Multicenter Study on the Diagnostic Performance of Native-T1 Cardiac Magnetic Resonance of Chronic Myocardial Infarctions at 3T

Native T1 Mapping for the Diagnosis of Myocardial Fibrosis in Patients With Chronic Myocardial Infarction

BACKGROUND

Myocardial fibrosis is a fundamental process in cardiac injury. Cardiac magnetic resonance native T1 mapping has been proposed for diagnosing myocardial fibrosis without the need for gadolinium contrast. However, recent studies suggest that T1 measurements can be erroneous in the presence of intramyocardial fat.

OBJECTIVES

The purpose of this study was to investigate whether the presence of fatty metaplasia affects the accuracy of native T1 maps for the diagnosis of myocardial replacement fibrosis in patients with chronic myocardial infarction (MI).

METHODS

Consecutive patients (n = 312) with documented chronic MI (>6 months old) and controls without MI (n = 50) were prospectively enrolled. Presence and size of regions with elevated native T1 and infarction were quantitatively determined (mean + 5SD) on modified look-locker inversion-recovery and delayed-enhancement images, respectively, at 3.0-T. The presence of fatty metaplasia was determined using an out-of-phase steady-state free-precession cine technique and further verified with standard fat-water Dixon methods.

RESULTS

Native T1 mapping detected chronic MI with markedly higher sensitivity in patients with fatty metaplasia than those without fatty metaplasia (85.6% vs 13.3%) with similar specificity (100% vs 98.9%). In patients with fatty metaplasia, the size of regions with elevated T1 significantly underestimated infarct size and there was a better correlation with fatty metaplasia size than infarct size (r = 0.76 vs r = 0.49). In patients without fatty metaplasia, most of the modest elevation in T1 appeared to be secondary to subchronic infarcts that were 6 to 12 months old; the T1 of infarcts >12 months old was not different from noninfarcted myocardium.

CONCLUSIONS

Native T1 mapping is poor at detecting replacement fibrosis but may indirectly detect chronic MI if there is associated fatty metaplasia. Native T1 mapping for the diagnosis and characterization of myocardial fibrosis is unreliable.

Artificial Intelligence for Contrast-Free MRI: Scar Assessment in Myocardial Infarction Using Deep Learning-Based Virtual Native Enhancement

BACKGROUND

Myocardial scars are assessed noninvasively using cardiovascular magnetic resonance late gadolinium enhancement (LGE) as an imaging gold standard. A contrast-free approach would provide many advantages, including a faster and cheaper scan without contrast-associated problems.

METHODS

Virtual native enhancement (VNE) is a novel technology that can produce virtual LGE-like images without the need for contrast. VNE combines cine imaging and native T1 maps to produce LGE-like images using artificial intelligence. VNE was developed for patients with previous myocardial infarction from 4271 data sets (912 patients); each data set comprises slice position-matched cine, T1 maps, and LGE images. After quality control, 3002 data sets (775 patients) were used for development and 291 data sets (68 patients) for testing. The VNE generator was trained using generative adversarial networks, using 2 adversarial discriminators to improve the image quality. The left ventricle was contoured semiautomatically. Myocardial scar volume was quantified using the full width at half maximum method. Scar transmurality was measured using the centerline chord method and visualized on bull's-eye plots. Lesion quantification by VNE and LGE was compared using linear regression, Pearson correlation (R), and intraclass correlation coefficients. Proof-of-principle histopathologic comparison of VNE in a porcine model of myocardial infarction also was performed.

RESULTS

VNE provided significantly better image quality than LGE on blinded analysis by 5 independent operators on 291 data sets (all P<0.001). VNE correlated strongly with LGE in quantifying scar size (R, 0.89; intraclass correlation coefficient, 0.94) and transmurality (R, 0.84; intraclass correlation coefficient, 0.90) in 66 patients (277 test data sets). Two cardiovascular magnetic resonance experts reviewed all test image slices and reported an overall accuracy of 84% for VNE in detecting scars when compared with LGE, with specificity of 100% and sensitivity of 77%. VNE also showed excellent visuospatial agreement with histopathology in 2 cases of a porcine model of myocardial infarction.

CONCLUSIONS

VNE demonstrated high agreement with LGE cardiovascular magnetic resonance for myocardial scar assessment in patients with previous myocardial infarction in visuospatial distribution and lesion quantification with superior image quality. VNE is a potentially transformative artificial intelligence-based technology with promise in reducing scan times and costs, increasing clinical throughput, and improving the accessibility of cardiovascular magnetic resonance in the near future.

SAturation-recovery and Variable-flip-Angle-based three-dimensional free-breathing cardiovascular magnetic resonance T1 mapping at 3 T

The purpose of the current study was to develop and validate a three-dimensional (3D) free-breathing cardiac T₁ -mapping sequence using SAturation-recovery and Variable-flip-Angle (SAVA). SAVA sequentially acquires multiple electrocardiogram-triggered volumes using a multishot spoiled gradient-echo sequence. The first volume samples the equilibrium signal of the longitudinal magnetization, where a flip angle of 2° is used to reduce the time for the magnetization to return to equilibrium. The succeeding three volumes are saturation prepared with variable delays, and are acquired using a 15° flip angle to maintain the signal-to-noise ratio. A diaphragmatic navigator is used to compensate the respiratory motion. T₁ is calculated using a saturation-recovery model that accounts for the flip angle. We validated SAVA by simulations, phantom, and human subject experiments at 3 T. SAVA was compared with modified Look-Locker inversion recovery (MOLLI) and saturation-recovery single-shot acquisition (SASHA) in vivo. In phantoms, T₁ by SAVA had good agreement with the reference (R²  = 0.99). In vivo 3D T₁ mapping by SAVA could achieve an imaging resolution of 1.25 × 1.25 × 8 mm³ . Both global and septal T₁ values by SAVA (1347 ± 37 and 1332 ± 42 ms) were in between those by SASHA (1612 ± 63 and 1618 ± 51 ms) and MOLLI (1143 ± 59 and 1188 ± 65 ms). According to the standard deviation (SD) and coefficient of variation (CV), T₁ precision measured by SAVA (SD: 99 ± 14 and 60 ± 8 ms; CV: 7.4% ± 0.9% and 4.5% ± 0.6%) was comparable with MOLLI (SD: 99 ± 25 and 46 ± 12 ms; CV: 8.8% ± 2.5% and 3.9% ± 1.1%) and superior to SASHA (SD: 222 ± 89 and 132 ± 33 ms; CV: 13.8% ± 5.5% and 8.1% ± 2.0%). It was concluded that the proposed free-breathing SAVA sequence enables more efficient 3D whole-heart T₁ estimation with good accuracy and precision.

Myocardial blood flow is the dominant factor influencing cardiac magnetic resonance adenosine stress T2

Stress imaging identifies ischemic myocardium by comparing hemodynamics during rest and hyperemic stress. Hyperemia affects multiple hemodynamic parameters in myocardium, including myocardial blood flow (MBF), myocardial blood volume (MBV), and venous blood oxygen levels (PvO2 ). Cardiac T2 is sensitive to these changes and therefore is a promising non-contrast option for stress imaging; however, the impact of individual hemodynamic factors on T2 is poorly understood, making the connection from altered T2 to changes within the tissue difficult. To better understand this interplay, we performed T2 mapping and measured various hemodynamic factors independently in healthy pigs at multiple levels of hyperemic stress, induced by different doses of adenosine (0.14-0.56 mg/kg/min). T1 mapping quantified changes in MBV. MBF was assessed with microspheres, and oxygen consumption was determined by the rate pressure product (RPP). Simulations were also run to better characterize individual contributions to T2. Myocardial T2, MBF, oxygen consumption, and MBV all changed to varying extents between each level of adenosine stress (T2 = 37.6-41.8 ms; MBF = 0.48-1.32 mL/min/g; RPP = 6507-4001 bmp*mmHg; maximum percent change in MBV = 1.31%). Multivariable analyses revealed MBF as the dominant influence on T2 during hyperemia (significant β-values >7). Myocardial oxygen consumption had almost no effect on T2 (β-values <0.002); since PvO2 is influenced by both oxygen consumption and MBF, PvO2 changes detected by T2 during adenosine stress can be attributed to MBF. Simulations varying PvO2 and MBV confirmed that PvO2 had the strongest influence on T2, but MBV became important at high PvO2 . Together, these data suggest a model where, during adenosine stress, myocardial T2 responds predominantly to changes in MBF, but at high hyperemia MBV is also influential. Thus, changes in adenosine stress T2 can now be interpreted in terms of the physiological changes that led to it, enabling T2 mapping to become a viable non-contrast option to detect ischemic myocardial tissue.

Segmental strain analysis for the detection of chronic ischemic scars in non-contrast cardiac MRI cine images

Cardiac magnetic resonance imaging (MRI) with late gadolinium enhancement (LGE) is considered the gold standard for scar detection after myocardial infarction. In times of increasing skepticism about gadolinium depositions in brain tissue and contraindications of gadolinium administration in some patient groups, tissue strain-based techniques for detecting ischemic scars should be further developed as part of clinical protocols. Therefore, the objective of the present work was to investigate whether segmental strain is noticeably affected in chronic infarcts and thus can be potentially used for infarct detection based on routinely acquired non-contrast cine images in patients with known coronary artery disease (CAD). Forty-six patients with known CAD and chronic scars in LGE images (5 female, mean age 52 ± 19 years) and 24 gender- and age-matched controls with normal cardiac MRI (2 female, mean age 47 ± 13 years) were retrospectively enrolled. Global (global peak circumferential [GPCS], global peak longitudinal [GPLS], global peak radial strain [GPRS]) and segmental (segmental peak circumferential [SPCS], segmental peak longitudinal [SPLS], segmental peak radial strain [SPRS]) strain parameters were calculated from standard non-contrast balanced SSFP cine sequences using commercially available software (Segment CMR, Medviso, Sweden). Visual wall motion assessment of short axis cine images as well as segmental circumferential strain calculations (endo-/epicardially contoured short axis cine and resulting polar plot strain map) of every patient and control were presented in random order to two independent blinded readers, which should localize potentially infarcted segments in those datasets blinded to LGE images and patient information. Global strain values were impaired in patients compared to controls (GPCS p = 0.02; GPLS p = 0.04; GPRS p = 0.01). Patients with preserved ejection fraction showed also impeded GPCS compared to healthy individuals (p = 0.04). In patients, mean SPCS was significantly impaired in subendocardially (-  5.4% ± 2) and in transmurally infarcted segments (- 1.2% ± 3) compared to remote myocardium (- 12.9% ± 3, p = 0.02 and 0.03, respectively). ROC analysis revealed an optimal cut-off value for SPCS for discriminating infarcted from remote myocardium of - 7.2% with a sensitivity of 89.4% and specificity of 85.7%. Mean SPRS was impeded in transmurally infarcted segments (15.9% ± 6) compared to SPRS of remote myocardium (31.4% ± 5; p = 0.02). The optimal cut-off value for SPRS for discriminating scar tissue from remote myocardium was 16.6% with a sensitivity of 83.3% and specificity of 76.5%. 80.3% of all in LGE infarcted segments (118/147) were correctly localized in segmental circumferential strain calculations based on non-contrast cine images compared to 53.7% (79/147) of infarcted segments detected by visual wall motion assessment (p > 0.01). Global strain parameters are impaired in patients with chronic infarcts compared to controls. Mean SPCS and SPRS in scar tissue is impeded compared to remote myocardium in infarcts patients. Blinded to LGE images, two readers correctly localized 80% of infarcted segments in segmental circumferential strain calculations based on non-contrast cine images, in contrast to only 54% of infarcted segments detected due to wall motion abnormalities in visual wall motion assessment. Analysis of segmental circumferential strain shows a promising method for detection of chronic scars in routinely acquired, non-contrast cine images for patients who cannot receive or decline gadolinium.

Machine learning phenotyping of scarred myocardium from cine in hypertrophic cardiomyopathy

AIMS

Cardiovascular magnetic resonance (CMR) with late-gadolinium enhancement (LGE) is increasingly being used in hypertrophic cardiomyopathy (HCM) for diagnosis, risk stratification, and monitoring. However, recent data demonstrating brain gadolinium deposits have raised safety concerns. We developed and validated a machine-learning (ML) method that incorporates features extracted from cine to identify HCM patients without fibrosis in whom gadolinium can be avoided.

METHODS AND RESULTS

An XGBoost ML model was developed using regional wall thickness and thickening, and radiomic features of myocardial signal intensity, texture, size, and shape from cine. A CMR dataset containing 1099 HCM patients collected using 1.5T CMR scanners from different vendors and centres was used for model development (n=882) and validation (n=217). Among the 2613 radiomic features, we identified 7 features that provided best discrimination between +LGE and -LGE using 10-fold stratified cross-validation in the development cohort. Subsequently, an XGBoost model was developed using these radiomic features, regional wall thickness and thickening. In the independent validation cohort, the ML model yielded an area under the curve of 0.83 (95% CI: 0.77-0.89), sensitivity of 91%, specificity of 62%, F1-score of 77%, true negatives rate (TNR) of 34%, and negative predictive value (NPV) of 89%. Optimization for sensitivity provided sensitivity of 96%, F2-score of 83%, TNR of 19% and NPV of 91%; false negatives halved from 4% to 2%.

CONCLUSION

An ML model incorporating novel radiomic markers of myocardium from cine can rule-out myocardial fibrosis in one-third of HCM patients referred for CMR reducing unnecessary gadolinium administration.