Deep learning formulation of electrocardiographic imaging integrating image and signal information with data-driven regularization.

AIMS: Electrocardiographic imaging (ECGI) is a promising tool to map the electrical activity of the heart non-invasively using body surface potentials (BSP). However, it is still challenging due to the mathematically ill-posed nature of the inverse problem to solve. Novel approaches leveraging progress in artificial intelligence could alleviate these difficulties. METHODS AND RESULTS: We propose a deep learning (DL) formulation of ECGI in order to learn the statistical relation between BSP and cardiac activation. The presented method is based on Conditional Variational AutoEncoders using deep generative neural networks. To quantify the accuracy of this method, we simulated activation maps and BSP data on six cardiac anatomies.We evaluated our model by training it on five different cardiac anatomies (5000 activation maps) and by testing it on a new patient anatomy over 200 activation maps. Due to the probabilistic property of our method, we predicted 10 distinct activation maps for each BSP data. The proposed method is able to generate volumetric activation maps with a good accuracy on the simulated data: the mean absolute error is 9.40 ms with 2.16 ms standard deviation on this testing set. CONCLUSION: The proposed formulation of ECGI enables to naturally include imaging information in the estimation of cardiac electrical activity from BSP. It naturally takes into account all the spatio-temporal correlations present in the data. We believe these features can help improve ECGI results.

Comparison of MR T1 and T2 mapping parameters to characterize myocardial and skeletal muscle involvement in systemic idiopathic inflammatory myopathy (IIM).

OBJECTIVES: To compare the performance of magnetic resonance (MR) relaxometry parameters to discriminate myocardial and skeletal muscle inflammation in idiopathic inflammatory myopathy (IIM) patients from healthy controls. MATERIALS AND METHODS: For this retrospective case-control study, 20 consecutive IIM patients (54 ± 18 years, 11 females) with cardiac involvement (troponin level > 50 ng/l) and 20 healthy controls (47 ± 12 years, 9 females) were included. All patients without cardiac MR imaging < 2 weeks prior to the laboratory testings were excluded. T1/T2 relaxation times, as well as T1-derived extracellular volume (ECV), relative tissue T1 shortening ΔT1 = (native T1tissue-post contrast T1tissue)/native T1tissue), and enhancement fraction EHF = (native T1tissue-post contrast T1tissue)/(native T1blood-post contrast T1blood), were compared using Mann-Whitney U test and ROC analysis. RESULTS: All measured MR relaxometry parameters significantly discriminated IIM patients and healthy controls, except T2 in skeletal muscles and ECV in the myocardium. In skeletal muscles, post contrast T1 and T1-derived parameters showed the best performance to discriminate IIM patients from healthy controls (AUC = 0.98 for post contrast T1 and AUC 0.94-0.97 for T1-derived parameters). Inversely, in the myocardium, native T1 and T2 showed better diagnostic performance (AUC = 0.89) than post contrast T1 (AUC = 0.76), ECV (AUC = 0.58), ΔT1 (AUC = 0.80) and EHF (0.82). CONCLUSIONS: MR relaxometry parameters applied to the myocardium and skeletal muscles might be useful to separate IIM patients from healthy controls. However, different tissue composition and vascularization should be taken into account for their interpretation. ΔT1 and EHF may be simple alternatives to ECV in highly vascularized tissues such as the myocardium. KEY POINTS: • MR relaxometry parameters applied to the myocardium and skeletal muscles are highly useful to separate IIM patients from healthy controls. • Different tissue composition and vascularization should be taken into account for T1 and T2 mapping parameter interpretation. • ΔT1 and EHF may be simple alternatives to ECV in highly vascularized tissues such as the myocardium.

Non-invasive differentiation of idiopathic inflammatory myopathy with cardiac involvement from acute viral myocarditis using cardiovascular magnetic resonance imaging T1 and T2 mapping.

BACKGROUND: Idiopathic inflammatory myopathy (IIM) is a group of autoimmune diseases with systemic myositis which may involve the myocardium. Cardiac involvement in IIM, although often subclinical, may mimic clinical manifestations of acute viral myocarditis (AVM). Our aim was to investigate the usefulness of the combined analysis of cardiovascular magnetic resonance (CMR) T1 and T2 mapping parameters measured both in the myocardium and in the thoracic skeletal muscles to differentiate AVM from IIM cardiac involvement. METHODS: Sixty subjects were included in this retrospective study (36 male, age 45 ± 16 years): twenty patients with AVM, twenty patients with IIM and cardiac involvement and twenty healthy controls. Study participants underwent CMR imaging with modified Look-Locker inversion-recovery (MOLLI) T1 mapping and 3-point balanced steady-state-free precession T2 mapping. Relaxation times were quantified after endocardial and epicardial delineation on basal and medial short-axis slices, as well as in different thoracic skeletal muscle groups present in the CMR field-of-view. ROC-Analysis was performed to assess the ability of mapping indices to discriminate the study groups. RESULTS: Mapping parameters in the thoracic skeletal muscles were able to discriminate between AVM and IIM patients. Best skeletal muscle parameters to identify IIM from AVM patients were reduced post-contrast T1 and increased extracellular volume (ECV), resulting in an area under the ROC curve (AUC) of 0.95 for post-contrast T1 and 0.96 for ECV. Conversely, myocardial mapping parameters did not discriminate IIM from AVM patients but increased native T1 (AUC 0.89 for AVM; 0.84 for IIM) and increased T2 (AUC 0.82 for AVM; 0.88 for IIM) could differentiate both patient groups from healthy controls. CONCLUSION: CMR myocardial mapping detects cardiac inflammation in AVM and IIM compared to normal myocardium in healthy controls but does not differentiate IIM from AVM. However, thoracic skeletal muscle mapping was able to accurately discern IIM from AVM.