Download PDF

CLINICAL RESEARCH
Europace (2015) 17, 938–945
doi:10.1093/europace/euu310
Cardiac electrophysiology
3D delayed-enhanced magnetic resonance
sequences improve conducting channel
delineation prior to ventricular tachycardia
ablation
David Andreu 1, Jose T. Ortiz-Pérez 1, Juan Fernández-Armenta 1, Esther Guiu 1,
Juan Acosta 1, Susanna Prat-González 1, Teresa M. De Caralt 2, Rosario J. Perea2,
César Garrido 2, Lluis Mont 1, Josep Brugada 1, and Antonio Berruezo 1*
1
Arrhythmia Section, Cardiology Department, Thorax Institute, Hospital Clinic, Universitat de Barcelona, IDIBAPS (Institut d’Investigacions Biomèdiques August Pi i Sunyer),
C/Villarroel 170, 08036 Barcelona, Spain; and 2Radiology Department, Hospital Clinic, Universitat de Barcelona, Barcelona, Spain
Received 28 May 2014; accepted after revision 8 October 2014; online publish-ahead-of-print 23 January 2015
Aims
Non-invasive depiction of conducting channels (CCs) is gaining interest for its usefulness in ventricular tachycardia (VT)
ablation. The best imaging approach has not been determined. We compared characterization of myocardial scar with
late-gadolinium enhancement cardiac magnetic resonance using a navigator-gated 3D sequence (3D-GRE) and conventional 2D imaging using either a single shot inversion recovery steady-state-free-precession (2D-SSFP) or inversionrecovery gradient echo (2D-GRE) sequence.
.....................................................................................................................................................................................
Methods
We included 30 consecutive patients with structural heart disease referred for VT ablation. Preprocedural myocardial
characterization was conducted in a 3 T-scanner using 2D-GRE, 2D-SSFP and 3D-GRE sequences, yielding a spatial resoand results
lution of 1.4 × 1.4 × 5 mm, 2 × 2 × 5 mm, and 1.4 × 1.4 × 1.4 mm, respectively. The core and border zone (BZ) scar
components were quantified using the 60% and 40% threshold of maximum pixel intensity, respectively. A 3D scar reconstruction was obtained for each sequence. An electrophysiologist identified potential CC and compared them
with results obtained with the electroanatomic map (EAM). We found no significant differences in the scar core mass
between the 2D-GRE, 2D-SSFP, and 3D-GRE sequences (mean 7.48 + 6.68 vs. 8.26 + 5.69 and 6.26 + 4.37 g, respectively, P ¼ 0.084). However, the BZ mass was smaller in the 2D-GRE and 2D-SSFP than in the 3D-GRE sequence
(9.22 + 5.97 and 9.39 + 6.33 vs. 10.92 + 5.98 g, respectively; P ¼ 0.042). The matching between the CC observed in
the EAM and in 3D-GRE was 79.2%; when comparing the EAM and the 2D-GRE and the 2D-SSFP sequence, the matching
decreased to 61.8% and 37.7%, respectively.
.....................................................................................................................................................................................
Conclusion
3D scar reconstruction using images from 3D-GRE sequence improves the overall delineation of CC prior to VT ablation.
----------------------------------------------------------------------------------------------------------------------------------------------------------Keywords
Magnetic resonance imaging † Ventricular tachycardia ablation † Conducting channels
Introduction
In recent years, many studies have clearly documented a relationship
between several myocardial fibrosis parameters derived from late
gadolinium enhancement cardiac magnetic resonance (LGE-CMR)
and ventricular arrhythmias occurrence or inducibility.1 – 5 From a
mechanistic point of view, conducting channels (CCs) are considered
a probable isthmus of ventricular tachycardia (VT), and substrate ablation techniques consider complete CC elimination critical to
achieving non-inducibility of ventricular arrhythmias.6 – 8
Several studies have demonstrated the utility of LGE-CMR imaging
to identify CC using 3D scar reconstructions to characterize the
myocardial fibrosis in two types of tissues: the scar core, represented
by dense hyperenhanced areas, and the border zone (BZ), reflecting
* Corresponding author. Tel: +34 93 2275551; fax: +34 93 4513045. E-mail address: [email protected]
Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2015. For permissions please email: [email protected].
939
3D CMR sequences improve conducting channel delineation
What’s new?
† Several studies have demonstrated the capability of late gadolinium enhancement magnetic resonance images (LGE-MRIs)
to identify the presence of border zone corridors (channels)
in the ventricular scar tissue. These border zone corridors
match in a high proportion of cases with conducting channels
identified on the electroanatomical maps. This is considered
the substrate for the ventricular arrhythmias (VA) and the
target for ablation.
† Various different LGE-MRI sequences are capable of identifying this VA substrate. However, no direct comparison has
been made between theses sequences. It could be important
to standardized image acquisition and to optimize VA substrate identification.
† This study demonstrates that 3D sequences provide the best
image quality to depict the VA substrate and that this is due to
the absence of shifting between consecutive slices but also to
the higher spatial resolution as compared with the standard
2D sequences.
areas of heterogeneous non-dense hyperenhancement likely capable
of transmitting the electric impulse with slow conduction.9,10 The
integration of these reconstructions into the navigation system gives
valuable information about the location and size of the CC prior to
electroanatomic mapping.11
Delayed-enhanced CMR imaging using an inversion recovery 2D
spoiled gradient echo sequence (2D-GRE) represents the standard for
clinical characterization of the myocardium.12 A single-shot inversionrecovery steady-state free-precession sequence (2D-SSFP), which
provides fast data acquisition that allows imaging in challenging
patients, has been clinically validated against 2D-GRE.13 An approach
based on free-breathing imaging with a navigator-gated inversion
recovery 3D sequence (3D-GRE) compares favourably with standard 2D and allows 3D reconstruction without slice-shifting
artifacts.14 In addition, the use of higher field strength at 3 T might
be exploited to reduce slice thickness to the level needed for appropriate CC depiction.
The aim of this study was to evaluate the capacity to identify the
presence of CC using 3D-GRE, 2D-GRE, and 2D-SSFP sequences
as compared with electroanatomical maps (EAMs) in patients referred for VT ablation.
Methods
Patient sample
As part of our ongoing clinical protocol, patients with ventricular arrhythmia and structural heart disease who are referred for catheter ablation
undergo a LGE-CMR examination. Our study consisted in the addition
of the 3D-GRE sequence to the clinical LGE-CMR protocol. The local
ethics committee approved the study protocol and all included patients
signed the informed consent. Patients with implantable devices or other
classical contraindications to LGE-CMR, such as clinical instability and
claustrophobia, and those referred for VT ablation and arrhythmogenic
right ventricular dysplasia were excluded from the study.
Delayed-enhanced CMR protocol
From June 2009 to September 2012, our hospital scheduled 31 patients
for LGE-CMR examination. One patient was excluded due to patient intolerance to the long acquisition time of the 3D-GRE sequence. The 3
Tesla scanner (Magnetom Trio a Tim System, Siemens Healthcare) was
equipped with 32 independent receiver channels and a 16-channel
cardiac coil. The system provides a strength of 45 mT/m, enabling a
slew rate of 200 T/m/s. No medications, other than those given for the
clinical VT, were administered prior to the study. The patients were
instructed to maintain a shallow, steady respiration. An intravenous
dose of 0.2 mmol/kg of gadodiamide (Omniscan, Amersham Health)
was given, and LGE-CMR imaging was started 7 min later to ensure the
correct acquisition of the three sequences. A TI scout was prescribed
in a mid-ventricular short-axis view to select the appropriate TI to null
healthy myocardium.
First, a whole-heart, high spatial resolution, delayed-enhanced study
was conducted using a commercially available free-breathing, 3D-GRE
inversion-recovery gradient echo technique as previously described
(Gradient echo readout).14 The 3D slab was acquired in the axial plane.
A Cartesian trajectory was used to fill the k-space with phase-encoding
(y) in the anteroposterior direction. The field of view was covered by a
256 × 240 pixel matrix, and in-plane reconstruction was allowed to
achieve an isotropic spatial resolution of 1.4 × 1.4 × 1.4 mm and a
voxel size of 2.74 mm3. To compensate for the long acquisition time
anticipated, we added 50 ms to the nominal value necessary to null
normal myocardium, as derived from the TI scout. Other typical sequence parameters were as follows: repetition time, 440 ms; echo
time, 1.29 ms; flip angle, 158; bandwidth, 810 Hz/pixel; and 51 k-space
lines filled per heartbeat. In some cases, a high temporal resolution,
4-chamber view cine, achieved by means of parallel imaging with an acceleration factor of 3, was used to select the optimal acquisition window and
minimize cardiac motion in late-diastole. Respiratory synchronization
was performed for every other heartbeat using a crossed-pair navigator
approach. The dataset was acquired during expiration and generalized,
autocalibrating, partially parallel acquisition with an acceleration factor
of 2 was used to speed up data acquisition. A set of images was reconstructed in the left ventricle (LV) short-axis orientation with 1.4 mm
slice thickness (typically 50 – 70 images) for subsequent image processing.
A new TI-scout was prescribed to select an updated TI.
Using a well-established, segmented 2D inversion-recovery gradientecho sequence (2D-GRE), a set of conventional delayed enhanced
images was obtained. Consecutive 5 mm thick slices with no gap
between them were prescribed to accomplish full LV coverage in the
short-axis orientation (one slice per breathhold). Typical parameters
were repetition time 790 ms; echo time 2 ms; flip angle, 258; bandwidth,
140 Hz/pixel; and 35 – 45 k space lines filled every other heartbeat,
needing 10 – 15 heartbeats per slice. This sequence has been validated
for clinical use at 3 T.15 The mean in-plane resolution was 1.4 ×
1.4 mm, and the voxel size was 9.80 mm3.
Finally, a stack of ultra-fast short-axis slices were obtained from the
same locations as the 2D-GRE, using a single-shot inversion-recovery
2D-SSFP without phase-sensitive reconstruction. This technique has
been validated against 2D-GRE at 1.5 T and 3 T field strength.16 Basic
parameters were as follows: repetition time, 830 ms; echo time,
2.27 ms; flip angle, 508; and bandwidth, 1240 Hz/pixel. Trigger was set
as every other heartbeat. Typically, 26 cardiac cycles were necessary to
image 13 slices during one single breathhold. A matrix size of 192 ×
192 pixels and a mean field of vision of 380 × 380 mm rendered an
in-plane resolution of 2.0 × 2.0 mm and a voxel size of 20.00 mm3.
Figure 1 shows a comparison of LGE-CMR images from the different
sequences with a similar slice position.
940
D. Andreu et al.
was also obtained by measuring the scar area in the EAM and counting
the total number of points in these areas.
A
Myocardial fibrosis quantification
The myocardial fibrosis region was delimited manually in all short-axis
slices by using the segmentation programme TCTKw. The myocardial fibrosis tissue was split into core and BZ using an algorithm based on the
pixel with the maximum signal intensity (MPI) of the myocardial fibrosis
region. Thresholds were calculated between core and BZ as 60% of MPI
and between BZ and healthy tissue as 40% of MPI, as previously
described.10 Total core and BZ mass were obtained by multiplying the
product of all the voxels for each region by mvoxel (further details in
Supplementary material online Files).
B
C
Figure 1 View of a representative slice of the same patient
obtained by the 2D-GRE sequence (A), the 2D-SSFP sequence
(B), and the 3D-GRE sequence (C).
Besides, standard 2-, 3- and 4-chamber views and LV functional assessment was performed using a standard cine SSFP.
Electrophysiological study
A CARTO navigation system (Biosense Webster, Diamond Bar) was
used to guide the ablation. A tetrapolar diagnostic 6F catheter was introduced through the femoral vein and placed at the RV apex. A 3.5 mm electrode irrigated-tip catheter (Thermocool Navistarw, Biosense Webster)
was introduced through transseptal or retrograde aortic access for endocardial mapping. When endocardial mapping did not identify a substrate
for VT or endocardial ablation was unsuccessful, nonsurgical transthoracic epicardial access was performed for epicardial mapping and ablation.
A high-density LV endocardial substrate EAM was obtained in each
case to identify the presence of CC. In the case of epicardial access, an
epicardial EAM was also obtained. The point density in the scar regions
Identification of conducting channels
Using an investigational software tool (ADAS, Galgo Medical SL), LV
endocardial and epicardial borders were delimited with a semiautomatic
segmentation algorithm. A manual adjustment was required to fit the
surface to the CMR images. Starting from the LV borders, five concentric
layers from the endocardium to epicardium were created (10%, 25%,
50%, 75%, and 90% of the LV wall thickness). After the LV surfaces
were obtained, the CMR information was projected over each surface
mesh node using a trilinear interpolation, and was colour-coded in
order to visualize the signal intensity distribution (see Supplementary material online Files for further details).
Using the same thresholds described previously to differentiate the scar
core from the BZ and the BZ from the healthy tissue (starting at 60% and
40% of the MPI, respectively, and using a maximum variation of +5%), the
colour scale was modified to mimic that used in the EAM.9,10,17
Shifting between consecutive images was evaluated qualitatively as
follows: (1) absent, when no spatial displacement between slices was
observed; (2) mild, when shifting between images was present but did
not compromise the segmentation; (3) moderate, when image shifting
caused inadequate segmentation in any region; and (4) severe, when
image shifting impeded correct 3D reconstruction of the LV.
All CCs were identified both in the LGE-CMR reconstructions and
EAM. A CC in the LGE-CMR reconstruction was defined as a corridor
of BZ between two core areas or between a core area and a valve
annulus, whether observed in the same layer or between consecutive
layers.9 An electrophysiologist, masked to the EAM data, analysed the
3D reconstructions and identified the CC according to their distribution
across the whole LV wall thickness. If the image quality of any set of images
was too low to be reconstructed (the presence of artefacts or severe
slice-shifting effects) the set was considered unreadable. Another electrophysiologist blinded to the LGE-CMR reconstructions labelled the
presence of CC in the EAM. The CC in the EAM was defined as
voltage CC and late-potential (LP) CC, following the same criteria
defined in a previous study.9 An example is shown in the Supplementary
material online Files. After that, both readings were compared side by side
by the two electrophysiologists and they determined the matching
between CC observed in the EAM and those observed in the
LGE-CMR reconstructions (see Figure 2). A third independent observer
assessed the agreement between LGE-CMR and EAM in CC depiction in
case of disagreement. In cases where only endocardial EAM were available, the CC identified in the epicardial LGE-CMR 3D reconstructions
(layers 75% and 90%) were not taken into account in the final analysis.10
Statistical analysis
Quantitative variables are expressed as mean value + standard deviation
and qualitative variables as number and percentage. Comparisons
between the different sequences were made using paired t-test and the
941
3D CMR sequences improve conducting channel delineation
A
10%
25%
50%
1
1
75%
B
90%
1
10%
50%
M
M
IRGE-2D
IRGE-2D
M
M
5
IRSSFP2D
2
2
IRSSFP2D
M
1
3DNAV
25%
1
5
1
5
2
3DNAV
2
Endocardial
electroanatomical map
2
His
3
M
Endocardial
electroanatomical map
Mitral annulus
Mitral annulus
1
2
Epicardial
electroanatomical map
M
1
His
5
4
2
Figure 2 Identification of CCs in the different 3D LV reconstructions using CMR sequences and in the EAM. (A) Example of an ischaemic patient
with inferior infarction. Layers at the 10%, 25%, and 50% of the LV wall thickness are shown (inferior view). In the EAM, two CCs) are identified,
displayed as white dotted lines. These CCs are also observed in the 3D LV reconstruction using the 3D-GRE sequence. The 3D LV reconstructions
using the other sequences only identified one CC of the two observed in the EAM. Additional channels observed only in the LV CMR reconstructions
from the 2D-GRE and 2D-SSFP are also shown (blue M tags and blue dotted lines) that can be considered false positive. (B) Example of an ischaemic
patient with a transmural inferior infarction. Layers at the 10%, 25%, 50%, 75%, and 90% of the LV wall thickness are shown (inferior view). In the EAM,
four CCs are identified in the endocardial EAM and another in the epicardial EAM, displayed as white dotted lines. The epicardial CC is observed in all
the 3D LV reconstructions using the three described sequences. The 3D LV reconstructions using the 3D-GRE sequence identifies two additional
CCs, whereas the 3D LV reconstructions using the 2D sequences only are able to identify one of the four endocardial CCs observed in the EAM. An
additional channels are observed only in the LV CMR reconstructions from the 2D acquisitions are shown (blue M tags and blue dotted lines).
repeated-measures analysis of variance test. The intra-observer agreement and inter-observer agreement were assessed using Lin’s coefficient.
Statistical significance was set at P , 0.05. All data were analysed using the
PASW Statistics 18.0 software (SPSS Inc).
Results
A total of 23 (76.7%) patients with ischaemic cardiomyopathy and 7
(23.3%) patients with non-ischaemic cardiomyopathy were included
in the study. Twenty-four (80.0%) patients were referred for VT
ablation and 6 (20.0%) patients were referred for scar-related premature ventricular contractions ablation. Patient characteristics are
shown in Table 1.
Delayed-enhanced cardiac magnetic
resonance results
The LGE-CMR study was performed a mean of 6.4 + 10.7 days prior
to the VT ablation procedure. The mean acquisition time of the
whole scan was 89.60 + 16.54 min (including the administration of
gadolinium). The mean acquisition time was 11.64 + 4.28 min for
the 2D-GRE sequence and 1.06 + 0.64 min and 16.43 + 7.19 min
for 2D-SSFP and 3D-GRE, respectively (P , 0.001). As expected,
the number of short-axis images available for analysis was higher
with the 3D-GRE sequence. Accordingly, the time required for LV
segmentation was higher with the 3D-GRE compared with the
others (Table 2).
There were no differences in total myocardial fibrosis mass
between the different LGE-CMR sequences. No significant differences were observed in the core mass images obtained with
3D-GRE as compared with the standard 2D-GRE sequence, but
there was a small difference (P ¼ 0.007) as compared with the
2D-SSFP sequence. No differences were found in the total core
mass between the 2D-GRE and 2D-SSFP images. In line with these
results, the BZ mass was higher in the 3D-GRE than in the 2D-GRE
and the 2D-SSFP sequences. More details are shown in Table 2.
The intra-observer agreement showed a good matching between
the different measures (Lin’s coefficient was .0.9 in all cases). See
Supplementary material online Table S1 for further details.
942
D. Andreu et al.
In four (13.3%) cases, the 2D-GRE sequence had severe shifting
and 3D reconstruction could not be performed. In five (16.7%)
cases, moderate shifting was observed. With regard to the
2D-SSFP sequence, moderate shifting was observed in five (16.7%)
studies with no cases of severe displacement. One patient (3.2%)
was excluded because the 3D-GRE sequence was not completed
due to claustrophobia.
Conducting channel analysis
A total of 30 endocardial EAMs and 9 epicardial EAMs were obtained.
The mean number of points for the endocardial and epicardial EAMs
was 422.6 + 194.1 and 475.0 + 242.7, respectively. The mean point
density in the scar areas was 7.45 + 2.84 points/cm2. The overall
number of CC identified in the EAM was 77 (mean 2.55 + 1.64). A
total of 74 CCs (mean 2.47 + 1.25) were identified in the 3D-GRE
dataset, whereas only 54 CCs (mean 2.08 + 1.26) and 37 CCs
(mean 1.23 + 0.97) were depicted using the 2D-GRE and 2DSSFP, respectively. The agreement (or sensibility) between the CC
present in the EAM and those identified in the 3D LV reconstructions
using the 2D-GRE, 2D-SSFP, and 3D-GRE sequences were 61.8%,
37.7%, and 79.2%, respectively (see Figure 3). The analysis of the
2D-GRE was made taking only into account the cases where the
Table 1 Patient demographics
Demographic data
N 5 30
Age (years)
65.6 + 9.3
Men
27 (90.0%)
Hypertension
Diabetes mellitus
23 (76.7%)
5 (16.7%)
Dyslipidemia
24 (80.0%)
................................................................................
Cardiac magnetic resonance data
LV ejection fraction (%)
30.3 + 10.9
LV end-diastolic volume (mL)
220.7 + 88.3
LV end-systolic volume (mL)
LV mass (g)
159.1 + 79.9
125.9 + 32.3
Antiarrhythmia treatment
Beta blockers
Amiodarone
20 (66.7%)
9 (30.0%)
3D reconstruction using the images provided by this sequence was
possible.
Data on baseline VT isthmus identification cannot be provided, as
the four-step ablation protocol in our centre (mapping, substrate ablation, re-mapping, and the induction and ablation of residual inducible VT if any) does not include a VT induction attempt before
ablation. We identified 24 VT isthmuses, most of them residual inducible VTs for which the VT isthmus was identified during the re-map
after substrate ablation. These VT isthmuses were also identified
in the 3D LV reconstructions in 15 (68.2%) VT cases when the
2D-GRE sequence was used, 7 (29.2%) VT cases using the 2D-SSFP
sequence, and 21 (87.5%) cases using the 3D-GRE sequence.
Discussion
Characterization of fibrotic tissue with LGE-CMR is widely performed using conventional imaging based on breathhold segmented
inversion recovery gradient echo or rapid SSFP techniques validated
for diagnosis and viability evaluation in multiple clinical scenarios.
However, imaging at a higher spatial resolution is becoming an imperative in the field of arrhythmias, where a higher level of detail is
mandatory to plan ablation strategy and technique. This study
shows that imaging with a free-breathing 3D acquisition at 3 T field
strength is superior to conventional LGE-CMR techniques to
depict CC among patients referred for ventricular arrhythmia
ablation. This higher capacity of the 3D-GRE sequence to detect
the arrhythmogenic substrate could be used for better sudden
cardiac death risk stratification and to better guide and plan ablation
procedures.
The unique ability of cardiac LGE-CMR to depict fibrotic myocardium with unprecedented precision makes the technique particularly
attractive in the field of arrhythmias. From the very beginning, even
coarse features reflecting myocardial fibrosis morphology on conventional 2D LGE-CMR images, such as infarct mass or infarct
surface area, have been identified as predictors of ventricular arrhythmia inducibility.1 Later on, a more refined approach to characterize
the myocardial fibrosis components into a dense core or an heterogeneous mixture of viable and patchy fibrotic myocardium, called
border zone, proved to be a stronger predictor of cardiac mortality
and ventricular arrhythmia inducibility than infarct sizing.2,3 In addition, several experimental studies have demonstrated distinctive
electrophysiological conduction properties of myocardial tissue
according to their characterization using high-resolution LGE-CMR
Table 2 Comparison between the three sequences studied
2D-GRE
2D-SSFP
3D-GRE
2D-GRE vs.
2D-SSFP (P value)
2D-GRE vs.
3D-GRE (P value)
2D-SSFP vs.
3D-GRE (P value)
...............................................................................................................................................................................
Myocardial fibrosis mass (g)
Core mass (g)
BZ mass (g)
Acquisition time (min)
Segmentation and
processing time (min)
16.70 + 10.89
17.65 + 10.70
17.18 + 9.32
P ¼ 0.168
P ¼ 0.689
P ¼ 0.641
7.48 + 6.68
9.22 + 5.97
8.26 + 5.69
9.39 + 6.33
6.26 + 4.37
10.92 + 5.98
P ¼ 0.280
P ¼ 0.815
P ¼ 0.522
P ¼ 0.028
P ¼ 0.007
P ¼ 0.040
11.64 + 4.28
1.06 + 0.64
16.43 + 7.19
P , 0.001
P ¼ 0.008
P , 0.001
1.85 + 0.38
1.70 + 0.42
9.01 + 2.83
P ¼ 0.002
P , 0.001
P , 0.001
943
3D CMR sequences improve conducting channel delineation
Conducting channels observed
90
80
74
77
IRGE-2D
70
60
IRSSFP-2D
3DNAV
EAM
61
54
48
50
40
42
37
29
30
26
16
20
10
12
8
13
0
Total observed
Both on EAM and on LGE-CMR Only on EAM (False negative)
(True positive)
Only on LGE-CMR (False
positive)
Figure 3 Matching between the CCs observed in the substrate EAM and the CCs identified in the 3D left ventricle reconstructions using different
CMR sequences.
imaging, which opens the possibility for the technique to become a
clinical tool for preprocedural planning of VT ablation. Ashikaga
and coworkers showed that LGE-CMR with submillimetre resolution is able to reveal the isthmus of the reentry circuits critical for
VT generation as areas of complex 3D disarrayed mixture of viable
and fibrotic tissue.8 Their study used postmortem cardiac imaging,
in which cardiac and respiratory motion was not at play, and acquisition time limitations were not an issue. Since then, a great interest in
developing fast, high-resolution imaging techniques has evolved into
the development of clinical scenarios that overcome the inherent
limitations of in vivo imaging.
So far, the accumulated evidence suggests that areas with intermediate grades of viable tissue within the scar are the substrate for
slow conduction channels.18 These areas can be depicted by
LGE-CMR, although the equipment and sequences used, as well as
the spatial and temporal resolution achieved, affect the imaging
results. In this regard, imaging with sufficient spatial resolution
remains a critical issue. Previous work demonstrated that the
amount of myocardium defined as BZ within the myocardial fibrosis
varies substantially depending on the spatial resolution achieved.
Using delayed-enhancement imaging in a murine model at high field
strength, Schelbert and co-workers observed an excellent correlation between the gadolinium deposition and the histological
pattern of fibrosis distribution nearly at the cellular level. Interestingly, when image resolution was degraded to resemble the clinical situation, the BZ increased but total infarct size remained unchanged.19
The authors claimed that partial volume effects accounted for the increase in BZ when image resolution was pushed down. In line with
these results and those obtained in other studies where comparison
in scar quantification between 2D and 3D sequences was performed,
our study did not find significant differences in total infarct sizing with
the 3D and 2D sequences.20 However, despite a higher spatial resolution with the 3D-GRE technique, we found that BZ was significantly
larger with the 3D-GRE sequence than with the 2D sequences, a
finding that cannot be explained by an increase in partial volume
effect. This might have accounted for an increase in the amount of
tissue showing intermediate signal intensity, finally classified as BZ.
Another factor to consider is the long acquisition time needed for
3D-GRE datasets, leading to contrast diffusion and suboptimal T1
nulling effects. Thus, 3D images visually look less sharp and contrasted than conventional 2D images, but when automatic postprocessing is conducted, a higher number of CC can be identified.
The 2D-GRE and 2D-SSFP sequences provide acceptable myocardial
fibrosis mass quantification while 3D-GRE sequences best depict the
CC, compared with CC observed in EAM.
Conducting channels delineation
Correct CC visualization using LGE-CMR images before the ablation
procedure can be a useful tool to plan and guide the intervention.9,10,21 Nevertheless, different LGE-CMR sequences were
described in each of those studies. 2D-GRE sequences present the
problem of displacement between slices due to respiratory motion
that limits a correct 3D reconstruction of the LV. In this study, 30%
of the cases had severe or moderate displacement between slices
that impeded a correct 3D reconstruction. 2D-SSFP theoretically
corrects this undesirable effect by making the full acquisition in a
single breathhold. However, in our study 16.7% of cases have moderate displacement between images, probably as the result of the need
of an excessively long breathhold. The effect of displacement
between consecutive slices of a set of images may result in critical
errors in the 3D reconstructions of the LV surface. If a slice registration method to correct the displacement is not applied to the set of
images, erroneous projection of CMR information over the 3D LV
reconstructions could lead to incorrect interpretation of the data.
The 3D LV reconstructions using the 3D-GRE sequences provided
the highest number of CC also observed in the EAM. The percentage
of CC identified both in the EAM and in the 3D LV reconstruction was
similar to that reported in previous studies.9,10 This can be explained
by the higher spatial resolution of this sequence, which facilitates the
depiction of CC in the 3D LV reconstruction. For the same reason,
944
D. Andreu et al.
Slice A
Ao
RV
LV
Slice B
RV
LV
Slice C
RV
LV
90%
Mitral annulus
Slice A
Slice B
Slice C
Figure 4 Identification of CCs in the original LGE-CMR images.
Three consecutive slices are shown in this image. The 3D LV reconstruction shown corresponds to the 90% layer of case A,
Figure 2. The area marked by a white dotted circle in the
LGE-CMR image corresponds to the white dotted circle region
in the 3D LV reconstruction, related to CC number 5 identified
in the electroanatomic map (Figure 2). The area marked by a
cyan dotted circle in the LGE-CMR image corresponds to the
cyan dotted circle region in the 3D LV reconstruction, related to
a CC identified only in the 3D LV reconstructions. Ao, aorta;
RV, right ventricle.
2D-SSFP provided the lowest sensitivity in identifying CC. The percentage of false positives (20%) was very similar in the three
sequences.
The results from this study support that the higher sensitivity in CC
detection is obtained at the cost of longer acquisition and processing
times, especially with the 3D-GRE sequence. The 2D-GRE sequence
could provide an alternative because of its capability to identify up to
60% of CC with lower acquisition and processing times. In those
cases in which the acquisition time of 3D-GRE and even 2D-GRE
could be problematic (e.g. claustrophobia, breathing-related difficulties), the 2D-SSFP sequence could be a useful alternative, although
the sensitivity of this sequence is modest.
In some cases, the length/width of the CC identified was close to
the spatial resolution of the sequence because the 3D LV reconstructions were used for observation of CC, rather than the raw CMR
images. The trilinear interpolation applied to the LV surface nodes
made this possible, smoothing the transition between the different
SI of the voxels over the LV reconstruction and sometimes
showing CC with a length close to the spatial resolution of the
CMR sequence (see Figure 4). In addition, a previous study described
the capability of a 2D sequence similar to that used in this study and
with the same spatial resolution to identify CC related to VT isthmuses.22 In that study, 100% of the CCs identified in the EAM were
observed in the LGE-CMR images. The reason for the mismatch
between that study and our results was mainly due to differences in
methodology. All CCs identified in the EAM by Pérez-David et al.
were voltage CCs, whereas in our study both voltage and LP CCs
were considered for the comparative analysis.
Although most of the CCs observed in the EAM can be identified
using the 3D-GRE sequence, nearly 20% of CCs still not visible in the
3D LV reconstructions. Higher LGE-CMR spatial resolution could
improve the identification capacity.
3 T scans provide a higher signal level, reducing time scans and/or improving the quality of the images obtained by means of reducing the
voxel size. This theoretically results in an increased number of CCs
identified in the CMR images. However, as these CMR studies were
not also done in a 1.5 T scan, we do not have data to support this statement. Nevertheless, the sequences described can be performed in a
1.5 T scan, although the results of CC depiction might be different.
Point density in the EAM is critical to obtain detailed information of
the tissue. Sometimes certain areas of the LV are poorly mapped and
this may result in losing important information about the presence/
absence of CC. In fact, this may be critical for the ablation success.
In this context, the CMR could not only correlate with the EAM,
but also to show additional tissue details to the EAM. This is why a
perfect match should probably not be pursued as the endpoint, but
an acceptable similarity.
Limitations
Our study has several limitations. In 3.2% of the cases, the 3D-GRE
could not be completed due to respiratory motion or excessively
long acquisition time. Factors related to patient cooperation and arrhythmia may increase substantially the acquisition time, which ends
up limiting image quality and the clinical applicability of the technique.
In addition, we did not test other validated methods for classifying BZ,
such as the standard deviation approach. In a recent study, the BZ
945
3D CMR sequences improve conducting channel delineation
definition using the standard deviation method was superior to that
achieved with methods proposed by Yan and Schmidt to predict VT
inducibility.22 Nonetheless, we used the method that in our hands has
provided a better correlation compared with the EAM in previous
studies.9 Another limitation of the study is the comparison of the
CCs identified in the EAM and those observed in the LGE-CMR. A
side-by-side comparison may be subjective despite the presence of
a third observer to solve disagreements. Besides, resolution of
EAM may be sometimes not enough to identify all the CC.
Conclusions
3D scar reconstruction using images from 3D-GRE sequence
improves the overall delineation of CCs prior to VT ablation.
Supplementary material
6.
7.
8.
9.
10.
11.
Supplementary material is available at Europace online.
Acknowledgements
We are indebted to the Medical Imaging core facility of the
IDIBAPS for technical assistance. We thank Luis Lasalvia, MD, MIB;
Okan Ekinci MD, MBA and Andreas Greiser, PhD for their great
support and collaboration.
12.
13.
14.
Conflict of interest: David Andreu is employed by Biosense
Webster.
15.
Funding
16.
This study was partially supported by a grant from the ‘Fondo de Investigación Sanitaria’ (FIS 2012-2014, PI11/02049). This study has been supported by Siemens Healthcare and performed in collaboration with the
Clinical Competence Center Cardiology, Erlangen, Germany.
17.
18.
References
1. Bello D, Fieno DS, Kim RJ, Pereles FS, Passman R, Song G et al. Infarct morphology
identifies patients with substrate for sustained ventricular tachycardia. J Am Coll
Cardiol 2005;45:1104 –8.
2. Yan AT, Shayne AJ, Brown KA, Gupta SN, Chan CW, Luu TM et al. Characterization
of the peri-infarct zone by contrast-enhanced cardiac magnetic resonance imaging is
a powerful predictor of post-myocardial infarction mortality. Circulation 2006;114:
32 –9.
3. Schmidt A, Azevedo CF, Cheng A, Gupta SN, Bluemke DA, Foo TK et al. Infarct tissue
heterogeneity by magnetic resonance imaging identifies enhanced cardiac arrhythmia susceptibility in patients with left ventricular dysfunction. Circulation 2007;115:
2006– 14.
4. Ranjan R, McGann CJ, Jeong EK, Hong K, Kholmovski EG, Blauer J et al. Wideband late
gadolinium enhanced magnetic resonance imaging for imaging myocardial scar
without image artefacts induced by implantable cardioverter-defibrillator: a feasibility study at 3 T. Europace 2015;17:483 –8.
5. Alexandre J, Saloux E, Lebon A, Dugué AE, Lemaitre A, Roule V et al. Scar extent as a
predictive factor of ventricular tachycardia cycle length after myocardial infarction:
19.
20.
21.
22.
implications for implantable cardioverter-defibrillator programming optimization.
Europace 2014;16:220 –6.
Berruezo A, Fernández-Armenta J, Mont L, Zeljko H, Andreu D, Herczku C et al.
Combined Endocardial and Epicardial Catheter Ablation in Arrhythmogenic Right
Ventricular Dysplasia Incorporating Scar Dechanneling Technique. Circ Arrhythm
Electrophysiol 2012;5:111–21.
Di Biase L, Santangeli P, Burkhardt DJ, Bai R, Mohanty P, Carbucicchio C et al.
Endo-epicardial homogenization of the scar versus limited substrate ablation for
the treatment of electrical storms in patients with ischemic cardiomyopathy. J Am
Coll Cardiol 2012;60:132 – 41.
Ashikaga H, Sasano T, Dong J, Zviman MM, Evers R, Hopenfeld B et al. Magnetic
resonance-based anatomical analysis of scar-related ventricular tachycardia: implications for catheter ablation. Circ Res 2007;101:939 – 47.
Fernandez-Armenta J, Berruezo A, Andreu D, Camara O, Silva E, Serra L et al. Threedimensional architecture of scar and conducting channels based on high resolution
CE-CMR: insights for ventricular tachycardia ablation. Circ Arrhythm Electrophysiol
2013;6:528 –37.
Andreu D, Berruezo A, Ortiz-Perez JT, Silva E, Mont L, Borras R et al. Integration of
3D electroanatomic maps and magnetic resonance scar characterization into the
navigation system to guide ventricular tachycardia ablation. Circ Arrhythm Electrophysiol 2011;4:674 –83.
Gupta S, Desjardins B, Baman T, Ilg K, Good E, Crawford T et al. Delayed-enhanced
MR scar imaging and intraprocedural registration into an electroanatomical mapping
system in post-infarction patients. JACC Cardiovasc Imaging 2012;5:207 –10.
Simonetti OP, Kim RJ, Fieno DS, Hillenbrand HB, Wu E, Bundy JM et al. An improved
MR imaging technique for the visualization of myocardial infarction. Radiology 2001;
218:215–23.
Sievers B, Elliott MD, Hurwitz LM, Albert TS, Klem I, Rehwald WG et al. Rapid
detection of myocardial infarction by subsecond, free-breathing delayed contrastenhancement cardiovascular magnetic resonance. Circulation 2007;115:236 –44.
Nguyen TD, Spincemaille P, Weinsaft JW, Ho BY, Cham MD, Prince MR et al. A fast
navigator-gated 3D sequence for delayed enhancement MRI of the myocardium:
comparison with breathhold 2D imaging. J Magn Reson Imaging 2008;27:802–8.
Bauner KU, Muehling O, Wintersperger BJ, Winnik E, Reiser MF, Huber A. Inversion
recovery single-shot TurboFLASH for assessment of myocardial infarction at 3 Tesla.
Invest Radiol 2007;42:361 – 71.
Huber A, Bauner K, Wintersperger BJ, Reeder SB, Stadie F, Mueller E et al. Phasesensitive inversion recovery (PSIR) single-shot TrueFISP for assessment of myocardial infarction at 3 tesla. Invest Radiol 2006;41:148 –53.
Bisbal F, Guiu E, Cabanas-Grandio P, Berruezo A, Prat-Gonzalez S, Vidal B et al. MRIguided approach to localize and ablate gaps in repeat AF ablation procedure. JACC
Cardiovasc Imaging 2014;7:653–63.
de Bakker JM, van Capelle FJ, Janse MJ, Wilde AA, Coronel R, Becker AE et al. Reentry
as a cause of ventricular tachycardia in patients with chronic ischemic heart disease:
electrophysiologic and anatomic correlation. Circulation 1988;77:589 –606.
Schelbert EB, Hsu LY, Anderson SA, Mohanty BD, Karim SM, Kellman P et al. Late
gadolinium-enhancement cardiac magnetic resonance identifies postinfarction myocardial fibrosis and the border zone at the near cellular level in ex vivo rat heart. Circ
Cardiovasc Imaging 2010;3:743–52.
Dewey M, Laule M, Taupitz M, Kaufels N, Hamm B, Kivelitz D. Myocardial viability:
assessment with three-dimensional MR imaging in pigs and patients. Radiology
2006;239:703 –9.
Perez-David E, Arenal A, Rubio-Guivernau JL, del Castillo R, Atea L, Arbelo E et al.
Noninvasive identification of ventricular tachycardia-related conducting channels
using contrast-enhanced magnetic resonance imaging in patients with chronic myocardial infarction: comparison of signal intensity scar mapping and endocardial
voltage mapping. J Am Coll Cardiol 2011;57:184 –94.
Rubenstein JC, Lee DC, Wu E, Kadish AH, Passman R, Bello D et al. A comparison of
cardiac magnetic resonance imaging peri-infarct border zone quantification strategies for the prediction of ventricular tachyarrhythmia inducibility. Cardiol J 2013;
20:68 –77.