects were assumed to be negligible.
The in vivo UTE scans were performed with two different TEs, the short-TE of 21 µs and the long-
TE of 4 ms. This resulted in ΔUTE images in which contrast enhancement was observed in the
myocardial wall of mice with a myocardial infarction, but no contrast enhancement was observed in
the healthy controls. However, the images with the long-TE of 4 ms often showed artefacts.
Therefore, at a later time point in this study it was decided to measure 5 different TEs, 0.021, 0.1,
0.3, 0.714, and 1.429 ms. The maximum TE of 1.429 ms reduced artefacts due to the earlier readout
while conserving contrast in the ΔUTE images. Also, this approach may allow for in vivo T2* mapping,
although no attempts have been made yet to do so, as this requires registration of the various 3D
UTE datasets. The TEs 0.714 ms and 1.429 ms were derived from the TE based phase cancelation
technique for water and fat64. With these echo-times the water and fat are in phase, reducing the

39
phase cancelation artefact, which is particularly useful in gradient echo techniques. Although the 3D
UTE sequence is no gradient echo technique, it was decided that it would be better to choose these
echo-times.
The observed contrast enhancement in the ΔUTE images of the mice with a 6 days old
permanent occlusion corresponds well to the LGE MR images. It should be mentioned, however, that
the visualization of signal enhancement on the LGE scans in PO mice was not straightforward, most
probably due to the poor perfusion of the infarct area. Therefore, no LGE measurements were done
in the mice 14 days post permanent occlusion.
5.4 Histology
The Picrosirius Red stained healthy control, myocardial infarction hearts of 2, 6 and 14 days after
PO showed an increase in red color in the apical region in the infarct hearts of 6 and 14 days after PO,
indicating the increase in the amount of collagen as the infarct ages and no red enhancement in the
apical region of the control heart and the 2 day PO heart. The red color in the PO heart of 6 days
after myocardial infarction is of rather low intensity, indicating that the collagen formation is still
ongoing at this time point. Also, this less intense red color in these hearts made it difficult to use a
fully automated RGB threshold based color detection algorithm. Thus, the threshold had to be
chosen for each heart individually, which made the custom build red color detection algorithm more
semi-automatic and time consuming. For the control and PO hearts of 14 days post myocardial
infarction the color detection algorithm worked well. Iron deposition in the infarct area was checked
with the Prussian Blue staining. Some iron was present, but in such small amounts that the influence
on the UTE MRI measurements is assumed small. Both the determined ratio of the enhanced over
total volume in the ex vivo 3D UTE datasets and the enhanced over total volume in the Picrosirius
Red staining showed a strong correlation (r2 = 0.99, P<0.001), supporting our intention to use the 3D
UTE MRI technique as a direct read out of collagen in mouse hearts with formation of myocardial
fibrosis due to a myocardial infarction. This correlation was so far based on the histology Picrosirius
Red and ex vivo 3D UTE images of four mouse hearts, so more hearts need to be analyzed to make
the above proposal more convincing.

40
6 Conclusion
The aim of this study was to demonstrate visualization of myocardial fibrosis in the infarcted
mouse myocardium by ultra short echo time (UTE) MRI. Myocardial fibrosis is an important hallmark
of various cardiac pathologies and is caused by an excessive accumulation of extracellular matrix
proteins, such as collagen. Endogenous contrast in the infarct area can be obtained with UTE MRI
sequences by exploiting the increased amount of collagen, having short T2 values. UTE MRI of
myocardial fibrosis, may provide a more specific and more direct detection method for myocardial
fibrosis as compared to the already available late gadolinium enhanced based MRI methods. The UTE
sequence was optimized, in phantom experiments, and tested for in vivo cardiac MR imaging of the
mouse. As proof-of-principle a mouse model of permanent occlusion of the left descending coronary
artery was used to induce myocardial infarction resulting in the formation of replacement fibrosis.
Both ex vivo and in vivo clear contrast was observed between remote and infarct mouse myocardium
with UTE MRI, likely caused by the presence of significant amounts of collagen in the infarct. To
validate that the observed enhancement was indeed due to the presence of collagen visual
comparison of the ex vivo UTE contrast enhancement in the UTE difference images was done with
the Picrosirius Red staining for collagen content and showed good agreement. Importantly, also a
strong correlation was obtained between the contrast enhanced infarct volume determined in the
UTE difference images and the red enhanced, collagen representing, infarct volume determined in
the Picrosirius Red histology images confirming that the enhanced area on the UTE difference images
indeed corresponded to areas with large amounts of collagen. In vivo UTE contrast enhancement
corresponds with delayed enhancement on LGE MR images.
In conclusion, this study showed that in vivo and ex vivo contrast in the infarcted heart can be
obtained using UTE MRI. In future research the possibility could be explored to use this sequence to
obtain endogenous contrast based on collagen content in mouse models of diffuse fibrosis. This may
improve knowledge and provide more insights in one of the most important hallmarks of various
cardiac pathologies.

41
7 Dankwoord
Bij het afronden van dit afstudeerproject met het schrijven van dit dankwoord realiseer je pas
hoeveel mensen hebben bijgedragen aan dit werk. Allereerst wil ik Bastiaan van Nierop bedanken
voor de mogelijkheid dit cardiovasculaire UTE project onder zijn supervisie te doen, voor zijn kritiek,
tips, matlab cursus, proefdier cursus, uitrij pasje, stroopwafels, nachtelijke MRI sessies, het veelvuldig
vullen van de koffie mok, Ultrashort TE: ‘’less is more’’ uitspraak, ISMRM Benelux praatje, en vooral
hoe je wetenschappelijk onderzoek doet. Leonie Niesen wil ik bedanken voor alle operaties die zij
voor dit project heeft gedaan. Larry de Graaf als onmisbare technische man van de groep wil ik
vooral danken voor de hulp met de triggerbox (ontwikkeld voor mijn project door Gerard Harkema
van de TUeDACS) maar ook voor het altijd snel oplossen van andere problemen of vragen. Noortje
Bax en Ariane van Spreeuwel, bedankt voor de hulp met de histology, zo mooi had ik zelf de hartjes
nooit kunnen snijden, plakken en kleuren. Verder wil ik Bastiaan van Nierop, Gustav Strijkers, Noortje
Bax, Marc Kouwenhoven en Klaas Nicolay bedanken voor hun deelname aan mijn afstudeer
commissie. Tevens wil ik Gustav Strijkers bedanken voor de wekelijkse bijeenkomsten en algemene
supervisie en Klaas Nicolay voor het mogelijk maken van het afstuderen binnen de Biomedical NMR
Group en zijn inzet voor de studie beurs van de Nederlandse Hartstichting die het voor mij mogelijk
maakt een externe stage te gaan doen in de groep van prof. Botnar in het St Thomas’ Hospital / Kings
College in London. Ook wil ik Rene van Donkelaar, en Anne Römgens bedanken voor het beschikbaar
stellen van samples zoals een meniscus en vele wervels van koeienstaarten die helaas het
uiteindelijke verslag niet gehaald hebben maar wel mooie UTE MRI beelden opleverden. Verder dank
ik alle (oud-)studenten van de Master room voor het halen van koffie en thee, vullen van de
snoeppot en het de vele leuke momenten. Als laatste wil ik de medewerkers van Vers Voordeel
Vaessen VOF, Huub , Marigold, Jeske, Marieke, en Lindsay bedanken voor hun altijd aanwezige steun
en vertrouwen.

42
8 Literature
1. Organization, W.H. Fact sheet No 317. (2008).
2. Roger, V.L. et al. Executive Summary: Heart Disease and Stroke Statistics-2012 Update A
Report From the American Heart Association. Circulation 125, 188-197 (2012).
3. Mendis, S. Global Atlas on Cardiovascular Disease Prevention and Control. (World Health
Organization, Geneva, 2011).
4. Mudd, J.O. & Kass, D.A. Tackling heart failure in the twenty-first century. Nature 451, 919-
928 (2008).
5. Fact sheet No 317. World Health Organization (2008).
6. McMurray, J.J. & Pfeffer, M.A. Heart failure. Lancet 365, 1877-89 (2005).
7. McMurray, J.J. & Stewart, S. Epidemiology, aetiology, and prognosis of heart failure. Heart
83, 596-602 (2000).
8. Gjesdal, O., Bluemke, D.A. & Lima, J.A. Cardiac remodeling at the population level -risk
factors, screening, and outcomes. Nature Reviews Cardiology 8, 673-685 (2011).
9. Drakos, S.G. et al. Impact of mechanical unloading on microvasculature and associated
central remodeling features of the failing human heart. Journal of the American College of
Cardiology 56, 382-91 (2010).
10. Opie, L.H., Commerford, P.J., Gersh, B.J. & Pfeffer, M.A. Controversies in ventricular
remodelling. Lancet 367, 356-67 (2006).
11. Krum, H. & Abraham, W.T. Heart failure. Lancet 373, 941-55 (2009).
12. Kaye, D.M. & Krum, H. Drug discovery for heart failure: a new era or the end of the pipeline?
Nat Rev Drug Discov 6, 127-39 (2007).
13. Horwich, T.B. & Fonarow, G.C. Glucose, obesity, metabolic syndrome, and diabetes relevance
to incidence of heart failure. Journal of the American College of Cardiology 55, 283-93 (2010).
14. Diez, J., Gonzalez, A., Lopez, B. & Querejeta, R. Mechanisms of disease: pathologic structural
remodeling is more than adaptive hypertrophy in hypertensive heart disease. Nat Clin Pract
Cardiovasc Med 2, 209-16 (2005).
15. Jourdan-Lesaux, C., Zhang, J. & Lindsey, M.L. Extracellular matrix roles during cardiac repair.
Life Sci 87, 391-400 (2010).
16. Berk, B.C., Fujiwara, K. & Lehoux, S. ECM remodeling in hypertensive heart disease. Journal of
Clinical Investigation 117, 568-575 (2007).
17. Creemers, E.E. & Pinto, Y.M. Molecular mechanisms that control interstitial fibrosis in the
pressure-overloaded heart. Cardiovascular Research 89, 265-72 (2011).
18. Diez, J. Mechanisms of cardiac fibrosis in hypertension. J Clin Hypertens (Greenwich) 9, 546-
50 (2007).
19. Ho, C.Y. et al. Myocardial Fibrosis as an Early Manifestation of Hypertrophic Cardiomyopathy.
New England Journal of Medicine 363, 552-563 (2010).
20. Broberg, C.S., Chugh, S.S., Conklin, C., Sahn, D.J. & Jerosch-Herold, M. Quantification of
diffuse myocardial fibrosis and its association with myocardial dysfunction in congenital
heart disease. Circ Cardiovasc Imaging 3, 727-34 (2010).
21. Weber, K.T., Sun, Y. & Diez, J. Fibrosis: a living tissue and the infarcted heart. Journal of the
American College of Cardiology 52, 2029-31 (2008).
22. Lopez, B., Gonzalez, A., Querejeta, R., Larman, M. & Diez, J. Alterations in the pattern of
collagen deposition may contribute to the deterioration of systolic function in hypertensive
patients with heart failure. Journal of the American College of Cardiology 48, 89-96 (2006).
23. Weber, K.T. Fibrosis in hypertensive heart disease: focus on cardiac fibroblasts. Journal of
Hypertension 22, 47-50 (2004).
24. Mewton, N., Liu, C.Y., Croisille, P., Bluemke, D. & Lima, J.A.C. Assessment of Myocardial
Fibrosis With Cardiovascular Magnetic Resonance. Journal of the American College of
Cardiology 57, 891-903 (2011).

43
25. Jellis, C., Martin, J., Narula, J. & Marwick, T.H. Assessment of Nonischemic Myocardial
Fibrosis. Journal of the American College of Cardiology 56, 89-97 (2010).
26. Sado, D.M., Flett, A.S. & Moon, J.C. Novel imaging techniques for diffuse myocardial fibrosis.
Future Cardiol 7, 643-50 (2011).
27. Hasenfuss, G. Animal models of human cardiovascular disease, heart failure and
hypertrophy. Cardiovascular Research 39, 60-76 (1998).
28. Nahrendorf, M. et al. Cardiac magnetic resonance imaging in small animal models of human
heart failure. Medical Image Analysis 7, 369-75 (2003).
29. Patten, R.D. & Hall-Porter, M.R. Small animal models of heart failure: development of novel
therapies, past and present. Circ Heart Fail 2, 138-44 (2009).
30. Osorio, J. New CMR approach to measure diffuse myocardial fibrosis. Nat Rev Cardiol 7, 542
(2010).
31. Assomull, R.G., Pennell, D.J. & Prasad, S.K. Cardiovascular magnetic nce in the evaluation of
heart failure. Heart 93, 985-992 (2007).
32. Assomull, R.G. et al. Cardiovascular magnetic resonance, fibrosis, and prognosis in dilated
cardiomyopathy. Journal of the American College of Cardiology 48, 1977-1985 (2006).
33. Coolen, B.F., Paulis, L.E., Geelen, T., Nicolay, K. & Strijkers, G.J. Contrast-enhanced MRI of
murine myocardial infarction - Part II. Nmr in Biomedicine (2012).
34. Geelen, T., Paulis, L.E., Coolen, B.F., Nicolay, K. & Strijkers, G.J. Contrast-enhanced MRI of
murine myocardial infarction - Part I. Nmr in Biomedicine (2012).
35. Jackson, E., Bellenger, N., Seddon, M., Harden, S. & Peebles, C. Ischaemic and non-ischaemic
cardiomyopathies--cardiac MRI appearances with delayed enhancement. Clin Radiol 62, 395-
403 (2007).
36. Kim, R.J. et al. Relationship of MRI delayed contrast enhancement to irreversible injury,
infarct age, and contractile function. Circulation 100, 1992-2002 (1999).
37. Moon, J.C. et al. The histologic basis of late gadolinium enhancement cardiovascular
magnetic resonance in hypertrophic cardiomyopathy. Journal of the American College of
Cardiology 43, 2260-4 (2004).
38. Kramer, C.M. Cardiac magnetic resonance imaging identification of myocardial fibrosis: the
need for standardization and therapies. Journal of the American College of Cardiology 56,
288-9 (2010).
39. Flett, A.S. et al. Equilibrium contrast cardiovascular magnetic resonance for the
measurement of diffuse myocardial fibrosis: preliminary validation in humans. Circulation
122, 138-44 (2010).
40. Iles, L. et al. Evaluation of Diffuse Myocardial Fibrosis in Heart Failure With Cardiac Magnetic
Resonance Contrast-Enhanced T-1 Mapping. Journal of the American College of Cardiology
52, 1574-1580 (2008).
41. Iles, L. et al. Evaluation of diffuse myocardial fibrosis in heart failure with cardiac magnetic
resonance contrast-enhanced T1 mapping. Journal of the American College of Cardiology 52,
1574-80 (2008).
42. Jellis, C. et al. Association of imaging markers of myocardial fibrosis with metabolic and
functional disturbances in early diabetic cardiomyopathy. Circ Cardiovasc Imaging 4, 693-702
(2011).
43. de Jong, S. et al. Direct detection of myocardial fibrosis by MRI. Journal of Molecular and
Cellular Cardiology 51, 974-9 (2011).
44. Tyler, D.J., Robson, M.D., Henkelman, R.M., Young, I.R. & Bydder, G.M. Magnetic resonance
imaging with ultrashort TE (UTE) PULSE sequences: technical considerations. Journal of
Magnetic Resonance Imaging 25, 279-89 (2007).
45. Bydder, M., Znamirowski, R.M. & Bydder, G.M. Magnetic resonance imaging of short T2
relaxation components in tissue using ultrashort echo time (UTE) pulse sequences. Current
Medical Imaging Reviews 2, 79-90 (2006).

44
46. Gatehouse, P.D. & Bydder, G.M. Magnetic resonance imaging of short T2 components in
tissue. Clin Radiol 58, 1-19 (2003).
47. Robson, M.D., Gatehouse, P.D., Bydder, M. & Bydder, G.M. Magnetic resonance: an
introduction to ultrashort TE (UTE) imaging. J Comput Assist Tomogr 27, 825-46 (2003).
48. Grenier, D., Pascui, O. & Briguet, A. Dipolar contrast for dense tissues imaging. Journal of
Magnetic Resonance 147, 353-6 (2000).
49. Bergin, C.J., Glover, G.H. & Pauly, J.M. Lung parenchyma: magnetic susceptibility in MR
imaging. Radiology 180, 845-8 (1991).
50. Du, J.A. et al. Qualitative and quantitative ultrashort echo time (UTE) imaging of cortical
bone. Journal of Magnetic Resonance 207, 304-311 (2010).
51. Chan, C.F. et al. Ultra-short echo time cardiovascular magnetic resonance of atherosclerotic
carotid plaque. Journal of Cardiovascular Magnetic Resonance 12 (2010).
52. Ernst, R.R. & Anderson, W.A. Application of Fourier Transform Spectroscopy to Magnetic
Resonance. Review of Scientific Instruments 37, 93-+ (1966).
53. Techawiboonwong, A., Song, H.K. & Wehrli, F.W. In vivo MRI of submillisecond T(2) species
with two-dimensional and three-dimensional radial sequences and applications to the
measurement of cortical bone water. Nmr in Biomedicine 21, 59-70 (2008).
54. Du, J. et al. Short T2 contrast with three-dimensional ultrashort echo time imaging. Magnetic
Resonance Imaging 29, 470-482 (2011).
55. Du, J., Bydder, M., Takahashi, A.M. & Chung, C.B. Two-dimensional ultrashort echo time
imaging using a spiral trajectory. Magnetic Resonance Imaging 26, 304-12 (2008).
56. Rahmer, J., Bornert, P., Groen, J. & Bos, C. Three-dimensional radial ultrashort echo-time
imaging with T(2) adapted sampling. Magnetic Resonance in Medicine 55, 1075-1082 (2006).
57. Michael, L.H. et al. Myocardial ischemia and reperfusion: a murine model. American Journal
of Physiology 269, H2147-54 (1995).
58. Patten, R.D. et al. Ventricular remodeling in a mouse model of myocardial infarction.
American Journal of Physiology 274, H1812-20 (1998).
59. Heijman, E. et al. Comparison between prospective and retrospective triggering for mouse
cardiac MRI. Nmr in Biomedicine 20, 439-47 (2007).
60. Coolen, B.F. et al. Mouse myocardial first-pass perfusion MR imaging. Magnetic resonance in
medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of
Magnetic Resonance in Medicine 64, 1658-63 (2010).
61. Bax, N.A.M. et al. Cardiac Malformations in Pdgfr alpha Mutant Embryos Are Associated With
Increased Expression of WT1 and Nkx2.5 in the Second Heart Field. Developmental Dynamics
239, 2307-2317 (2010).
62. van den Bos, E.J. et al. Magnetic resonance imaging of haemorrhage within reperfused
myocardial infarcts: possible interference with iron oxide-labelled cell tracking? European
Heart Journal 27, 1620-6 (2006).
63. Aguor, E.N. et al. Quantitative T (2) (*) assessment of acute and chronic myocardial
ischemia/reperfusion injury in mice. MAGMA (2012).
64. Woodward, P. MRI for Technologists (McGraw-Hill, Medical Pub., 2000).

45
9 Appendix
In vivo ultra short TE (UTE) MRI of mouse myocardial infarction
Adapted from: J.L. Nelissen, B.J. van Nierop, N.A.M. Bax, L. de Graaf, K. Nicolay and G.J. Strijkers.
In vivo ultra short TE (UTE) MRI of mouse myocardial infarction. In “Proceedings of the Annual
Meeting of the Benelux ISMRM Chapter”, January 2012, Leuven, Belgium. (oral presentation)