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.

Non-invasive assessment of portal hypertension by multi-parametric magnetic resonance imaging of the spleen: A proof of concept study.

BACKGROUND AND AIMS: Non-invasive assessment of portal hypertension is an area of unmet need. This proof of concept study aimed to evaluate the diagnostic accuracy of a multi-parametric magnetic resonance technique in the assessment of portal hypertension. Comparison to other non-invasive technologies was a secondary aim. METHODS: T1 and T2* maps through the liver and spleen were acquired prior to trans-jugular liver biopsy and hepatic vein pressure gradient (HVPG) measurement. T1 measurements reflect changes in tissue water content, but this relationship is confounded by the presence of iron, which in turn can be quantified accurately from T2* maps. Data were analysed using LiverMultiScan (Perspectum Diagnostics, Oxford, UK) which applies an algorithm to remove the confounding effect of iron, yielding the "iron corrected T1" (cT1). Sensitivity, specificity, diagnostic values and area under the curve were derived for spleen cT1, liver cT1, transient elastography, and serum fibrosis scores. HVPG was the reference standard. RESULTS: Nineteen patients (15 men) with median age 57 years were included. Liver disease aetiologies included non-alcoholic fatty liver disease (n = 9; 47%) and viral hepatitis (n = 4; 21%). There was strong correlation between spleen cT1 and HVPG (r = 0.69; p = 0.001). Other non-invasive biomarkers did not correlate with HVPG. Spleen cT1 had excellent diagnostic accuracy for portal hypertension (HVPG >5 mmHg) and clinically significant portal hypertension (HVPG ≥10 mmHg) with an area under the receiver operating characteristic curve of 0.92 for both. CONCLUSION: Spleen cT1 is a promising biomarker of portal pressure that outperforms other non-invasive scores and should be explored further.

Reproducibility of native myocardial T1 mapping in the assessment of Fabry disease and its role in early detection of cardiac involvement by cardiovascular magnetic resonance.

BACKGROUND: Cardiovascular magnetic resonance (CMR) derived native myocardial T1 is decreased in patients with Fabry disease even before left ventricular hypertrophy (LVH) occurs and may be the first non-invasive measure of myocyte sphingolipid storage. The relationship of native T1 lowering prior to hypertrophy and other candidate early phenotype markers are unknown. Furthermore, the reproducibility of T1 mapping has never been assessed in Fabry disease. METHODS: Sixty-three patients, 34 (54%) female, mean age 48±15 years with confirmed (genotyped) Fabry disease underwent CMR, ECG and echocardiographic assessment. LVH was absent in 25 (40%) patients. Native T1 mapping was performed with both Modified Look-Locker Inversion recovery (MOLLI) sequences and a shortened version (ShMOLLI) at 1.5 Tesla. Twenty-one patients underwent a second scan within 24 hours to assess inter-study reproducibility. Results were compared with 63 healthy age and gender-matched volunteers. RESULTS: Mean native T1 in Fabry disease (LVH positive), (LVH negative) and healthy volunteers was 853±50 ms, 904±46 ms and 968±32 ms (for all p<0.0001) by ShMOLLI sequences. Native T1 showed high inter-study, intra-observer and inter-observer agreement with intra-class correlation coefficients (ICC) of 0.99, 0.98, 0.97 (ShMOLLI) and 0.98, 0.98, 0.98 (MOLLI). In Fabry disease LVH negative individuals, low native T1 was associated with reduced echocardiographic-based global longitudinal speckle tracking strain (-18±2% vs -22±2%, p=0.001) and early diastolic function impairment (E/E'=7 [6-8] vs 5 [5-6], p=0.028). CONCLUSION: Native T1 mapping in Fabry disease is a reproducible technique. T1 reduction prior to the onset of LVH is associated with early diastolic and systolic changes measured by echocardiography.

Identification and assessment of Anderson-Fabry disease by cardiovascular magnetic resonance noncontrast myocardial T1 mapping.

BACKGROUND: Anderson-Fabry disease (AFD) is a rare but underdiagnosed intracellular lipid disorder that can cause left ventricular hypertrophy (LVH). Lipid is known to shorten the magnetic resonance imaging parameter T1. We hypothesized that noncontrast T1 mapping by cardiovascular magnetic resonance would provide a novel and useful measure in this disease with potential to detect early cardiac involvement and distinguish AFD LVH from other causes. METHODS AND RESULTS: Two hundred twenty-seven subjects were studied: patients with AFD (n=44; 55% with LVH), healthy volunteers (n=67; 0% with LVH), patients with hypertension (n=41; 24% with LVH), patients with hypertrophic cardiomyopathy (n=34; 100% with LVH), those with severe aortic stenosis (n=21; 81% with LVH), and patients with definite amyloid light-chain (AL) cardiac amyloidosis (n=20; 100% with LVH). T1 mapping was performed using the shortened modified Look-Locker inversion sequence on a 1.5-T magnet before gadolinium administration with primary results derived from the basal and midseptum. Compared with health volunteers, septal T1 was lower in AFD and higher in other diseases (AFD versus healthy volunteers versus other patients, 882±47, 968±32, 1018±74 milliseconds; P<0.0001). In patients with LVH (n=105), T1 discriminated completely between AFD and other diseases with no overlap. In AFD, T1 correlated inversely with wall thickness (r=-0.51; P=0.0004) and was abnormal in 40% of subjects who did not have LVH. Segmentally, AFD showed pseudonormalization or elevation of T1 in the left ventricular inferolateral wall, correlating with the presence or absence of late gadolinium enhancement (1001±82 versus 891±38 milliseconds; P<0.0001). CONCLUSIONS: Noncontrast T1 mapping shows potential as a unique and powerful measurement in the imaging assessment of LVH and AFD.

Cardiovascular magnetic resonance by non contrast T1-mapping allows assessment of severity of injury in acute myocardial infarction.

BACKGROUND: Current cardiovascular magnetic resonance (CMR) methods, such as late gadolinium enhancement (LGE) and oedema imaging (T2W) used to depict myocardial ischemia, have limitations. Novel quantitative T1-mapping techniques have the potential to further characterize the components of ischemic injury. In patients with myocardial infarction (MI) we sought to investigate whether state-of the art pre-contrast T1-mapping (1) detects acute myocardial injury, (2) allows for quantification of the severity of damage when compared to standard techniques such as LGE and T2W, and (3) has the ability to predict long term functional recovery. METHODS: 3T CMR including T2W, T1-mapping and LGE was performed in 41 patients [of these, 78% were ST elevation MI (STEMI)] with acute MI at 12-48 hour after chest pain onset and at 6 months (6M). Patients with STEMI underwent primary PCI prior to CMR. Assessment of acute regional wall motion abnormalities, acute segmental damaged fraction by T2W and LGE and mean segmental T1 values was performed on matching short axis slices. LGE and improvement in regional wall motion at 6M were also obtained. RESULTS: We found that the variability of T1 measurements was significantly lower compared to T2W and that, while the diagnostic performance of acute T1-mapping for detecting myocardial injury was at least as good as that of T2W-CMR in STEMI patients, it was superior to T2W imaging in NSTEMI. There was a significant relationship between the segmental damaged fraction assessed by either by LGE or T2W, and mean segmental T1 values (P < 0.01). The index of salvaged myocardium derived by acute T1-mapping and 6M LGE was not different to the one derived from T2W (P = 0.88). Furthermore, the likelihood of improvement of segmental function at 6M decreased progressively as acute T1 values increased (P < 0.0004). CONCLUSIONS: In acute MI, pre-contrast T1-mapping allows assessment of the extent of myocardial damage. T1-mapping might become an important complementary technique to LGE and T2W for identification of reversible myocardial injury and prediction of functional recovery in acute MI.