Author + information
- Received March 2, 2015
- Revision received April 26, 2015
- Accepted April 29, 2015
- Published online September 1, 2015.
- Avinash Kali, MS∗,†,
- Eui-Young Choi, MD‡,
- Behzad Sharif, PhD∗,
- Young Jin Kim, MD§,
- Xiaoming Bi, PhD‖,
- Bruce Spottiswoode, PhD¶,
- Ivan Cokic, MD∗,
- Hsin-Jung Yang, MS∗,†,
- Mourad Tighiouart, PhD#,
- Antonio Hernandez Conte, MD, MBA∗∗,
- Debiao Li, PhD∗,†,††,
- Daniel S. Berman, MD∗,††,‡‡,
- Byoung Wook Choi, MD§,
- Hyuk-Jae Chang, MD‡∗ ( and )
- Rohan Dharmakumar, PhD∗,†,††,‡‡∗∗ ()
- ∗Biomedical Imaging Research Institute, Department of Biomedical Sciences, Cedars-Sinai Medical Center, Los Angeles, California
- †Department of Bioengineering, University of California, Los Angeles, California
- ‡Division of Cardiology, Yonsei University College of Medicine, Seoul, South Korea
- §Department of Radiology, Yonsei University College of Medicine, Seoul, South Korea
- ‖MR Research & Development, Siemens Healthcare, Los Angeles, California
- ¶MR Research & Development, Siemens Healthcare, Chicago, Illinois
- #Biostatistics and Bioinformatics Research Center, Cedars-Sinai Medical Center, Los Angeles, California
- ∗∗Department of Anesthesiology, Cedars-Sinai Medical Center, Los Angeles, California
- ††Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, California
- ‡‡Cedars-Sinai Heart Institute, Cedars-Sinai Medical Center, Los Angeles, California
- ↵∗Reprint requests and correspondence:
Dr. Hyuk-Jae Chang, Division of Cardiology, Severance Hospital, Yonsei University College of Medicine, 250 Seongsanno, Seodaemun-gu, Seoul, South Korea 120-752.
- ↵∗∗Dr. Rohan Dharmakumar, Department of Biomedical Sciences, Cedars-Sinai Medical Center, Biomedical Imaging Research Institute, PACT Building—Suite 800, 8700 Beverly Boulevard, Los Angeles, California 90048.
Objectives The purpose of this study was to investigate whether native T1 maps at 3-T can reliably characterize chronic myocardial infarctions (MIs) in patients with prior ST-segment elevation myocardial infarction (STEMI) or non-ST-segment elevation myocardial infarction (NSTEMI).
Background Late gadolinium enhancement (LGE) cardiac magnetic resonance is the gold standard for characterizing chronic MIs, but it is contraindicated in patients with end-stage chronic kidney disease.
Methods Native T1 and LGE images were acquired at 3-T in patients with prior STEMI (n = 13) and NSTEMI (n = 12) at a median of 13.6 years post-MI. Infarct location, size, and transmurality were measured using mean ± 5 SDs thresholding criterion from LGE images and T1 maps and compared against one another. Independent reviewers assessed visual conspicuity of MIs on LGE images and T1 maps.
Results Native T1 maps and LGE images were not different for measuring infarct size (STEMI: p = 0.46; NSTEMI: p = 0.27) and transmurality (STEMI: p = 0.13; NSTEMI: p = 0.21) using thresholding criterion. Using thresholding criterion, good agreement was observed between LGE images and T1 maps for measuring infarct size (STEMI: bias = 0.6 ± 3.1%; R2 = 0.93; NSTEMI: bias = −0.4 ± 4.4%; R2 = 0.85) and transmurality (STEMI: bias = 2.0 ± 4.2%; R2 = 0.89; NSTEMI: bias = −2.7 ± 7.9%; R2 = 0.68). Sensitivity and specificity of T1 maps for detecting chronic MIs based on thresholding criterion were 89% and 98%, respectively (STEMI), and 87% and 95%, respectively (NSTEMI). Relative to LGE images, the mean visual conspicuity score for detecting chronic MIs was significantly lower for T1 maps (p < 0.001 for both cases). Median infarct-to-remote myocardium contrast-to-noise ratio was 2.5-fold higher for LGE images relative to T1 maps (p < 0.001). Sensitivity and specificity of T1 maps for visual detection were 60% and 86%, respectively (STEMI), and 64% and 91% (NSTEMI), respectively.
Conclusions Chronic MIs in STEMI and NSTEMI patients can be reliably characterized using threshold-based detection on native T1 maps at 3-T. Visual detection of chronic MIs on native T1 maps in both patient populations has high specificity, but modest sensitivity.
Determining infarct location, size, and transmurality can be instrumental in the clinical management of patients with prior myocardial infarction (MI) (1,2). In patients with prior MI, the risk of sudden cardiac death (3) or major adverse complications, such as heart failure (1), have been directly related to chronic MI characteristics (infarct size, location, and transmurality). Visualization and quantification of MI by late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR) has been vital to building this understanding (4,5). Several studies now support the belief that LGE CMR can be helpful in identifying chronic MI patients for implantable cardiac defibrillators for primary prevention of sudden cardiac death (6).
In spite of its capabilities, a well-known limitation of LGE CMR is the requirement for gadolinium-based contrast agents, which are contraindicated in patients with severe kidney dysfunction (7–9). In fact, given the comorbidity of late-stage renal failure in patients with ischemic heart disease (10,11), LGE CMR is estimated to be contraindicated in nearly 20% of MI patients. Hence, a CMR approach that could visually detect and quantify chronic MI without exogenous contrast media would be immensely valuable in the overall diagnosis and management of MI patients.
Using a canine model of reperfused chronic MI, we previously demonstrated that native T1 mapping at 3-T provides significantly greater sensitivity and specificity for characterizing chronic MI over native T1 mapping at 1.5-T (12). In this study, we examined whether native T1 mapping at 3-T can be clinically useful in the characterization of chronic MI in patients with a prior history of healed ST-segment elevation myocardial infarction (STEMI) or non-ST-segment elevation myocardial infarction (NSTEMI) relative to LGE imaging.
Patients (13 STEMI and 12 NSTEMI) with prior MI were studied according to the protocols approved by the Institutional Review Board of Severance Hospital, Yonsei University Health System, at a median of 13.6 years (interquartile range [IQR]: 7.5 to 18.5 years) after acute MI. CMR studies were performed on a 3-T clinical magnetic resonance system (MAGNETOM Trio, Siemens Healthcare, Erlangen, Germany) after obtaining informed consent. Patients were excluded from the study if they had symptoms of chest pain, electrocardiogram changes, or cardiac enzyme elevation within 1 year before the date of CMR examination; multiple infarctions; or were contraindicated in a magnetic resonance study (claustrophobia, metallic implants, glomerular filtration rate <45 ml/min/1.73 m2, and so on). The patients’ clinical features are summarized in Table 1. Electrocardiogram-triggered breath-held 2-dimensional cine balanced steady-state free precession (25 to 30 cardiac phases, repetition time/echo time = 2.92/1.46 ms, flip angle = 50°, bandwidth = 888 Hz/pixel, voxel size = 1.3 × 1.3 × 8 mm3), pre-contrast modified Look-Locker inversion recovery (MOLLI) (8 inversion times [TI] with 2 Look-Locker cycles of 3 + 5 images ; Siemens Works in Progress Package number 448, minimum TI = 120 ms, TI increment = 80 ms, flip angle = 35°, bandwidth = 1,085 Hz/pixel, voxel size = 1.5 × 1.5 × 8 mm3), and LGE images (inversion recovery prepared segmented fast low-angle shot, acquired 10 to 12 min following intravenous administration of 0.2 mmol/kg of gadobutrol [Gadovist, Bayer Schering Pharma, Berlin, Germany]; and optimal TI for nulling the remote myocardium; repetition time/echo time = 6.54/3.27 ms; flip angle = 20°; bandwidth = 460 Hz/pixel; voxel size = 1.2 × 1.2 × 8 mm3) were acquired along the short-axis direction covering the entire left ventricle.
Motion-corrected native T1 maps were constructed from the nonrigid motion-corrected pre-contrast MOLLI images as previously described (14) by the scanner’s image reconstruction system. All image analyses were performed on cvi42 (Circle Cardiovascular Imaging Inc., Calgary, Canada). LGE images and T1 maps were randomized and independently analyzed by 2 blinded reviewers in consensus. The locations of remote myocardium on both techniques were identified as the regions showing no hyperintensity on the respective images and reference regions of interest were drawn in the remote myocardium. In both techniques, infarct was identified using the mean ± 5 SD criterion relative to the respective reference regions of interest (12,15–18). Hypointense cores on T1 maps, suggestive of chronic iron deposition (12) or fat deposition (19–21), that were not detected as infarcted using the mean ± 5 SD criterion were manually included in the final analysis.
Infarct size, as a percentage of total left ventricular volume, was measured from both techniques. To evaluate the concordance between the 2 techniques for detecting infarct on a regional basis, infarct size was also measured within the first 16 segments of the American Heart Association (AHA) 17-segment model. The apical cap was excluded to avoid partial volume effects. Mean infarct transmurality was measured using the centerline chord method by calculating the extent of the scar along 100 equally placed chords drawn on each slice (22).
T1 values of infarct and remote myocardium were measured. Percentage changes in the native T1 value and LGE signal intensity (LGE-SI) of infarct relative to the remote myocardium were also measured. Contrast-to-noise ratios (CNRs) were calculated for the 2 techniques as follows:
Visualization of infarcted myocardium
Two blinded, independent reviewers with >5 years’ experience in reading CMR images scored randomized LGE images and T1 maps for the visual conspicuity of infarct. Each reviewer was presented with basal, midventricular, and apical slices from every patient for both techniques. The reviewers were allowed to freely window both the LGE images and the T1 maps to their preference. Using the AHA 17-segment model, basal and midventricular slices were divided into 6 segments each, whereas the apical slices were divided into 4 segments. The apical cap was excluded to avoid partial volume effects. Each reviewer scored for the visual conspicuity of the MI in each segment using the following scale: 1, absent; 2, uncertain; and 3, present.
Statistical analyses were performed on IBM SPSS Statistics version 21.0 (IBM Corporation, Armonk, New York). Shapiro-Wilk test and quantile-quantile plots were used to test the normality of the data. Whole-heart infarct size and transmurality were compared between LGE images and T1 maps using paired Student t test if the data was normal or Wilcoxon signed rank test if the data was non-normal. Infarct size measured on the basis of the AHA 17-segment model was averaged for a given AHA segment across all the canines for both the techniques. LGE images and T1 maps were then compared for the differences in “averaged” AHA segmental infarct size. Bland-Altman and linear regression analyses were performed to determine the agreement between the 2 techniques. The slope of the best-fit line from linear regression was tested to be equal to 1, and the intercept was tested to be equal to 0. Sensitivity and specificity of T1 maps for threshold-based detection of infarct were measured using LGE images as the gold standard and AHA segmental infarct size as the predictor variable. For this purpose, an AHA segment with an infarct size >1% by volume was considered positive for infarction, whereas an AHA segment with an infarct size <1% by volume was considered negative. A 1% cutoff was used to eliminate segments with spurious hyperintense pixels from being considered as infarcted. Receiver-operating characteristic analysis was performed to calculate the area under the curve by using the AHA segmental infarct size on T1 maps as the continuous variable and the presence or absence of infarction using the 1% cutoff on LGE images as the status variable. Native T1 values of infarct and remote myocardium, percentage changes in the native T1 value and LGE-SI of infarct relative to remote myocardium, and CNR measures were compared.
Cohen’s kappa coefficient was calculated to evaluate the agreement between the 2 reviewers on the visual conspicuity of infarction from both techniques. The mean score for each segment was calculated for both methods by averaging the scores for that segment from the 2 reviewers and were compared using a mixed-model analysis of variance. Sensitivity and specificity for visual identification of chronic MIs on T1 maps were measured using LGE images as the gold standard. For this purpose, segments with a mean score >2 were considered positive for MI, segments with a mean score <2 were considered negative for MI, and segments with a mean score equal to 2 either on LGE images or T1 maps were not included in the final analysis. Statistical significance for all analyses was set at p < 0.05. Normal data are expressed as mean ± SD, whereas non-normal data are expressed as median with IQR.
Representative LGE images and native T1 maps of the basal, midventricular, and apical slices acquired from a STEMI and an NSTEMI patient are shown in Figures 1 and 2, respectively. In both cases, visually conspicuous T1 increases could be observed within the MI territories identified on LGE images (yellow arrows in raw images in Figures 1 and 2). Semiautomatic threshold analysis using mean ± 5 SD criterion showed excellent visual agreement between the 2 techniques in terms of MI location and spatial extent in both cases (highlighted pixels in the processed images in Figures 1 and 2). Bulls-eye plots depicting infarct size and transmurality also showed good agreement between the 2 techniques.
Infarct size comparisons
LGE images and T1 maps were not different for measuring whole left ventricle infarct size in both STEMI (LGE: median 13.8%, IQR: 10.2% to 17.0%; T1: median 14.8%, IQR: 12.8% to 17.9%; p = 0.46) (Figure 3A) and NSTEMI (LGE: median 10.9%, IQR: 4.9% to 13.4%; T1: median 12.6%, IQR: 6.2% to 14.7%; p = 0.27) (Figure 3B) patients. In both patient pools, Bland-Altman (STEMI: bias = 0.6 ± 3.1%; NSTEMI: bias = −0.4 ± 4.4%) (Figures 3C and 3D, respectively) and linear regression analyses (STEMI: R2 = 0.93; slope = 0.91, p = 0.25; intercept = 2.12%, p = 0.19; NSTEMI: R2 = 0.85; slope = 0.62, p = 0.001; intercept = 4.02%, p = 0.009) (Figures 3E and 3F, respectively) showed good agreement between the 2 techniques for measuring infarct size. In 1 NSTEMI patient with inferolateral infarct, significant underestimations of the infarct size were observed in T1 maps, primarily from banding artifacts in the inferior segment (LGE: 32.7% vs. T1: 20.4%). Averaged segmental infarct size was not different between the 2 techniques in STEMI (p = 0.09) and NSTEMI (p = 0.37) patients.
LGE images and T1 maps were not different for measuring transmurality in both STEMI (LGE: median 55.6%, IQR: 54.0% to 65.8%; T1: median 57.3%, IQR: 53.1% to 69.5%; p = 0.13) (Figure 4A) and NSTEMI (LGE: median 65.1%, IQR: 56.9% to 73.3%; T1: median 59.3%, IQR: 55.9% to 68.8%; p = 0.21) (Figure 4B) patients. In both patient pools, Bland-Altman (STEMI: bias = 2.0 ± 4.2%; NSTEMI: bias = −2.7 ± 7.9%) (Figures 4C and 4D, respectively) and linear regression analyses (STEMI: R2 = 0.89; slope = 0.97, p = 0.79; intercept = 3.7%; p = 0.57; NSTEMI: R2 = 0.68; slope = 0.76, p = 0.18; intercept = 12.4%, p = 0.27) (Figures 4E and 4F, respectively) showed good agreement between the 2 techniques for measuring transmurality.
Diagnostic performance of native T1 maps using threshold-based detection
Approximately 5% of AHA segments (10 segments from STEMI patients; 13 segments from NSTEMI patients) were uninterpretable because of contamination from banding artifacts and were excluded from further analysis. Sensitivity and specificity of the native T1 maps to detect chronic STEMIs on a segmental basis were 91% (108 of 119 true positives; 95% confidence interval [CI]: 86% to 96%) and 97% (77 of 79 true negatives; 95% CI: 94% to 100%) respectively. Area under the curve was 0.96 (p < 0.001) (Figure 5A). Sensitivity and specificity of the native T1 maps to detect chronic NSTEMIs on a segmental basis were 91% (82 of 90 true positives; 95% CI: 85% to 97%) and 94% (84 of 89 true negatives; 95% CI: 90% to 99%), respectively. Area under the curve was 0.95 (p < 0.001) (Figure 5B).
Image contrast characteristics between infarct and remote myocardium
Relative to the remote myocardium, median T1 of infarct was 271 ms higher (IQR: 197 to 332 ms) in STEMI patients (infarct: median 1,517 ms, IQR: 1,443 to 1,627 ms; remote: median 1,247 ms, IQR: 1,210 to 1,302 ms; p < 0.001) (Figure 6A) and 229 ms (IQR: 190 to 323 ms) higher in NSTEMI patients (infarct: median 1,549 ms, IQR: 1,399 to 1,624 ms; remote: median 1,262 ms, IQR: 1,209 to 1,326 ms; p < 0.001) (Figure 6B). The median percentage change in LGE-SI of infarct relative to remote myocardium was significantly higher than the percentage change in T1 in both STEMI (LGE: median 465%, IQR: 362% to 629%; T1: median 21%, IQR: 17% to 27%; p < 0.001) (Figure 6C) and NSTEMI (LGE: median 441%, IQR: 343% to 569%; T1: median 20%, IQR: 16% to 25%; p < 0.001) (Figure 6D) patients. Median CNR of LGE images was also 2.5-fold higher relative to that of T1 maps in both STEMI (LGE: 23.1, IQR: 15.6 to 39.7; T1: 9.2, IQR: 7.0 to 12.3; p < 0.001) (Figure 6E) and NSTEMI (LGE: 25.3, IQR: 16.4 to 32.5; T1: 9.7, IQR: 7.1 to 12.1; p < 0.001) (Figure 6F) patients.
Visual detection of chronic MI on LGE images and native T1 maps
Cohen’s kappa coefficient showed good interobserver agreement for the visual conspicuity of chronic MIs on LGE images and T1 maps in both STEMI (κ = 0.86, p = 0.023) and NSTEMI (κ = 0.86, p = 0.024) patients. The mean score for LGE images was significantly higher than those for T1 maps in both cases (STEMI 1.96 ± 0.93 vs. 1.71 ± 0.71, p = 0.021; NSTEMI 1.83 ± 0.93 vs. 1.66 ± 0.89, p = 0.024) (Figure 7). Fewer than 5% of AHA segments (8 segments from STEMI patients; 10 segments from NSTEMI patients) were uninterpretable because of contamination from banding artifacts. Fourteen segments from STEMI patients and 3 segments from NSTEMI patients were excluded from the analysis because the mean visual score was equal to 2 (indicating uncertainty for the presence of infarct) either on LGE images or T1 maps. Sensitivity and specificity for visual detection of chronic MIs using native T1 maps were 60% (54 of 90 true positives; 95% CI: 50% to 70%) and 86% (89 of 104 true negatives; 95% CI: 79% to 92%), respectively, in STEMI patients and 64% (47 of 73 true positives; 95% CI: 53% to 75%) and 91% (105 of 116 true negatives; 95% CI: 87% to 98%), respectively, in NSTEMI patients.
Recent studies in a canine model of MI have shown that native T1 mapping at 3-T can reliably characterize chronic MIs and can therefore be a potential alternative to LGE CMR (12). In this study, we investigated the clinical validity of the canine findings in 2 pilot patient populations with prior STEMI and NSTEMI, respectively. We found that the mean native T1 value of chronic MIs in both patient populations is approximately 21% higher than that of remote myocardium at 3-T. We also found that the threshold-based criterion applied to native T1 maps can be used to determine the chronic MI location, size, and transmurality with high diagnostic accuracy without contrast agents. Although the native T1 contrast enables accurate threshold-based detection, the technique had modest sensitivity but high specificity for visual detection of chronic infarction relative to LGE.
The feasibility of using native T1 mapping for characterizing chronic MIs in patients has been previously evaluated at 1.5-T (23,24). However, the low native T1 contrast between chronic MIs and remote myocardium resulted in poor diagnostic accuracy of native T1 maps. These results have been confirmed in a recent canine study (12). The same canine study showed that increasing the field strength from 1.5-T to 3-T markedly improves native T1 contrast for chronic MI detection; hence, our observation of improved diagnostic accuracy of T1 maps for detecting chronic MIs in patients compared with 1.5-T is consistent with animal studies.
Although the biophysical mechanisms responsible for such native T1 contrast in chronic MIs and its dependence on field strength remain to be elucidated, diffusion and magnetization transfer effects seem to be plausible explanations. Previous studies have shown that diffusion coefficient of protons in chronic MIs is ∼2-fold larger relative to remote myocardium (25) and that the coefficient of protons is directly related to T1(26). Chronic MIs are also known to have significantly reduced magnetization transfer effects (27). This could lead to an apparent increase in T1 values of chronic MIs when a magnetization transfer–sensitive T1 mapping sequence such as MOLLI-balanced steady-state free procession is used (28). The field-dependent magnetization transfer–based T1 bias (29) and T1 elongations (30) could potentially explain the increase in native T1 sensitivity and specificity that we observed in this study compared with the previous study in patients at 1.5-T (23).
Although native T1 maps have high specificity, their modest sensitivity for visually detecting chronic MIs relative to LGE may be limiting in some patients if visual analysis is the only mode used for MI characterization. Further increases in image contrast (for example, through inversion-recovery preparation to null the remote myocardium, analogous to inversion-recovery preparation used for LGE) needs to be investigated. Another potential difficulty with the proposed approach is image artifacts (off-resonance bands or motion). To mitigate these artifacts, careful cardiac shimming and optimal acquisition strategies need to be used. Moreover, partial volume effects, particularly in the apical slices, may complicate accurate characterization of relatively thin and scarred myocardium. This may lead to overestimation of the true infarct size and transmurality, but this limitation also exists with LGE imaging. A more stringent thresholding criterion and high-resolution native T1 mapping can potentially obviate these complications.
The ability of native T1 maps to detect pathologies beyond those detected on LGE images also remains to be explored. A particular pathology that we have consistently observed in T1 maps in some chronic MI patients is the focal decreases in T1 values (well below the T1 of normal myocardium and infarct) within the infarct. Based on previous studies in canines and acute MI patients (12,31), we suspect that this T1 loss may be due to chronic iron deposition following acute intramyocardial hemorrhage or fat deposition in healed MIs. This is further supported by recent reports that T1 is inversely related with both iron (32) and fat deposition (19–21) in MIs. Nonetheless, further studies are needed to examine the capability of native T1 maps to consistently detect chronic iron and fat depositions within the infarct.
First, this is a single-center study with a small sample size, but is comparable to previous studies that attempt to establish clinical validation (23,24). A multicenter study in a larger patient cohort is necessary to assess the reliability of the approach for clinical adoption. Second, because of imaging time constraints, we did not acquire native T2 maps to confirm the resolution of edema. Nevertheless, our exclusion criteria of not including patients with clinically recorded symptoms for MIs within 1 year before the CMR examination ensured that the observed MI territories were indeed chronic. These observations, along with recent pre-clinical studies (12), support the belief that native T1 elongations seen in chronic MIs in our patients are likely from replacement fibrosis. Third, to minimize imaging time, a frequency-scouting scheme was not used to reduce banding artifacts (33). Finally, for calculating the sensitivity and specificity of visual analysis, we have excluded segments rated as 2 because of reader ambiguity. Nevertheless, retrospective analyses including such segments as either positive or negative for infarction have shown no significant differences in sensitivities and specificities compared with those measured after the excluding the ambiguous segments.
Chronic MIs in STEMI and NSTEMI patients may be accurately characterized without contrast agents using threshold-based detection on native T1 maps. Our findings justify a study in a larger patient cohort.
COMPETENCY IN MEDICAL KNOWLEDGE: In relation to LGE CMR, native T1 mapping at 3-T can characterize chronic STEMIs and NSTEMIs with high diagnostic accuracy when semiautomated (threshold-based) detection is used. Native T1 mapping at 3-T can be a potential alternative to LGE CMR for characterizing chronic MIs in patients who are contraindicated for gadolinium administration.
TRANSLATIONAL OUTLOOK: Additional studies, preferably in a larger patient cohort, are also needed to assess the intraobserver and interobserver differences of native T1 mapping at 3-T for detecting and characterizing chronic MI. The ability of this technique to detect pathologies beyond those detected by LGE imaging also needs to be investigated. The modest sensitivity of native T1 mapping for visually detecting chronic MIs relative to LGE CMR at 3-T is a limitation, and future efforts should focus on improving image contrast.
This work was supported, in part, by grants from the American Heart Association (13PRE17210049); the National Heart, Lung, and Blood Institute (HL091989); and the National Research Foundation of Korea (MEST No. 2012027176). Dr. Dharmakumar and Mr. Kali are coinventors on a patent pending (PCT/US14/53938) on the use of T1 mapping at 3-T for characterizing chronic myocardial infarction. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Drs. Kali and E.-Y. Choi contributed equally to this work.
- Abbreviations and Acronyms
- American Heart Association
- confidence interval
- cardiac magnetic resonance
- contrast-to-noise ratio
- late gadolinium enhancement
- myocardial infarction
- interquartile range
- modified Look-Locker inversion recovery
- non–ST-segment elevation myocardial infarction
- signal intensity
- ST-segment elevation myocardial infarction
- Received March 2, 2015.
- Revision received April 26, 2015.
- Accepted April 29, 2015.
- American College of Cardiology Foundation
- Wu E.,
- Ortiz J.T.,
- Tejedor P.,
- et al.
- Bello D.,
- Fieno D.S.,
- Kim R.J.,
- et al.
- Kim R.J.,
- Fieno D.S.,
- Parrish T.B.,
- et al.
- Klem I.,
- Weinsaft J.W.,
- Bahnson T.D.,
- et al.
- Grobner T.
- Fox C.S.,
- Muntner P.,
- Chen A.Y.,
- et al.
- U.S. Renal Data System
- Kali A.,
- Cokic I.,
- Tang R.L.,
- et al.
- Amado L.C.,
- Gerber B.L.,
- Gupta S.N.,
- et al.
- Schulz-Menger J.,
- Bluemke D.A.,
- Bremerich J.,
- et al.
- Kali A.,
- Cokic I.,
- Yang H.J.,
- Sharif B.,
- Dharmakumar R.
- Ferreira V.M.,
- Holloway C.J.,
- Piechnik S.K.,
- Karamitsos T.D.,
- Neubauer S.
- Dall'Armellina E.,
- Ferreira V.M.,
- Kharbanda R.K.,
- et al.
- Kali A.,
- Kumar A.,
- Cokic I.,
- et al.