Late gadolinium enhancement cardiac imaging on a 3T scanner with ...

2 downloads 30 Views 1MB Size Report
One hundred and sixty participants prospectively enrolled underwent a 3T cardiac MRI with 3 different LGE sequences: 3D Phase-Sensitive Inversion-Recovery ...
Eur Radiol DOI 10.1007/s00330-015-4002-y

CARDIAC

Late gadolinium enhancement cardiac imaging on a 3T scanner with parallel RF transmission technique: prospective comparison of 3D-PSIR and 3D-IR Anthony Schultz 1,5 & Thibault Caspar 2 & Mickaël Schaeffer 3 & Aïssam Labani 1 & Mi-Young Jeung 1 & Soraya El Ghannudi 1 & Catherine Roy 1 & Mickaël Ohana 1,4

Received: 26 March 2015 / Revised: 10 August 2015 / Accepted: 2 September 2015 # European Society of Radiology 2015

Abstract Objective To qualitatively and quantitatively compare different late gadolinium enhancement (LGE) sequences acquired at 3T with a parallel RF transmission technique. Methods One hundred and sixty participants prospectively enrolled underwent a 3T cardiac MRI with 3 different LGE sequences: 3D Phase-Sensitive Inversion-Recovery (3DPSIR) acquired 5 minutes after injection, 3D InversionRecovery (3D-IR) at 9 minutes and 3D-PSIR at 13 minutes. All LGE-positive patients were qualitatively evaluated both independently and blindly by two radiologists using a 4level scale, and quantitatively assessed with measurement of contrast-to-noise ratio and LGE maximal surface. Statistical analyses were calculated under a Bayesian paradigm using MCMC methods. Results Fifty patients (70 % men, 56yo±19) exhibited LGE (62 % were post-ischemic, 30 % related to cardiomyopathy and 8 % post-myocarditis). Early and late 3D-PSIR were superior to 3D-IR sequences (global quality, estimated coefficient IR>early-PSIR : -2.37 CI=[-3.46 ; -1.38], prob(coef>

* Anthony Schultz [email protected] 1

Radiology Department, Nouvel Hôpital Civil, Strasbourg University Hospital, Strasbourg Cedex, France

2

Cardiology Department, Nouvel Hôpital Civil, Strasbourg University Hospital, Strasbourg Cedex, France

3

Public Health and Biostatistics Department, Nouvel Hôpital Civil, Strasbourg University Hospital, Strasbourg Cedex, France

4

iCube Laboratory, Université de Strasbourg / CNRS, UMR 7357, 67400 Illkirch, France

5

Service de Radiologie, Nouvel Hôpital Civil, 1 Place de l’Hôpital, 67091 Strasbourg Cedex, France

0) = 0 % and late-PSIR > IR : 3.12 CI = [0.62 ; 4.41], prob(coef>0)=100 %), LGE surface estimated coefficient IR>early-PSIR: -0.09 CI=[-1.11; -0.74], prob(coef>0)=0 % and late-PSIR>IR : 0.96 CI=[0.77; 1.15], prob(coef>0)= 100 %). Probabilities for late PSIR being superior to early PSIR concerning global quality and CNR were over 90 %, regardless of the aetiological subgroup. Conclusions In 3T cardiac MRI acquired with parallel RF transmission technique, 3D-PSIR is qualitatively and quantitatively superior to 3D-IR. Key Points • Late gadolinium enhancement is an essential part of a cardiac MRI examination • PSIR and IR sequences are the two possible options for LGE imaging • At 3T with parallel RF transmission, PSIR sequences are significantly better • One LGE sequence is sufficient, allowing an optimization of the acquisition time

Keywords Cardiac imaging techniques . Magnetic resonance imaging . Late gadolinium enhancement . Parallel radiofrequency transmission . Phase-sensitive inversion recovery

Abbreviations CAD Coronary artery disease CMR Cardiac magnetic resonance CNR Contrast to noise ratio LGE Late Gadolinium enhancement RF Radiofrequency TI Inversion time

Eur Radiol Table 1 Acquisition parameters of LGE sequences Sequence type Repetition time Echo time TI Flip angle Field of view Acquisition matrix Reconstruction matrix size Slice thickness acquisition Acquisition time per slice TFE factor Approximate acquisition time

IR 3D SENSE Turbo Field Echo

PSIR 3D SENSE Turbo Field Echo

1.3 ms 4 ms 250-330 ms (visually assessed with Look Locker sequence) 25° 320x320x70 mm 1.75 x1.75 mm 1.33 mm 5 mm 1 sec 146 ms 3 min

5 ms 2.4 ms 250 ms 25° 5° 320x320x42 mm 1.8 x1.85 mm 1.33 mm 6 mm 2.3 sec 160 ms 3 min

Introduction Cardiac MRI (CMR) is a key imaging modality in the modern assessment of cardiovascular diseases [1–4]. In particular, late gadolinium enhancement (LGE) analysis has a strong diagnostic and prognostic value [5–10], with a list of clinical indications constantly growing [1]. In our academic hospital (Nouvel Hôpital Civil, Strasbourg, France), 594 patients underwent a cardiac MRI in 2010 and 890 in 2014, a 50 % increase that will likely continue over the coming years. Yet, MR accessibility remains limited by availability of scanners and by time-consuming protocols, subjecting current equipment to stress and making it difficult for radiology departments to respond to needs. Consequently, there is a strong interest in optimizing the cardiac MRI acquisition protocols so as to shorten them as much as reasonably achievable [11–13] and a push towards utilizing all available MR resources, whether they are 1.5 T or 3T scanners. Recent adoption of clinical 3T MR systems [14–17] , driven by the potential for better signal-to-noise ratio and image contrast, can currently be implemented beyond the initial scope of neuroradiology. Technological innovations such as dual-source parallel radiofrequency (RF) transmission with patient-adaptive local RF shimming result in an improved

Table 2

Qualitative 4-level scale

image quality and a reduction of artefacts, which are particularly useful for 3T cardiovascular imaging [18, 19]. With the objective of optimizing our acquisition protocol to cope with the increase in CMR indications, our study primary endpoint was to qualitatively and quantitatively compare different LGE sequences - early 3D-PSIR (at 5 minutes), late 3DIR (at 9 minutes) and late 3D-PSIR (at 13 minutes) - acquired at 3T with a parallel RF transmission technique, in order to determine which one offers the best image and diagnostic quality.

Material and methods Our institutional review board approved this study and written informed consent was obtained from all participants. Study population For 9 consecutive months (January to September 2013), all patients referred to our department for a gadolinium-enhanced cardiac MRI were prospectively enrolled. Exclusion criteria were: subject younger than 18 years old, classical contraindications to MRI or to gadolinium injection, or any condition which could interfere with the patient’s ability to comply with the examination. All LGE-positive patients were further analyzed with qualitative and quantitative assessment.

Overall image quality

LGE contrast and margin sharpness

Artifacts

Imaging protocol

1

Poor

Poor

2

Fair

Fair

3 4

Acceptable Good

Acceptable Good

Severe, interfering with the evaluation Moderate, partially interfering Mild, not interfering Minimum to no artifact

All patients underwent a 3T cardiac MRI (Achieva 3.0 T Xseries, Phillips Medical Systems, Best, The Netherlands) equipped with a fully flexible dual-source RF transmission technology (i.e., Tx MultiTransmit®) and a dedicated 6channel SENSE torso coil. The same acquisition protocol was applied to all patients. In addition to the conventional cine B-TFE sequences and

Eur Radiol

Fig. 1 Signal intensity measurement on a LV short-axis view of early-3D-PSIR (a), 3D-IR (b) and late 3D-PSIR (c), with a 51.6 mm2 ROI in the mid anteroseptal wall (LGE) and mid inferoseptal wall (healthy myocardium)

morphological Spin Echo sequences, three consecutive LGE sequences were acquired in the three cardiac planes (LV shortand long-axis, 4-chamber) after the injection of a gadoliniumbased contrast agent (Gadovist, Bayer Healthcare, Leverkusen, Germany) at a dose of 0.1 mmol/kg: – – –

early 3D Phase-Sensitive Inversion-Recovery (early 3DPSIR), acquired at 5 minutes; 3D Inversion-Recovery (3D-IR), acquired at 9 minutes, after a Look-Locker sequence to visually determine the inversion time; late 3D-PSIR, acquired at 13 minutes.

For both early and late 3D-PSIR sequences, only the phase images were analyzed. Acquisition parameters are summarized in Table 1.

LGE evaluation All LGE-positive patients were independently and blindly evaluated by two chest radiologists (AS with 1 year of experience in CMR and MO with 5 years of experience), in a randomized order. Qualitative assessment was performed on a PACS workstation (Centricity RA1000, GE, Milwaukee, USA) using a 4level Likert scale (Table 2) evaluating the overall image quality, the contrast between the LGE and the healthy myocardium and the artefacts. Quantitative evaluation was obtained by the signal intensity measurement of the LGE and the adjacent healthy myocardium, using a circular region of interest (ROI) averaging at least 4 mm2 (Fig. 1), allowing for calculation of a contrast-tonoise ratio (CNR) with the following formula:

Fig. 2 Example of a 68-year-old male with LGE in the subendocardium of the LVanterior wall, typical of infarction secondary to coronary artery disease (a - early 3D-PSIR; b - 3D-IR; c - late 3D-PSIR)

Eur Radiol

Fig. 3 Example of a 45-year-old female with arrhythmogenic biventricular cardiomyopathy and large LGE lesions in the LV lateral wall and the RV wall (a - early 3D-PSIR; b - 3D-IR; c - late 3D-PSIR)

CNR ¼

ðLGE signal−myocardium signalÞ . ðLGE SDþmyocardium SDÞ

2

The largest surface of LGE was also measured, in mm2, for all sequences on the same cardiac view for every patient by drawing the LGE contours with a manual ROI. Optimal adaptation of the window settings for each sequence was performed by the same operator so as to get an optimal and reproducible visualization of the LGE. Statistical analysis Descriptive analyses were calculated using generic statistics, such as mean, standard deviation, min, and max for quantitative variables, whereas effectives and percentages were used

for qualitative variables. Cohen’s Kappa was used to measure concordance between readers [20]. Inferential analyses were calculated under a Bayesian paradigm [21–23], by using MCMC methods to evaluate posterior distributions and to measure the probability for a coefficient to be positive. MCMC methods are based on simulations under specific assumptions, and empirical posterior distributions are studied for conclusions. Quantitative evaluation scores on a scale of 4 levels were grouped into two categories, 0 for both levels 1 and 2 (insufficient quality), and 1 for levels 3 and 4 (sufficient diagnostic quality). Then, comparisons between sequences were made using hierarchical logistic regression models in order to take into account two random effects and thus intraclass variability for the reader and the subject effect. Posterior mean and 95 % confidence intervals were then estimated. Coefficients corresponding to the

Fig. 4 Example of a 20-year-old male with patchy and nodular LGE lesions in the subepicardium of the LV lateral wall and the midwall of basal septum, typical of myocarditis (a - early 3D-PSIR; b - 3D-IR; c - late 3D-PSIR)

Eur Radiol Table 3

Descriptive analysis of qualitative evaluation

Table 5

Descriptive analysis for the quantitative evaluation

Early PSIR

IR

Late PSIR

Overall quality

3.39±0.6

2.68±0.63

3.46±0.62

CNR

Artifacts

3.14±0.69

2.5±0.53

3.16±0.74

Surface (cm2)

Margin

3.16±0.63

2.69±0.73

3.31±0.65

probability of being positive that were greater than 97.5 % or less than 2.5 % were considered as non-zero. This is equivalent to checking if the 95 % confidence interval was different from 0 or not. Analyses were performed on both software WinBUGS 1.4 [24] and R 3.0.1.

Early PSIR

IR

Late PSIR

10.1±6 1.5±1.3

4.2±3.1 1.4±1.2

10.2±5.3 1.6±1.5

Results of the qualitative evaluation with a 4-level scale are summarized in Table 3. Results of the inferential statistical analysis are listed in Table 4. 3D-PSIR sequences are superior to 3D-IR. Moreover, and even though there is no statistical difference between early3D-PSIR and late-3D-PSIR sequences, the probability of late3D-PSIR being superior to early-3D-PSIR in terms of overall quality and margin is relatively high (superior to 90 %, inferior to 97.5 %).

Results Quantitative evaluation

Population One hundred and sixty patients underwent a gadoliniumenhanced cardiac MRI on our 3T scanner over the 9 months of the study. Out of these 160 patients, 54 exhibited LGE. Four LGE-positive examinations could not be analyzed, due to respiratory motion artefacts (n=1) or due to the absence of one or two LGE sequences (n=3). Fifty patients (70 % men, 56 years old ±19, range 18-88) were thus ultimately included. Among these pathological LGE findings, 62 % were related to infarction secondary to coronary artery disease (Fig. 2), 30 % were related to idiopathic cardiomyopathy, either dilated, hypertrophic or restrictive (Fig. 3), and 8 % were related to myocarditis (Fig. 4). Qualitative evaluation For the overall image quality, as well as for the artefacts and the LGE contrast and sharpness, early and late 3D-PSIR sequences were significantly superior to 3D-IR sequences. Table 4

Quantitative assessment confirmed the superiority of 3DPSIR over 3D-IR (descriptive analysis in Table 5, inference analysis in Table 6). There was a significant difference between IR and PSIR sequences in terms of LGE surface measurement, with larger surfaces in the latter case. In terms of CNR, late-3D-PSIR and early-3D-PSIR sequences were superior to 3D-IR with a probability of 100 % and 86 %, respectively.

Inter-observer agreement Kappa tests concerning overall image quality showed a low to moderate agreement between the two readers (0.31 for the early-3D-PSIR sequence, 0.44 for the 3DIR sequence, 0.62 for the late-3D-PSIR). When analyzed more specifically, the disagreement was in more than 90 % of cases on an examination ranked 3 by one reader and 4 by the other one, meaning that diagnostic

Results of the inferential statistical analysis for the qualitative evaluation

IR>Early PSIR

Late PSIR>Early PSIR

Late PSIR>IR

Mean of difference Credible interval [2.5 ; 97.5] Likelihood Mean of difference Credible interval [2.5 ; 97.5] Likelihood Mean of difference Credible interval [2.5 ; 97.5] Likelihood

Overall quality

Artifacts

Margin

-2.3690 [-3.4590 ; -1.3840] 0.000* 0.7479 [-0.3607 ; 1.9160] 0.904 3.1169 [0.6173 ; 4.4060] 1.000*

-2.4322 [-3.4400 ; -1.49970] 0.000* -0.0542 [-0.9855 ; -0.8858] 0.45 2.3780 [0.4985 ; 3.4020] 1.000*

-1.2437 [-2.0030; -0.5108] 0.004* 0.5494 [-0.2582 ; 1.3830] 0.9071 1.7931 [0.9916 ; 2.6440] 1.000*

Eur Radiol Table 6 Inferential analysis results for the quantitative evaluation IR>Early PSIR

Late PSIR>Early PSIR

Late PSIR>IR

Mean of difference Credible interval [2.5 ; 97.5]

-0.09289 [-1.1120 ; -0.7377]

0.1413

0.000*

0.1342 [0.0549; 0.2144]

0.0346 [-0.1544; 0.2254]

Likelihood

0.9991*

0.6406

Mean of difference Credible interval [2.5 ; 97.5]

0.1777 [0.0979; 0.2553]

0.9636 [0.7748; 1.1500]

Likelihood

1.000*

1.000*

The same trends were observed in the different etiological subgroups, with significant superiority of the 3D-PSIR over the 3D-IR (Tables 7, 8, and 9).

Discussion The comparison between different LGE sequences demonstrated the quantitative and qualitative superiority of the 3DPSIR over the 3D-IR sequences. These results are in agreement with previous studies made at 1.5 Tesla, where the robustness of PSIR sequences for LGE imaging is well-established [25–30]. Few studies are performed at 3T, and their results are disparate. Kido et al. [31] evaluated 56 patients with LGE on a 3T scanner, and showed no significant difference in overall image quality between free-breathing 3D-PSIR and breath-held 3D-IR. A strong correlation between both methods was found (R2 =0.96), with larger LGE volumes detected with PSIR compared to IR in

patients suspected of non-ischemic cardiomyopathy, whereas LGE volumes did not differ significantly between both sequences in patients with ischemic cardiac disease. Morita et al. [32] studied 30 patients with hypertrophic cardiomyopathy at 3T and found that myocardium-LGE contrast and overall image quality were significantly higher on 3D-PSIR than on 2D-IR images (pIR

-0.0435 [-0.1207; 0.0364]

Mean of difference Credible interval [2.5 ; 97.5]

Subgroup analysis

Late PSIR>Early PSIR

Surface

Likelihood

(examinations graded 3 or 4) and non-diagnostic examinations (graded 1-2) were correctly ranked by both readers.

IR>Early PSIR

CNR

Mean of difference Credible interval [2.5 ; 97.5] Likelihood Mean of difference Credible interval [2.5 ; 97.5] Likelihood Mean of difference Credible interval [2.5 ; 97.5] Likelihood

Overall quality

CNR

-3.2396 [-4.7760; -1.8860] 0.000* 0.7703 [-0.8184; 2.4770] 0.8220 4.0100 [2.3820; 5.9320] 1.000*

-0.9218 [-1.1665; -0.6796] 0.0000* 0.0775 [-0.1712 ; 0.3318] 0.7278 0.1879 [0.0647; 0.3109] 1.0000*

Eur Radiol Table 8 Results of Bayesian analysis for the myocarditis subgroup IR>Early PSIR

Late PSIR>Early PSIR

Late PSIR>IR

Overall quality

CNR

-3.3169 [-7.5900; 0.5818]

-0.8372 [-1.5326 ; -0.1318]

Likelihood

0.0497*

0.0091*

Mean of difference Credible interval [2.5 ; 97.5]

1.6791 [-3.4140; 7.4880]

-0.0239 [-0.7265 ; 0.6648]

Likelihood

0.7173

0.4689

Mean of difference Credible interval [2.5 ; 97.5]

4.9960 [-0.5298; 11.7800]

0.8134 [0.1143 ; 1.5091]

Likelihood

0.9583*

0.9889*

Mean of difference Credible interval [2.5 ; 97.5]

myocardium [41]. Its performance, therefore, greatly relies on the TI selected, and an inappropriate selection of the optimal TI will result in a reduction in contrast, which can lead to a reduced LGE area and an underestimation of the extent of myocardial fibrosis. The PSIR sequence offsets any reduction in CNR due to TI changes and maintains signal as well as contrast consistent during scanning, avoiding the need for a narrow definition of optimal TI. The 3D-acquisition technique allows for complete heart coverage with a high spatial resolution. Therefore the 3D PSIR sequence is able to provide operator-independent and reproducible images with a high spatial resolution, and perform accurate lesion detection as well as reliable quantification of myocardial fibrosis. In our work, late-3D-PSIR at 13 minutes showed a high likelihood (90 to 99 %) of being superior to early-3D-PSIR acquired at 5 minutes, in terms of overall image quality, LGE margin sharpness, and CNR, even though the difference did not reach statistical significance. In all our patients, LGE positive diagnosis was possible with both early and late PSIR sequences, and these two sequences showed no difference in terms of LGE surface evaluation. This result might be explained by the high proportion of ischemic cardiomyopathy in our population. Indeed, the gadolinium kinetics of myocardial infarction [42], with myocardial late enhancement due to the relative slow accumulation of contrast agent in areas of

fibrosed myocardial interstitium, is different from those of acute myocarditis [43, 44]. Acute myocarditis exhibits early myocardial enhancement secondary to relative fast accumulation of contrast agent in the inflamed myocardial interstitial areas, due to hyperemia and capillary leakage or other cardiomyopathies. Furthermore, the equivalence between early and late 3D-PSIR sequences in terms of artefacts (which are not influenced by the delay of the gadolinium injection) corroborates this hypothesis. The results of our etiological subgroup analysis draw the same conclusions: 3D-PSIR is superior to 3D-IR, both for overall image quality and for CNR. Bayesian inference analysis showed a strong yet non-significant trend for the superiority of late PSIR over early PSIR sequences regardless of the subgroup of patients considered, which is concordant with the literature data for ischemic cardiomyopathy [42, 43], but more surprising for the myocarditis subgroup where early LGE sequences are thought to be more sensitive [9]. This study has several limitations. First of all, this study is vendor-specific, and limited to a 3T scanner with dual-source RF transmission technology. It is, however, likely that our results could be applied to any 3T scanner. One can also point out the difference in the slice thickness between both types of sequences, to the advantage of 3D-

Table 9 Results of Bayesian analysis for the cardiomyopathy subgroup IR>Early PSIR

Late PSIR>Early PSIR

Late PSIR>IR

Mean of difference Credible interval [2.5 ; 97.5] Likelihood Mean of difference Credible interval [2.5 ; 97.5] Likelihood Mean of difference Credible interval [2.5 ; 97.5] Likelihood

Overall quality

CNR

-1.4600 [-2.9560; -0.0645] 0.0198* 0.9008 [-0.6396;2.5330] 0.8694 2.3608 [0.7947; 4.0990] 0.9988*

-0.9746 [-1.3251; -0.6214] 0.0000* -0.0530 [-0.4113; 0.3079] 0.3858 0.9216 [0.1798; 1.2730] 1.0000*

Eur Radiol

PSIR (6 mm versus 5 mm for 3D-IR), which could result in a higher signal and a lower noise. We chose the acquisition parameters based on the vendor’s recommendations, with the objective of keeping a similar acquisition time (i.e., 3 minutes). Moreover, our qualitative evaluation is based on the contrast-to-noise ratio, which should consequently be relatively independent of the slice thickness, as a proportional increase in the signal and the noise would lead to an equivalent CNR. Even though Bayesian inference analysis enables superior statistic power and gives us a probability evaluation of the qualitative difference between all studied sequences, a binary clustering of qualitative assessment data (class 1 and 2 grouped in category 0, class 3 and 4 grouped in category 1) was needed, and de facto leading to a loss of information. Moreover, concordance analysis showed moderate agreement between the two radiologists. This result needs to be put into perspective with (1) the number of categories -four- of our evaluation: the more categories, the lower is the concordance and (2) the fact that the disagreement was in more than 90 % of cases between class 3 and 4, which does not impact the diagnosis or work-up of patients. In conclusion, this study demonstrates that on a 3T scanner equipped with a parallel RF transmission technique, 3D-PSIR sequences are significantly superior to conventional 3D-IR, both in terms of subjective assessment of LGE and objective CNR. Late 3D-PSIR sequences are possibly better than or at least as informative as early 3D-PSIR, regardless of the etiological subgroup. Consequently, a systematic acquisition of all three sequences is redundant. We can, therefore, propose a faster cardiac MRI protocol relying only on late 3D-PSIR. Particularly suited for routine patients with ischemic cardiomyopathy, its implementation in our department has allowed a gain of roughly 10 minutes per examination that has, to some extent, helped us in responding to the growing need for CMR examinations.

2.

3.

4.

5.

6.

7.

8.

9.

10. 11.

12.

13.

14.

15. Acknowledgments The scientific guarantor of this publication is Pr. Catherine Roy. The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article. The authors state that this work has not received any funding. One of the authors is a biostatistician: Dr. Mickaël Schaeffer. Institutional Review Board approval was obtained. Written informed consent was obtained from all subjects (patients) in this study. No study subjects or cohorts have been previously reported. Methodology: prospective, observational, performed at one institution. Conflicts of interest None

16.

17. 18.

19. 20.

References 1.

American College of Cardiology Foundation Task Force on Expert Consensus D, Hundley WG, Bluemke DA et al (2010) ACCF/ACR/AHA/NASCI/SCMR 2010 expert consensus

21.

22.

document on cardiovascular magnetic resonance: a report of the American College of Cardiology Foundation Task Force on Expert Consensus Documents. J Am Coll Cardiol 55:2614–2662 Wu KC (2009) Variation on a theme: CMR as the "one-stop shop" for risk stratification after infarction? J Am Coll Cardiol Img 2:843– 845 Pennell DJ, Sechtem UP, Higgins CB et al (2004) Clinical indications for cardiovascular magnetic resonance (CMR): consensus panel report. Eur Heart J 25:1940–1965 Rehwald WG, Wagner A, Sievers B, Kim RJ, Judd RM (2007) Cardiovascular MRI: its current and future use in clinical practice. Expert Rev Cardiovasc Ther 5:307–321 Flett AS, Hasleton J, Cook C et al (2011) Evaluation of techniques for the quantification of myocardial scar of differing etiology using cardiac magnetic resonance. J Am Coll Cardiol Img 4:150–156 Kim RJ, Wu E, Rafael A et al (2000) The use of contrast-enhanced magnetic resonance imaging to identify reversible myocardial dysfunction. N Engl J Med 343:1445–1453 Klem I, Heitner JF, Shah DJ et al (2006) Improved detection of coronary artery disease by stress perfusion cardiovascular magnetic resonance with the use of delayed enhancement infarction imaging. J Am Coll Cardiol 47:1630–1638 Laissy JP, Pasi N, Bazeli R, Schouman-Claeys E, Serfaty JM (2010) Delayed myocardial enhancement and diagnosis of acute coronary syndrome with normal coronarography. J Radiol 91:602–608 Silvera S, Palangie E, Marmursztejn J et al (2010) Which nonischemic myocardial diseases show delayed enhancement, and what are their imaging characteristics. J Radiol 91:609–614 Weaver JC, McCrohon JA (2008) Contrast-enhanced cardiac MRI in myocardial infarction. Heart Lung Circ 17:290–298 Kramer CM, Barkhausen J, Flamm SD, Kim RJ, Nagel E, Society for Cardiovascular Magnetic Resonance Board of Trustees Task Force on Standardized P (2013) Standardized cardiovascular magnetic resonance (CMR) protocols 2013 update. J Cardiovasc Magn Reson Off J Soc Cardiovasc Magn Reson 15:91 Jacquier A, Bartoli B, Flavian A et al (2010) Delayed myocardial enhancement: optimizing the MR imaging protocol. J Radiol 91: 598–601 Xu J, Kim D, Otazo R et al (2013) Towards a five-minute comprehensive cardiac MR examination using highly accelerated parallel imaging with a 32-element coil array: feasibility and initial comparative evaluation. J Magn Reson Imaging JMRI 38:180–188 Cheng AS, Selvanayagam JB (2010) High field cardiac magnetic resonance imaging–current and future perspectives. Heart Lung Circ 19:145–153 Gerretsen SC, Versluis B, Bekkers SC, Leiner T (2008) Cardiac cine MRI: comparison of 1.5 T, non-enhanced 3.0 T and blood pool enhanced 3.0 T imaging. Eur J Radiol 65:80–85 Gutberlet M, Noeske R, Schwinge K, Freyhardt P, Felix R, Niendorf T (2006) Comprehensive cardiac magnetic resonance imaging at 3.0 Tesla: feasibility and implications for clinical applications. Invest Radiol 41:154–167 Willinek WA, Schild HH (2008) Clinical advantages of 3.0 T MRI over 1.5 T. Eur J Radiol 65:2–14 Rajiah P, Bolen MA (2014) Cardiovascular MR imaging at 3 T: opportunities, challenges, and solutions. Radiographics 34:1612– 1635 Wieben O, Francois C, Reeder SB (2008) Cardiac MRI of ischemic heart disease at 3 T: potential and challenges. Eur J Radiol 65:15–28 Cohen J (1960) A coefficient of agreement for nominal scales. Educ Psychol Meas 20:37–46 Ronald Christensen WJ, Branscum A, Hanson TE (2010) Bayesian ideas and data analysis: an introduction for scientists and statisticians. CRC Press Freedman L (1996) Bayesian statistical methods. BMJ 313:569– 570

Eur Radiol 23.

Ntzoufras I (2009) Bayesian modelling using WinBUGS. Wiley, Hoboken 24. Lunn DJ, Thomas A, Best N, Spiegelhalter D (2000) WinBUGS - a Bayesian modelling framework: concepts, structure, and extensibility. Stat Comput 10:325–337 25. Chen X, Lu M, Yin G et al (2013) Three-dimensional phase-sensitive inversion-recovery Turbo FLASH sequence for the assessment of left ventricular myocardial scar in swine. PLoS One 8, e78305 26. Kino A, Keeling AN, Farrelly CT et al (2011) Assessment of left ventricular myocardial scar in infiltrative and non-ischemic cardiac diseases by free breathing three dimensional phase sensitive inversion recovery (PSIR) TurboFLASH. Int J Cardiovasc Imaging 27: 527–537 27. Kino A, Zuehlsdorff S, Sheehan JJ et al (2009) Three-dimensional phase-sensitive inversion-recovery turbo FLASH sequence for the evaluation of left ventricular myocardial scar. AJR Am J Roentgenol 193:W381–W388 28. Jablonowski R, Nordlund D, Kanski M et al (2013) Infarct quantification using 3D inversion recovery and 2D phase sensitive inversion recovery; validation in patients and ex vivo. BMC Cardiovasc Disord 13:110 29. Elgeti T, Abdel-Aty H, Wagner M et al (2007) Assessment of late gadolinium enhancement in nonischemic cardiomyopathy: comparison of a fast Phase-Sensitive Inversion Recovery Sequence (PSIR) and a conventional segmented 2D gradient echo recall (GRE) sequence–preliminary findings. Invest Radiol 42:671–675 30. Huber A, Bauner K, Wintersperger BJ et al (2006) Phase-sensitive inversion recovery (PSIR) single-shot TrueFISP for assessment of myocardial infarction at 3 tesla. Invest Radiol 41:148–153 31. Kido T, Kido T, Nakamura M et al (2014) Three-dimensional phase-sensitive inversion recovery sequencing in the evaluation of left ventricular myocardial scars in ischemic and non-ischemic cardiomyopathy: comparison to three-dimensional inversion recovery sequencing. Eur J Radiol 83:2159–2166 32. Morita K, Utsunomiya D, Oda S et al (2013) Comparison of 3D phase-sensitive inversion-recovery and 2D inversion-recovery MRI at 3.0 T for the assessment of late gadolinium enhancement in patients with hypertrophic cardiomyopathy. Acad Radiol 20: 752–757 33. Katscher U, Bornert P (2006) Parallel RF transmission in MRI. NMR Biomed 19:393–400

34.

Willinek WA, Gieseke J, Kukuk GM et al (2010) Dual-source parallel radiofrequency excitation body MR imaging compared with standard MR imaging at 3.0 T: initial clinical experience. Radiology 256:966–975 35. Nelles M, Konig RS, Gieseke J et al (2010) Dual-source parallel RF transmission for clinical MR imaging of the spine at 3.0 T: intraindividual comparison with conventional single-source transmission. Radiology 257:743–753 36. Krishnamurthy R, Pednekar A, Kouwenhoven M, Cheong B, Muthupillai R (2013) Evaluation of a subject specific dualtransmit approach for improving B1 field homogeneity in cardiovascular magnetic resonance at 3T. J Cardiovasc Magn Reson Off J Soc Cardiovasc Magn Reson 15:68 37. Mueller A, Kouwenhoven M, Naehle CP et al (2012) Dual-source radiofrequency transmission with patient-adaptive local radiofrequency shimming for 3.0-T cardiac MR imaging: initial experience. Radiology 263:77–85 38. Strach K, Clauberg R, Muller A et al (2013) Feasibility of high-dose dobutamine stress SSFP Cine MRI at 3 Tesla with patient adaptive local RF Shimming using dual-source RF transmission: initial results. Röfo 185:34–39 39. Jia H, Wang C, Wang G et al (2013) Impact of 3.0 T Cardiac MR Imaging Using Dual-Source Parallel Radiofrequency Transmission with Patient-Adaptive B1 Shimming. PLoS One 8, e66946 40. Kellman P, Arai AE, McVeigh ER, Aletras AH (2002) Phasesensitive inversion recovery for detecting myocardial infarction using gadolinium-delayed hyperenhancement. Magn Reson Med 47:372–383 41. Kim RJ, Shah DJ, Judd RM (2003) How we perform delayed enhancement imaging. J Cardiovasc Magn Reson Off J Soc Cardiovasc Magn Reson 5:505–514 42. Doltra A, Skorin A, Hamdan A et al (2014) Comparison of acquisition time and dose for late gadolinium enhancement imaging at 3.0 T in patients with chronic myocardial infarction using GdBOPTA. Eur Radiol 24:2192–2200 43. Laissy JP, Hyafil F, Feldman LJ et al (2005) Differentiating acute myocardial infarction from myocarditis: diagnostic value of earlyand delayed-perfusion cardiac MR imaging. Radiology 237:75–82 44. Kellman P, Arai AE (2012) Cardiac imaging techniques for physicians: late enhancement. J Magn Reson Imaging JMRI 36:529–542