Download Image Quality of Cardiac Magnetic Resonance Imaging in Patients

Magnetic Resonance Imaging
Image Quality of Cardiac Magnetic Resonance
Imaging in Patients With an Implantable Cardioverter
Defibrillator System Designed for the Magnetic
Resonance Imaging Environment
Juerg Schwitter, MD; Michael R. Gold, MD, PhD; Ahmed Al Fagih, MD; Sung Lee, MD;
Michael Peterson, MD; Allen Ciuffo, MD; Yan Zhang, MS; Nina Kristiansen, PhD;
Emanuel Kanal, MD; Torsten Sommer, MD, PhD; on behalf of the Evera-MRI Study Investigators
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
Background—Recently, magnetic resonance (MR)–conditional implantable cardioverter defibrillator (ICD) systems have
become available. However, associated cardiac MR image (MRI) quality is unknown. The goal was to evaluate the image
quality performance of various cardiac MR sequences in a multicenter trial of patients implanted with an MR-conditional
ICD system.
Methods and Results—The Evera-MRI trial enrolled 275 patients in 42 centers worldwide. There were 263 patients
implanted with an Evera-MRI single- or dual-chamber ICD and randomized to controls (n=88) and MRI (n=175), 156
of whom underwent a protocol-required MRI (9–12 weeks post implant). Steady-state-free-precession (SSFP) and
fast-gradient-echo (FGE) sequences were acquired in short-axis and horizontal long-axis orientations. Qualitative and
quantitative assessment of image quality was performed by using a 7-point scale (grades 1–3: good quality, grades
6–7: nondiagnostic) and measuring ICD- and lead-related artifact size. Good to moderate image quality (grades 1–5)
was obtained in 53% and 74% of SSFP and FGE acquisitions, respectively, covering the left ventricle, and in 69% and
84%, respectively, covering the right ventricle. Odds for better image quality were greater for right ventricle versus left
ventricle (odds ratio, 1.8; 95% confidence interval, 1.5–2.2; P<0.0001) and greater for FGE versus SSFP (odds ratio, 3.5;
95% confidence interval, 2.5–4.8; P<0.0001). Compared with SSFP, ICD-related artifacts on FGE were smaller (141±65
versus 75±57 mm, respectively; P<0.0001). Lead artifacts were much smaller than ICD artifacts (P<0.0001).
Conclusions—FGE yields good to moderate quality in 74% of left ventricle and 84% of right ventricle acquisitions and
performs better than SSFP in patients with an MRI-conditional ICD system. In these patients, cardiac MRI can offer
diagnostic information in most cases.
Clinical Trial Registration—URL: http://www.clinicaltrials.gov. Unique identifier: NCT02117414. (Circ Cardiovasc Imaging. 2016;9:e004025. DOI: 10.1161/CIRCIMAGING.115.004025.)
Key Words: artifacts ◼ heart ventricles ◼ Image Quality Enhancement and Cardiac Function Test
◼ implantable cardioverter-defibrillator ◼ magnetic resonance imaging
C
ardiac magnetic resonance imaging (MRI) is generally
accepted as a valuable diagnostic tool to detect and monitor a large variety of cardiac diseases; it has both excellent
tissue characterization capabilities and high image quality in
most cases. In a large population of >9900 patients cardiac
image quality was shown to be diagnostic in 98.2%.1 Even
in the presence of a pacemaker system, the diagnostic quality
of cardiac MRI (CMR) of the left (LV) and right (RV) ventricles is preserved.2 Thus, the high reliability and robustness
See Editorial by Simonetti and Gross
See Clinical Perspective
of CMR is known for patients implanted with a pacemaker
system. It is estimated that >60% of patients implanted with
an ICD will require an MRI within 10 years post implant,3 and
recently an MR-conditional ICD system was shown to be safe
and effective when scanned at 1.5 Tesla (T).4 In comparison
with pacemakers, the larger size of the ICD pulse generator
Received August 17, 2015; accepted April 4, 2016.
From the Division of Cardiology and Director of the Cardiac Magnetic Resonance Center, University Hospital Lausanne, Switzerland (J.S.); Division
of Cardiology, Medical University of South Carolina, Charleston (M.R.G.); Department of Adult Cardiology, Prince Sultan Cardiac Center, Riyadh, Saudi
Arabia (A.A.F.); Departments of Cardiovascular Disease and Clinical Cardiac Electrophysiology, Washington Hospital Center, Washington, DC (S.L.);
United Heart and Vascular Clinic, Minneapolis, MN (M.P.); Sentara Norfolk General Hospital, VA (A.C.); Cardiac Rhythm and Heart Failure Management,
Medtronic, Minneapolis, MN (Y.Z.); Cardiac Rhythm and Heart Failure Management, Medtronic, Maastricht, The Netherlands (N.K.); Department of
Radiology and Neuroradiology, University of Pittsburgh Medical Center, PA (E.K.); and Department of Diagnostic and Interventional Radiology and
Nuclear Medicine, German Red Cross Hospital, Neuwied, Germany (T.S.).
Correspondence to Juerg Schwitter, MD, Division of Cardiology and Director CMR Center, University Hospital Lausanne (CHUV), Rue du Bugnon 46,
1011 Lausanne, Switzerland. E-mail [email protected]
© 2016 American Heart Association, Inc.
Circ Cardiovasc Imaging is available at http://circimaging.ahajournals.org
1
DOI: 10.1161/CIRCIMAGING.115.004025
2 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
and large defibrillation coil are expected to impact substantially on image quality. However, to our knowledge, no systematic prospective study evaluated to date the image quality
generated by different pulse sequences in patients implanted
with an MR-conditional ICD system. Therefore, this study
was performed to assess the image quality obtainable with 2
types of pulse sequences, ie, with steady-state free precession
(SSFP) and fast-gradient-echo (FGE) acquisitions, which are
the most frequently used for CMR. For this purpose, image
quality was graded, image artifacts caused by the ICD and its
leads were measured, and the percentage of examinations with
diagnostic quality was determined for SSFP and FGE acquisitions in the Evera-MRI trial.
Methods
General Electric, USA; and Toshiba, Japan). The MRI protocol was
primarily set up to test for safety of the MRI-conditional ICD system.
To apply the predefined specific absorption rate to all patients, it was
not allowed to repeat sequences or to use contrast media to optimize
image quality. Of the 10 different MRI sequences applied, the 2 clinically relevant cardiac pulse sequences evaluated in this analysis were
cine SSFP and cine FGE acquisitions (Table 1). The SSFP sequence
is nowadays the most commonly used sequence for cardiac functional
analyses. It was used to test whether this sequence is robust enough
to produce adequate quality in the presence of an ICD system. It was
compared with FGE sequences that are known to be less susceptible to
field inhomogeneities. Both sequences were ECG-triggered and were
acquired in short-axis (SA) and horizontal long-axis (HLA) orientations during breath holding. Before each site’s first patient examination,
a validation of a test MRI scan was performed by an independent Scan
Advisory Board to confirm standardization of the imaging parameters.
Qualitative and Quantitative Assessment
Study Population
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
The Evera-MRI trial was a multicenter, international, randomized
clinical study confirming the safety and efficacy of patients receiving
a novel ICD system and subjected to an MRI examination.4 The study
enrolled 275 subjects in 42 centers worldwide, of which 263 were
randomized to the MRI (n=175) or control group (n=88) at a 2:1 ratio
after a successful implant. In the MRI group, 156 subjects underwent
a predefined not clinically indicated MRI examination scheduled at
9 to 12 weeks post implant. Of them, 152 subjects had scan data collected for assessing cardiac image quality and therefore are included
in this analysis.
The study protocol was approved by the ethics committee at each
participating institution, and all subjects provided written informed
consent before study participation. The Declaration of Helsinki was
followed as well as laws and regulations of the participating countries. Trained center personnel collected the data, and its integrity was
maintained by programmatic edit checks and source data verification
by the sponsor.
Study Device
Consented patients received pectoral implantation of an EveraMRI single- or dual-chamber ICD (MR-ICD; Medtronic) connected to commercially available defibrillator leads (model 6935M
or 6947M) with/without a pacing lead (model 5076) that has been
demonstrated safe for MRI.5 The MR-ICD had specific design and
material modifications to reduce interaction with the MRI environment as described elsewhere.6 Briefly, ferromagnetic material was
reduced, a hall sensor replaced the mechanical reed switch, filters to
prevent gradient and radiofrequency energy coupling were added,
and battery circuitry protection was built in. The Evera-MRI study
protocol included testing of device functionality before and after the
MRI examination, and no influence of the MRI examination on the
device function was found.4
Cardiac MRI
The CMR scans were performed on a 1.5T scanner of any of the 4 most
common manufacturers (Siemens, Germany; Philips, Netherlands;
The CMR data were collected via CD during the 9- to 12-week
postimplant visit and were sent to a core laboratory (German Red
Cross Hospital, Neuwied, Germany, led by T.S.). The assessment
of cardiac image quality was originally done by evaluating the
SSFP and FGE sequences by an experienced CMR reader and was
redone independently by a second reader. Acquisitions with correct
slice orientations on both SA and HLA of at least 1 sequence type
were graded for image quality applying qualitative and quantitative criteria. For qualitative assessment a 7-point grading scale was
used (Table 2), which took into account whether artifacts from the
leads or the ICD impaired the ability to delineate the endocardial
and epicardial borders and to assess ventricular function via the observation of systolic myocardial wall motion and thickening. The
rating criteria were applied for both the SSFP and the FGE images
and for both ventricles. Images with incorrect acquisition of standard cardiac planes, severe wrap-around artifacts obscuring parts
of the LV or RV, or severe image blurring because of poor breath
holding or poor ECG-triggering were not included in the analysis
because these quality deficits are purely operator/patient dependent
(and not ICD related) and would be fixed in clinical practice by
repeating such acquisitions. Repeated acquisitions, however, were
not allowed in the trial to ensure that all patients were exposed to
the correct predefined specific absorption rate. For quantitative assessment of the size of ICD system–related artifacts, the maximum
diameter of the artifacts, ie, the signal void, susceptibility artifacts,
image distortion, and SSFP off-resonance banding artifacts of the
ICD and the leads, was measured. The window and level were established for best visualization of the endocardial borders of the
ventricles as typically done for reading and analysis of such CMR
data and without changing these settings the maximum diameter of
the artifact/signal void was measured (Figure 1).
Example Pulse Sequences in an ICD Phantom
An ICD (Evera-MRI XT DR SureScan connected to 6935M and 5076)
was positioned in water at ≈5 cm from a heart phantom representing a
SA view of the LV. Standard spin-echo, SSFP, FGE, compressed sensing accelerated SSFP and FGE,7 3-dimensional (3D) self-navigating
SSFP,8 and a silent zero-echo-time sequence9 were applied.
Table 1. Pulse Sequences and Parameters Applied in the Evera-Magnetic Resonance Imaging Trial
Sequence Name
FOV, mm
Matrix x
Matrix y
Flip Angle
Breath hold
Comment
Cine SSFP: short axis, 3 slices
320–400
168–224
144–224
50–63
Yes (<20 s)
No fat sat, no flow-comp
Cine FGE: short axis, 3 slices
300–360
192–256
126–180
15–30
Yes (<20 s)
No fat sat
Cine SSFP: horizontal long axis
320–360
168–256
160–256
50–55
Yes (<20 s)
No fat sat, flow-comp
Cine FGE: horizontal long axis
300–360
192–256
126–160
15–30
Yes (<20 s)
No fat sat
All acquisitions were performed during breath-holding (duration 10–20 s/slice). Slice thickness was 8 to 10 mm for all acquisitions. Repetition time and
echo time were set to minimum for all acquisitions. FGE indicates fast-gradient-echo; FOV, field-of-view; and SSFP, steady-state-free-precession.
3 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
Table 2. Image Quality Grading Criteria
Left Ventricle
Right Ventricle
Regional Function is Well
Depicted and Evaluable
in n of 13 Segments
Artifacts Interfering With
Endo- or Epicardial Borders
in n of 13 Segments
Regional Function is Well
Depicted and Evaluable
in n of 4 Segments
Artifacts Interfering With
Endo- or Epicardial Borders
in n of 4 Segments
Grade 1
13/13
0
4/4
0
Grade 2
13/13
≤3
4/4
≤2
Grade 3
13/13
>3
4/4
>2
Grade 4
10–12/13
…
3/4
…
Grade 5
7–9/13
…
2/4
…
Grade 6
4–6/13
…
1/4
…
Grade 7
≤3/13
…
0/4
…
Rating
Good quality
Moderate quality
Poor quality (ie, nondiagnostic)
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
Grading of steady-state free precession and fast-gradient-echo short-axis and long-axis 4-chamber view images.
Statistical Analysis
For each subject, the images from 2 sequences (SSFP or FGE) in 2
orientations (SA or HLA) and of the 2 ventricles (LV or RV) were
evaluated. An ordinal multinomial logistic regression model with
generalized estimating equation method was used to analyze the image quality score (grades 1–7) as a correlated ordinal categorical
variable, and the effects of sequence type and ventricle type and
their interaction on the score were considered. The odds ratios of
having a better image quality for different sequences and ventricles
were reported using a forest plot. Additional analyses included (1)
considering age, sex, and body mass index (BMI) in the abovementioned model and (2) re-classifying grades into good (grades
1–3), moderate (grades 4–5), and poor (grades 6–7; nondiagnostic)
and analyzing this correlated ordinal categorical variable using an
ordinal multinomial logistic regression model with generalized estimating equation.
For each subject, the presence and size of lead- or ICD-related artifacts were evaluated for the 2 orientations (SA and HLA) and the 2
sequences (SSFP and FGE). If no artifact was observed, its diameter
was given as 0 mm. A general linear mixed model was used to evaluate the effects of artifact type (lead or ICD related), orientation, and
pulse sequence and their interactions on the largest diameters of the
artifacts. An additional analysis considered age, sex, and BMI using
a linear mixed model.
The above-mentioned analyses were based on the assessment of
the original reader. In addition, the agreement on image quality scores
Figure 1. Representative cardiac images
from a study patient implanted with an
Evera-Magnetic Resonance Imaging (MRI)
implantable cardioverter defibrillator (ICD)
system. Four-chamber horizontal long-axis
(HLA) and short-axis (SA) acquisitions with
the fast-gradient-echo (FGE) and steadystate-free-precession (SSFP) sequences.
The FGE images were graded 1 for both
ventricles, and the SSFP images were
graded 3 and 2 for the left and right ventricle, respectively.
4 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
FGE versus SSFP. FGE yielded good quality in 34.4% and
50.6% for the LV and RV, respectively. If grade 4 or 5 is considered as acceptable quality (1–3 or 4–6 segments not evaluable, respectively), FGE yielded 58.9% and 74.4% successful
studies for the LV, respectively, and 67.5% and 84.4% for the
RV, respectively. The forest plot of Figure 3 shows higher
odds of having a better image quality with the FGE sequence
than with the SSFP, regardless of ventricle type (odds ratio,
3.48; 95% confidence interval, 2.52–4.80; P<0.0001); and the
odds for better image quality was greater for RV versus LV,
regardless of pulse sequence type (odds ratio, 1.78; 95% confidence interval, 1.46–2.16; P<0.0001). Furthermore, the performances of the pulse sequences were not related to the type
of ventricle (no significant interaction between type of pulse
sequence and type of ventricle, P=0.2752). Men were more
likely to present a better image quality than women (odds
ratio, 2.63; 95% confidence interval, 1.23–5.63; P=0.0147),
while age and BMI did not correlate with image quality
(P>0.05). Re-classifying the quality grades into good (grades
1–3), moderate (grades 4–5), and poor (grades 6–7; ie, nondiagnostic) led to similar results with a greater chance of obtaining better image quality with FGE versus SSFP (P<0.0001)
and for RV versus LV (P<0.0001).
between the original reader and the second reader was estimated using Fleiss κ statistics for each sequence by orientation by ventricle
combination.10 The method of Landis and Koch11 was used to interpret the κ values.
All analyses were performed using SAS version 9.4 (SAS Institute,
Cary, NC). A P<0.05 was considered statistically significant.
Results
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
This study included 152 subjects from the MRI group whose
image files for predefined MRI examination were collected
(Figure 2). The mean age was 59.7±13.8 years, BMI was
28.9±5.9 kg/m2, 21.7% (33/152) were women, and 11.2%
(17/152) of patients were paced during the MRI examination
(in 4.0% the pacing status was not known). Of the 152 subjects, 130 had uncorrupted image files that could be assessed,
ie, at least 1 pulse sequence type (SSFP or FGE) could be analyzed, and 105, 90, 104, and 77 subjects had assessable data at
both SA and HLA orientations for the LV SSFP, LV FGE, RV
SSFP, and RV FGE pulse sequence, respectively (Figure 2).
Image Quality and Type of Pulse Sequences for the
LV and RV
Table 3 presents the quality grading distribution in the study
population with higher percentages of diagnostic quality for
Randomized to MRI group and
underwent protocol-required
MRI examination (n = 156)
MRI scan files not available (n=4)
MRI scan files collected (n = 152)
MRI scan files corrupted (n=22)
MRI scan files not corrupted (n =130)
Sequence
not
analyzed:
(n = 9)
Sequence
not
analyzed:
(n = 16)
Sequence
not
analyzed:
(n = 24)
Sequence
not
analyzed:
(n = 25)
Sequence
not
analyzed:
(n = 10)
Sequence
not
analyzed:
(n = 18)
Sequence
not
analyzed:
(n = 38)
Sequence
not
analyzed:
(n = 29)
Sequence
not done:
(n = 4)
Sequence
not done:
(n = 4)
Sequence
not done:
(n = 6)
Sequence
not done:
(n = 5)
Sequence
not done:
(n = 4)
Sequence
not done:
(n = 4)
Sequence
not done:
(n = 6)
Sequence
not done:
(n = 5)
LV SSFP
SA graded
(n = 117)
77.0%*
LV SSFP
HLA graded
(n = 110)
72.4%*
LV SSFP
SA+HLA
graded
(n = 105)
LV FGE
SA graded
(n = 100)
65.8%*
LV FGE
HLA graded
(n = 100)
65.8%*
LV FGE
SA+HLA
graded
(n = 90)
RV SSFP
SA graded
(n = 116)
76.3%*
RV SSFP
HLA graded
(n = 108)
71.1%*
RV SSFP
SA+HLA
graded
(n = 104)
RV FGE
SA graded
(n = 86)
56.6%*
RV FGE
HLA graded
(n = 96)
63.2%*
RV FGE
SA+HLA
graded
(n = 77)
Figure 2. Flowchart of study participants. Acquisitions were graded if artifacts were related to the implantable cardioverter defibrillator (ICD) system itself (ie, signal void, susceptibility artifacts, image distortion, steady-state-free-precession [SSFP]–related off-resonance banding artifacts,
SSFP-related dark flow artifacts). Artifacts not related to the ICD system were not graded including those because of respiratory motion (incorrect breath-holding), ECG-mistriggering, incorrect field-of-view (causing wrap-around artifacts), or incorrect image plane orientations. These not
graded artifacts are typically corrected in clinical MRI examinations by repeated acquisitions (eventually with individual modifications of scan
parameters), which were not allowed in the study to achieve predefined specific absorption rate exposition in all patients. *Percentage calculated using n=152 as the denominator. FGE indicates fast-gradient-echo; HLA, horizontal long-axis; LV, left ventricle; and RV, right ventricle.
5 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
Table 3. Image Quality: Influence of Pulse Sequence Type
Left Ventricle
Right Ventricle
SSFP Cine
Sequences
FGE Cine
Sequences
Not graded
25
40
26
53
Graded (=100%)
105
90
104
77
Good IQ (grade
1–3),* n (%)
11 (10.5)
31 (34.4)
16 (15.4)
39 (50.6)
Image quality
grade 1
1 (1.0)
4 (4.4)
1 (1.0)
4 (5.2)
Image quality
grade 2
4 (3.8)
17 (18.9)
9 (8.7)
16 (20.8)
Image quality
grade 3
6 (5.7)
10 (11.1)
6 (5.8)
19 (24.7)
Moderate IQ
(grade 4–5),*
n (%)
45 (42.9)
36 (40.0)
56 (53.8)
26 (33.8)
Image quality
grade 4
10 (9.5)
22 (24.4)
27 (26.0)
13 (16.9)
Image quality
grade 5
35 (33.3)
14 (15.6)
29 (27.9)
13 (16.9)
Poor IQ (grade 6–7;
nondiagnostic),*
n (%)
49 (46.7)
23 (25.6)
32 (30.8)
12 (15.6)
Image quality
grade 6
21 (20.0)
9 (10.0)
14 (13.5)
4 (5.2)
Image quality
grade 7
28 (26.7)
14 (15.6)
18 (17.3)
8 (10.4)
Rating
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
Image file corrupted
SSFP Cine FGE Cine
Sequences Sequences
n=22
FGE indicates fast-gradient-echo; IQ, image quality; and SSFP, steady-statefree-precession.
*Better quality was more likely for the FGE sequence than for the SSFP sequence
(P<0.0001) and for the right ventricle than for the left ventricle (P<0.0001).
As shown in Table 4, for the 2 readers, substantial to
almost perfect agreement (κ=0.6–0.8 and κ>0.8, respectively) was observed for almost all acquisition types. Only
for 2 acquisitions (FGE of the LV and SSPF of the RV) at
the moderate quality level, the agreement was moderate
(κ=0.4–0.6).
Size of Artifacts of the ICD and the Leads and Type
of Pulse Sequences
Lead- and ICD-related artifacts were observed among most of
the subjects, except for the FGE sequence in the HLA plane
for which no ICD-related artifacts were present in 49.5% of
patients (Table 5).
The analysis of artifact size and type (ie, ICD related
and lead related) based on a general linear mixed model
showed significant interaction effects for artifact type with
pulse sequence and for artifact type with slice orientation.
Specifically, ICD-related artifacts on the FGE sequence
were smaller than those on the SSFP sequence (Figure 4A;
75.0±56.7 versus 140.9±65.2 mm; P<0.0001). Also, ICDrelated artifacts on the HLA orientation were smaller than
those on the SA orientation (Figure 4B; 78.8±69.6 versus
139.4±54.6 mm; P<0.0001). Lead-related artifacts were
smaller than ICD-related artifacts irrespective of the pulse
sequence type (Figure 4A; P<0.0001) or orientation of acquisition (Figure 4B; P<0.0001). The difference in the leadrelated artifact diameters between the FGE and the SSFP
sequence was small from a clinical point of view, but it was
statistically significant (Figure 4A; 10.2±3.2 versus 8.3±3.3
mm; P<0.0001). The difference in lead-related artifact diameters between HLA and SA orientations was even smaller
(Figure 4B; 8.7±3.7 versus 9.8±2.9 mm; P<0.0001). Age, sex,
and BMI were not associated with artifact size (P>0.05).
Examples of Pulse Sequences Not Used in the
Clinical Trial
In the phantom example in Figure 5, 2D-SSFP–based (Figure 5B) and 3D-SSFP–based (Figure 5D) sequences show
substantially larger artifacts than the FGE-based sequences
(Figure 5C and 5G). A novel 3D-zero-echo-time sequence (Figure 5H) sequence shows small ICD-related artifacts in comparison with the SSFP and FGE sequences used in the trial. In a
Figure 3. Odds ratio (OR) of better image
quality grades by pulse sequence type and
type of ventricle. For fast-gradient-echo
(FGE) higher ORs are observed than for
the steady-state-free-precession (SSFP)
sequence for both, the left (LV) and right
(RV) ventricles (OR, 3.48; 95% confidence
interval [CI], 2.52–4.80; P<0.0001). For RV
acquisitions, higher ORs are observed vs LV
acquisitions (OR, 1.78; 95% CI, 1.46–2.16;
P<0.0001) irrespective of pulse sequence
type. FGE indicates fast-gradient-echo;
LCL, lower confidence limit; SSFP, steadystate-free-precession; and UCL, upper confidence limit.
6 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
Table 4. Agreement Between 2 Readers: κ Statistics
κ Statistics
Left Ventricle
Right Ventricle
SSFP Cine Sequences
FGE Cine Sequences
SSFP Cine Sequences
FGE Cine Sequences
Could not be graded
0.81
0.74
0.80
0.89
Good IQ (grade 1–3)
0.83
0.70
0.76
0.81
Moderate IQ (grade 4–5)
0.70
0.50
0.56
0.67
Poor IQ (grade 6–7)
0.75
0.80
0.70
0.82
Overall (n=130)
0.76
0.69
0.68
0.81
Rating
Degree of agreement: κ≤0: poor; >0 to 0.2: slight; >0.2 to 0.4: fair; >0.4 to 0.6: moderate; >0.6 to 0.8: substantial; >0.80: almost perfect. FGE indicates fastgradient-echo; IQ, image quality; and SSFP, steady-state-free-precession.
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
patient example (Figure 6), also highly accelerated compressed
sensing acquisitions (Figure 5G and 5H) show potential to yield
adequate quality images. Moreover, the FGE-based viability
sequence postcontrast (Figure 6I) clearly delineates scar.
Discussion
This study is the first evaluating in a large multicenter setting
the diagnostic performance of SSFP- and FGE-based pulse
sequences in patients implanted with an MR-conditional ICD
system. It demonstrates that good to moderate image quality
is obtainable with FGE in 74% of LV and 84% of RV acquisitions and it yields substantially higher image quality than
SSFP acquisitions. Moreover, the ICD generator is the relevant
system component that can potentially degrade image quality,
and this is dependent on the pulse sequence type, while the
leads are of minor importance with regard to artifacts.
Quality of CMR Images: Influence of Pulse
Sequence Type and Orientation of Acquisition
In a single-center study, Sasaki et al12 found diagnostic quality with SSFP in 85.5% of cases (by using the criterion of
more than half of LV segments being distorted to define nondiagnostic image quality as in the current study), whereas the
SSFP sequence in the present study yielded diagnostic quality
in only 53% and 69% of the RV and LV, respectively. This
considerably lower performance of the SSFP sequence in the
current study may be, in part, because of the inclusion of 42
centers, whereas the study of Sasaki et al12 included 56 scans
in a single center.12 Nevertheless, when using the SSFP cine
sequence, the single-center study reported a mean artifact
diameter for the ICD device on SA images of 16.0 cm, which
is comparable with the size found in the current study of 16.7
cm (Table 5). In another small cohort of patients with metal
implants of various origin, a better image quality of CMR was
found for FGE acquisitions in comparison with SSFP.13 These
findings and the results of the current study are in agreement
with the fact that the SSFP sequence is more susceptible to
field inhomogeneities than the FGE sequence. Importantly,
many pulse sequences used in clinical MRI practice for
viability/scar imaging,14 myocardial perfusion imaging,15–18
tagging,19 and flow measurements20 are using FGE-type readouts. Also, highly accelerated cardiac acquisitions based on
novel compressed sensing strategies can use the FGE read-out
scheme,7 and thus are expected to work reliably in patients
with this implanted ICD system as shown in the example in
Figures 5 and 6.
Table 5. Presence and Size of Artifacts
Lead Related
ICD Related
SSFP SA
SSFP HLA
FGE SA
FGE HLA
SSFP SA
SSFP HLA
FGE SA
FGE HLA
Analyzed
118
114
103
108
118
115
107
111
Artifact not
present, n (%)
0 (0)
9 (7.9)
0 (0)
5 (4.6)
4 (3.4)
18 (15.7)
6 (5.6)
55 (49.5)
Artifact present,
n (%)
118 (100)
105 (92.1)
103 (100)
103 (95.4)
114 (96.6)
97 (84.3)
101 (94.4)
56 (50.5)
Mean±SD, mm
9.1±3.2
7.6±3.3
10.6±2.4
9.9±3.7
166.9±50.7
114.3±67.8
109.1±41.3
42.1±49.7
8.0
7.0
11.0
10.0
170.0
125.0
113.0
12.0
5.0–25.0
0.0–20.0
5.0–17.0
0.0–30.0
0.0–350.0
0.0–300.0
0.0–205.0
0.0–165.0
12
16
27
22
12
15
23
19
Median, mm
Min–max, mm
Could not be
analyzed*
Image file corrupted
n=22
*Images with incorrect acquisition of standard cardiac planes, severe wrap-around artifacts obscuring parts of the left or right (RV) ventricles or severe image blurring
caused by poor breath-holding or poor ECG-triggering were considered nonanalyzable. FGE indicates fast-gradient-echo; HLA, horizontal long-axis; ICD, implantable
cardioverter defibrillator; SA, short-axis; and SSFP, steady-state-free-precession.
7 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
Figure 4. Artifacts by sequence type and
orientation of acquisition. Influence of pulse
sequence type (A) and slice orientation (B)
on artifact size (mean of largest diameter)
caused by the implantable cardioverter
defibrillator (ICD; blue) and leads (brown).
ICD-related artifacts associated with fastgradient-echo (FGE) are smaller (P<0.0001,
A) than with steady-state-free-precession
(SSFP). ICD-related artifacts on horizontal
long-axis (HLA) acquisitions are smaller
(P<0.0001, B) than on short-axis (SA). The
ICD component (blue line) is causing larger
artifacts (P<0.0001) than the leads (brown
line). Data are given as arithmetic means and
SD; P values are derived from the general linear mixed model that takes into account the
effects of artifact type (lead or ICD related),
orientation (HLA or SA), and pulse sequence
(FGE or SSFP).
The diagnostic performance in the patients with an
MR-conditional ICD system of 53% and 69% for the LV and
RV, respectively, reported in this study is also considerably
lower than that observed for an MR-conditional pacemaker
evaluated in a multicenter trial and ranging from 95% of diagnostic studies for the LV to 98% for the RV.2 This difference is
most likely explained by the amount of ferromagnetic materials contained in the ICD.6
It is noteworthy that no contrast media was applied (in
order not to confound the safety analysis of the study). It might
be recommended to improve the diagnostic performance of
FGE by administering contrast media to increase the endocardial border delineation on FGE-based images, particularly for
long-axis acquisitions.
Based on this large trial performed at 42 centers, the FGE
is the pulse sequence of choice for patients with an implanted
ICD system and the HLA acquisition yielded even higher
quality than SA acquisitions.
Image Artifacts: Influence of ICD System
Components
The study clearly identifies the ICD generator as the relevant
component of the ICD system to cause imaging artifacts.
The fact that the ICD is the limiting factor on artifacts also
explains why image quality was higher for the RV than for
the LV, as the ICD is located closer to the LV than the RV
as shown in Figures 1 and 6. Consequently, in ICD patients
who have insufficient echocardiographic quality, it seems reasonable to undergo a CMR, especially, if RV pathologies are
suspected. Although the leads are particularly important with
regard to side effects such as heating, they are not contributing
much to image artifacts, and thus, they are generally not compromising image quality. This is in agreement with a recent
study on image quality of MR-conditional pacemakers, where
a high diagnostic quality was documented.2
The spectrum of artifacts observed in the current study
can be divided into those related to patients’ collaboration and
8 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
Figure 5. Pulse sequences in an ICD phantom. Standard spin-echo (TSE; A) shows high image quality with almost no artifacts on the
heart phantom (T1/T2 values representing blood, fat, and myocardium from center out). For comparison, an steady-state-free-precession
(SSFP) image (E) is shown from a patient implanted with an Evera-Magnetic Resonance Imaging implantable cardioverter defibrillator
(ICD) used in the trial. Considerable off-resonance artifacts of SSFP degrade the visualization of the patient’s heart (E) and the heart phantom (B/F). ICD-related artifacts are smaller with fast-gradient-echo (FGE) (C and G). The compressed-sensing versions cs-SSFP (F) and
cs-FGE show the same artifact behavior as the standard nonaccelerated versions. 3-dimensional (3D)-SSFP represents a self-navigating
free-breathing 3D version of SSFP (D). A silent 3D zero-echo-time (3D-ZTE; H) sequence shows great potential to reduce ICD-related artifacts. The small lead-related artifacts are indicated by arrows.
anatomy (inadequate breath-holding, wrap-around artifacts,
ECG-mistriggering) and those related to the device, ie, the
ICD generator (susceptibility artifacts, signal voids, SSFPrelated off-resonance banding artifacts, and SSFP-related dark
flow artifacts). It is highly desirable to develop and design new
pulse sequences that are dedicated to image the heart in the
presence of ICD systems, ie, with optimal robustness against
field distortions. New pulse sequences are available to image
with less artifacts near metallic implants.21 However, these
novel sequences are often spin-echo–based and as such they
Figure 6. Representative cardiac images from a patient implanted with an Evera-Magnetic Resonance Imaging system. Diastolic (A,
D, and G) and systolic (B, E, and H) short-axis images with steady-state-free-precession (SSFP) and fast-gradient-echo (FGE) pulse
sequences. Off-resonance artifacts are visible on the SSFP acquisitions, which are inexistent on the FGE acquisitions. The highly accelerated FGE version (cs-FGE, G and H, acquisition duration is 2 heart beats) shows good image quality. C and F, Arrows point to the pacer
(PM) and defibrillator (Def) leads in the atrium (C) and in the right ventricle (F). Inversion-recovery FGE (IR-FGE) viability sequence demonstrates a transmural scar in the inferolateral wall of the left ventricle (I, arrows). This scar tissue is also detected by cs-FGE, which was
acquired after contrast medium injection (H, arrows).
9 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
are not time efficient and not well suited to provide functional
information of the heart. As shown in the phantom example in
Figure 5, alternative types of pulse sequences that collect data
during the free induction decay of the signal after excitation,
the so-called ultrashort or zero-echo-time techniques show the
potential for a robust acquisition in inhomogeneous magnetic
fields.22 They could present a novel approach to yield functional images of the heart in the presence of devices.23 Also
novel shimming procedures could reduce the occurrence of
such device-related artifacts. Future research on novel dedicated pulse sequences and shimming procedures are needed
and with the advent of MRI-conditional ICDs such approaches
can now be tested in patients.
Limitations
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
This study should be interpreted in light of certain limitations.
The higher robustness of the FGE sequence in comparison with
SSFP comes at the cost of reduced signal-to-noise in the blood
pool, and thus, discrimination of the blood–myocardium interface is typically lower than with SSFP-type pulse sequences.24
Accordingly, administration of contrast media would be of value
for the FGE sequence,24 but was not tested. The performance
of machines of different vendors was not evaluated as this was
not within the scope of the study. Finally, image quality was an
ancillary objective of the study, and therefore, it was not allowed
to repeat acquisitions. As inadequately acquired series (because
of imperfect breath-holding, wrap-around artifacts, or incorrect
image plane) are often repeated in clinical routine CMR examinations, an underestimation of achievable diagnostic image quality in the studied ICD-implanted population cannot be excluded.
Finally, the order of SSFP and FGE acquisitions was not randomized, and FGE acquisitions were acquired after the SSFP
acquisitions by protocol. This fact might explain the trend for a
higher drop-out rate for the FGE sequence because suboptimal
patient collaborations occur more often toward the end of MRI
examinations and related artifacts were an exclusion criterion.
Conclusions
FGE produces better quality and smaller ICD-related artifacts
for CMR than SSFP in patients with an MRI-conditional ICD
system. FGE yields good to moderate image quality in this
multicenter setting in 74% of LV and 84% of RV acquisitions.
In these patients implanted with ICD systems designed for the
MR environment, CMR can offer diagnostic information in
most cases.
Acknowledgments
We are grateful to Jean Delacoste, MSc, Department of Radiology,
University Hospital Lausanne, Switzerland, for performing the phantom acquisitions.
Sources of Funding
The Evera-MRI Study was sponsored by Medtronic, Minneapolis,
MN.
Disclosures
Dr Gold is a consultant to and receives clinical trials funds from
Medtronic, Boston Scientific, and St. Jude Medical. Dr Kanal is a
consultant to Medtronic, Boston Scientific, and St. Jude Medical.
Dr Ciuffo is a consultant to Medtronic and Boston Scientific. Dr
Schwitter and Dr Sommer are consultants to Medtronic. Drs Peterson,
Lee, and Fagih are study Investigators. Y. Zhang and Dr Kristiansen
are employees of Medtronic.
References
1. Bruder O, Schneider S, Nothnagel D, Dill T, Hombach V, Schulz-Menger
J, Nagel E, Lombardi M, van Rossum AC, Wagner A, Schwitter J, Senges J,
Sabin GV, Sechtem U, Mahrholdt H. EuroCMR (European Cardiovascular
Magnetic Resonance) registry: results of the German pilot phase. J Am
Coll Cardiol. 2009;54:1457–1466. doi: 10.1016/j.jacc.2009.07.003.
2. Schwitter J, Kanal E, Schmitt M, Anselme F, Albert T, Hayes DL, Bello
D, Tóth A, Chang Y, van Osch D, Sommer T; Advisa MRI System Study
Investigators. Impact of the Advisa MRI pacing system on the diagnostic
quality of cardiac MR images and contraction patterns of cardiac muscle
during scans: Advisa MRI randomized clinical multicenter study results.
Heart Rhythm. 2013;10:864–872. doi: 10.1016/j.hrthm.2013.02.019.
3. Nazarian S, Reynolds MR, Ryan MP, Wolff SD, Mollenkopf SA, Turakhia
MP. Utilization and likelihood of radiologic diagnostic imaging in patients with implantable cardiac defibrillators. J Magn Reson Imaging.
2016;43:115–127. doi: 10.1002/jmri.24971.
4. Gold MR, Sommer T, Schwitter J, Al Fagih A, Albert T, Merkely B,
Peterson M, Ciuffo A, Lee S, Landborg L, Cerkvenik J, Kanal E; Evera
MRI Study Investigators. Full-body MRI in patients with an implantable
cardioverter-defibrillator: primary results of a randomized study. J Am
Coll Cardiol. 2015;65:2581–2588. doi: 10.1016/j.jacc.2015.04.047.
5. Shenthar J, Milasinovic G, Al Fagih A, Götte M, Engel G, Wolff S, Tse
HF, Herr J, Carrithers J, Cerkvenik J, Nähle CP. MRI scanning in patients
with new and existing CapSureFix Novus 5076 pacemaker leads: randomized trial results. Heart Rhythm. 2015;12:759–765. doi: 10.1016/j.
hrthm.2014.12.035.
6.Gold MR, Kanal E, Schwitter J, Sommer T, Yoon H, Ellingson M,
Landborg L, Bratten T. Preclinical evaluation of implantable cardioverter-defibrillator developed for magnetic resonance imaging use. Heart
Rhythm. 2015;12:631–638. doi: 10.1016/j.hrthm.2014.12.012.
7. Vincenti G, Monney P, Chaptinel J, Rutz T, Coppo S, Zenge MO, Schmidt
M, Nadar MS, Piccini D, Chèvre P, Stuber M, Schwitter J. Compressed
sensing single-breath-hold CMR for fast quantification of LV function,
volumes, and mass. JACC Cardiovasc Imaging. 2014;7:882–892. doi:
10.1016/j.jcmg.2014.04.016.
8. Monney P, Piccini D, Rutz T, Vincenti G, Coppo S, Koestner SC, Sekarski
N, Di Bernardo S, Bouchardy J, Stuber M, Schwitter J. Single centre experience of the application of self navigated 3D whole heart cardiovascular
magnetic resonance for the assessment of cardiac anatomy in congenital heart disease. J Cardiovasc Magn Reson. 2015;17:55. doi: 10.1186/
s12968-015-0156-7.
9.Dournes G, Grodzki D, Macey J, Girodet PO, Fayon M, Chateil JF,
Montaudon M, Berger P, Laurent F. Quiet submillimeter MR imaging of the lung is feasible with a PETRA sequence at 1.5 T. Radiology.
2015;276:258–265. doi: 10.1148/radiol.15141655.
10. Fleiss JL. Statistical Methods for Rates and Proportions: New York: John
Wiley; 1981:38–46.
11. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174.
12. Sasaki T, Hansford R, Zviman MM, Kolandaivelu A, Bluemke DA, Berger
RD, Calkins H, Halperin HR, Nazarian S. Quantitative assessment of artifacts on cardiac magnetic resonance imaging of patients with pacemakers and implantable cardioverter-defibrillators. Circ Cardiovasc Imaging.
2011;4:662–670. doi: 10.1161/CIRCIMAGING.111.965764.
13. Olivieri LJ, Cross RR, O’Brien KE, Ratnayaka K, Hansen MS. Optimized
protocols for cardiac magnetic resonance imaging in patients with thoracic metallic implants. Pediatr Radiol. 2015;45:1455–1464. doi: 10.1007/
s00247-015-3366-0.
14. Kim RJ, Wu E, Rafael A, Chen EL, Parker MA, Simonetti O, Klocke FJ,
Bonow RO, Judd RM. The use of contrast-enhanced magnetic resonance
imaging to identify reversible myocardial dysfunction. N Engl J Med.
2000;343:1445–1453. doi: 10.1056/NEJM200011163432003.
15. Coelho-Filho OR, Seabra LF, Mongeon FP, Abdullah SM, Francis SA,
Blankstein R, Di Carli MF, Jerosch-Herold M, Kwong RY. Stress myocardial perfusion imaging by CMR provides strong prognostic value to
cardiac events regardless of patient’s sex. JACC Cardiovasc Imaging.
2011;4:850–861. doi: 10.1016/j.jcmg.2011.04.015.
10 Schwitter et al Quality of Cardiac MR Imaging in ICD Patients
16. Schwitter J, Arai AE. Assessment of cardiac ischaemia and viability: role
of cardiovascular magnetic resonance. Eur Heart J. 2011;32:799–809.
doi: 10.1093/eurheartj/ehq481.
17. Jogiya R, Kozerke S, Morton G, De Silva K, Redwood S, Perera D, Nagel
E, Plein S. Validation of dynamic 3-dimensional whole heart magnetic
resonance myocardial perfusion imaging against fractional flow reserve
for the detection of significant coronary artery disease. J Am Coll Cardiol.
2012;60:756–765. doi: 10.1016/j.jacc.2012.02.075.
18.
Schwitter J, Wacker CM, Wilke N, Al-Saadi N, Sauer E,
Huettle K, Schönberg SO, Debl K, Strohm O, Ahlstrom H, Dill T, Hoebel
N, Simor T; MR-IMPACT Investigators. Superior diagnostic performance
of perfusion-cardiovascular magnetic resonance versus SPECT to detect
coronary artery disease: The secondary endpoints of the multicenter multivendor MR-IMPACT II (Magnetic Resonance Imaging for Myocardial
Perfusion Assessment in Coronary Artery Disease Trial). J Cardiovasc
Magn Reson. 2012;14:61. doi: 10.1186/1532-429X-14-61.
19. Rutz AK, Manka R, Kozerke S, Roas S, Boesiger P, Schwitter J. Left
ventricular dyssynchrony in patients with left bundle branch block and
patients after myocardial infarction: integration of mechanics and viability by cardiac magnetic resonance. Eur Heart J. 2009;30:2117–2127. doi:
10.1093/eurheartj/ehp212.
20. Gatehouse PD, Rolf MP, Graves MJ, Hofman MB, Totman J, Werner B,
Quest RA, Liu Y, von Spiczak J, Dieringer M, Firmin DN, van Rossum A,
Lombardi M, Schwitter J, Schulz-Menger J, Kilner PJ. Flow measurement
by cardiovascular magnetic resonance: a multi-centre multi-vendor study
of background phase offset errors that can compromise the accuracy of derived regurgitant or shunt flow measurements. J Cardiovasc Magn Reson.
2010;12:5. doi: 10.1186/1532-429X-12-5.
21. Butts K, Pauly JM, Gold GE. Reduction of blurring in view angle tilting
MRI. Magn Reson Med. 2005;53:418–424. doi: 10.1002/mrm.20375.
22. Carl M, Koch K, Du J. MR imaging near metal with undersampled 3D
radial UTE-MAVRIC sequences. Magn Reson Med. 2013;69:27–36. doi:
10.1002/mrm.24219.
23. Motaal AG, Noorman N, de Graaf WL, Hoerr V, Florack LM, Nicolay
K, Strijkers GJ. Functional imaging of murine hearts using accelerated
self-gated UTE cine MRI. Int J Cardiovasc Imaging. 2015;31:83–94. doi:
10.1007/s10554-014-0531-8.
24.Thiele H, Nagel E, Paetsch I, Schnackenburg B, Bornstedt A,
Kouwenhoven M, Wahl A, Schuler G, Fleck E. Functional cardiac MR
imaging with steady-state free precession (SSFP) significantly improves
endocardial border delineation without contrast agents. J Magn Reson
Imaging. 2001;14:362–367.
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
CLINICAL PERSPECTIVE
Magnetic resonance imaging (MRI) of the heart is increasing in importance, particularly for the evaluation of heart failure
and coronary artery disease. MRI-conditional pacemakers and, more recently, implantable cardioverter defibrillator (ICD)
systems, are available. Imaging at sites remote from the pacemaker/ICD is rarely a problem. However, little is known about
cardiac imaging. This is of particular concern for ICDs given the large pulse generator and more complex leads, including
metallic coils. The Evera-MRI trial included 175 patients who were randomized to undergo a protocol-required MRI at 9 to
12 weeks post implant. Steady-state-free-precession and fast-gradient-echo sequences were acquired, and image quality was
assessed qualitatively and quantitatively (ICD- and lead-related artifact size). Good to moderate image quality was obtained
in 74% and 84% of fast-gradient-echo acquisitions covering the left and right ventricle, respectively. ICD-related artifacts
were much larger than lead artifacts on both steady-state-free-precession and fast-gradient-echo. These results indicate that
fast-gradient-echo is currently the pulse sequence of choice in ICD-implanted patients, yielding good to moderate quality in
the majority of examinations. Nevertheless, these results also indicate that there is a need for further development of pulse
sequences and novel shimming procedures dedicated to minimize ICD-related artifacts. With the advent of MRI-conditional
ICDs, new strategies can now be tested in ICD-implanted patients.
Downloaded from http://circimaging.ahajournals.org/ by guest on May 2, 2017
Image Quality of Cardiac Magnetic Resonance Imaging in Patients With an Implantable
Cardioverter Defibrillator System Designed for the Magnetic Resonance Imaging
Environment
Juerg Schwitter, Michael R. Gold, Ahmed Al Fagih, Sung Lee, Michael Peterson, Allen Ciuffo,
Yan Zhang, Nina Kristiansen, Emanuel Kanal and Torsten Sommer
on behalf of the Evera-MRI Study Investigators
Circ Cardiovasc Imaging. 2016;9:
doi: 10.1161/CIRCIMAGING.115.004025
Circulation: Cardiovascular Imaging is published by the American Heart Association, 7272 Greenville Avenue,
Dallas, TX 75231
Copyright © 2016 American Heart Association, Inc. All rights reserved.
Print ISSN: 1941-9651. Online ISSN: 1942-0080
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circimaging.ahajournals.org/content/9/5/e004025
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation: Cardiovascular Imaging can be obtained via RightsLink, a service of the Copyright Clearance
Center, not the Editorial Office. Once the online version of the published article for which permission is being
requested is located, click Request Permissions in the middle column of the Web page under Services. Further
information about this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation: Cardiovascular Imaging is online at:
http://circimaging.ahajournals.org//subscriptions/