Background: Late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) imaging is the gold standard for non-invasive myocardial tissue characterization, but requires intravenous contrast agent administration. It is highly desired to develop a contrast-agent-free technology to replace LGE for faster and cheaper CMR scans. Methods: A CMR Virtual Native Enhancement (VNE) imaging technology was developed using artificial intelligence. The deep learning model for generating VNE uses multiple streams of convolutional neural networks to exploit and enhance the existing signals in native T1-maps (pixel-wise maps of tissue T1 relaxation times) and cine imaging of cardiac structure and function, presenting them as LGE-equivalent images. The VNE generator was trained using generative adversarial networks. This technology was first developed on CMR datasets from the multi-center Hypertrophic Cardiomyopathy Registry (HCMR), using HCM as an exemplar. The datasets were randomized into two independent groups for deep learning training and testing. The test data of VNE and LGE were scored and contoured by experienced human operators to assess image quality, visuospatial agreement and myocardial lesion burden quantification. Image quality was compared using nonparametric Wilcoxon test. Intra- and inter-observer agreement was analyzed using intraclass correlation coefficients (ICC). Lesion quantification by VNE and LGE were compared using linear regression and ICC. Results: 1348 HCM patients provided 4093 triplets of matched T1-maps, cines, and LGE datasets. After randomization and data quality control, 2695 datasets were used for VNE method development, and 345 for independent testing. VNE had significantly better image quality than LGE, as assessed by 4 operators (n=345 datasets, p<0.001, Wilcoxon test). VNE revealed characteristic HCM lesions in high visuospatial agreement with LGE. In 121 patients (n=326 datasets), VNE correlated with LGE in detecting and quantifying both hyper-intensity myocardial lesions (r=0.77-0.79, ICC=0.77-0.87; p<0.001) and intermediate-intensity lesions (r=0.70-0.76, ICC=0.82-0.85; p<0.001). The native CMR images (cine plus T1-map) required for VNE can be acquired within 15 minutes. Producing a VNE image takes less than one second. Conclusions: VNE is a new CMR technology that resembles conventional LGE, without the need for contrast administration. VNE achieved high agreement with LGE in the distribution and quantification of lesions, with significantly better image quality.

Background: Quantitative cardiovascular magnetic resonance T1-mapping is increasingly used for myocardial tissue characterization. However, the lack of standardization limits direct comparability between centers and wider roll-out for clinical use or trials. Purpose: To develop a quality assurance (QA) program assuring standardized T1 measurements for clinical use. Methods: MR phantoms manufactured in 2013 were distributed, including ShMOLLI T1-mapping and reference T1 and T2 protocols. We first studied the T1 and T2 dependency on temperature and phantom aging using phantom datasets from a single site over 4 years. Based on this, we developed a multiparametric QA model, which was then applied to 78 scans from 28 other multi-national sites. Results: T1 temperature sensitivity followed a second-order polynomial to baseline T1 values (R 2 > 0.996). Some phantoms showed aging effects, where T1 drifted up to 49% over 40 months. The correlation model based on reference T1 and T2, developed on 1004 dedicated phantom scans, predicted ShMOLLI-T1 with high consistency (coefficient of variation 1.54%), and was robust to temperature variations and phantom aging. Using the 95% confidence interval of the correlation model residuals as the tolerance range, we analyzed 390 ShMOLLI T1-maps and confirmed accurate sequence deployment in 90%(70/78) of QA scans across 28 multiple centers, and categorized the rest with specific remedial actions. Conclusions: The proposed phantom QA for T1-mapping can assure correct method implementation and protocol adherence, and is robust to temperature variation and phantom aging. This QA program circumvents the need of frequent phantom replacements, and can be readily deployed in multicenter trials.

Background: Myocardial scar is currently assessed non-invasively using cardiovascular magnetic resonance (CMR) late gadolinium enhancement (LGE) as an imaging gold-standard. However, 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 (AI). VNE was developed for patients with prior myocardial infarction on 4271 datasets (912 patients), where each dataset is comprised of slice position-matched cine, T1-maps and LGE images. After quality control, 3002 datasets (775 patients) were used for development, and 291 datasets (68 patients) for testing. The VNE generator was trained using generative adversarial networks, employing two adversarial discriminators to improve the image quality. The left ventricle was contoured semi-automatically. 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 were compared using linear regression, Pearson correlation (R) and intraclass correlation coefficients (ICC). Proof-of-principle histopathological comparison of VNE in a porcine model of myocardial infarction was also performed. Results: VNE provided significantly better image quality than LGE on blinded analysis by 5 independent operators on 291 datasets (all p<0.001). VNE correlated strongly with LGE in quantifying scar size (R=0.89, ICC=0.94) and transmurality (R=0.84, ICC=0.90) in 66 patients (277 test datasets). Two CMR experts reviewed all test image slices and reported an overall accuracy of 84% of VNE in detecting scar 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-CMR for myocardial scar assessment in patients with prior myocardial infarction in visuospatial distribution and lesion quantification, with superior image quality. VNE is a potentially transformative AI-based technology, with promise to reduce scan times and costs, increase clinical throughput, and improve the accessibility of CMR in the very-near future.

Recent progress on deep learning (DL)-based medical image segmentation can enable fast extraction of clinical parameters for efficient clinical workflows. However, current DL methods can still fail and require manual visual inspection of outputs, which is time-consuming and diminishes the advantages of automation. For clinical applications, it is essential to develop DL approaches that can not only perform accurate segmentation, but also predict the segmentation quality and flag poor-quality results to avoid errors in diagnosis. To achieve robust performance, DL-based methods often require large datasets, which are not always readily available. It would be highly desirable to be able to train DL models using only small datasets, but this requires a quality prediction method to ensure reliability. We present a novel segmentation framework utilizing an ensemble of deep convolutional neural networks with Monte Carlo sampling. The proposed framework merges the advantages of both state-of-the-art deep ensembles and Bayesian approaches, to provide robust segmentation with inherent quality control. We successfully developed and tested this framework using just a small MRI dataset of 45 subjects. The framework obtained high mean Dice similarity coefficients (DSC) for segmentation of the endocardium (0.922) and the epicardium (0.942); importantly, segmentation DSC can be accurately predicted with low mean absolute errors (≤0.035), in the absence of the manual ground truth. Furthermore, binary classification of segmentation quality achieved a near-perfect accuracy of 99%. The proposed framework can enable fast and reliable medical image analysis with accurate quality control, and training of DL-based methods using even small datasets.