Background— The relation between infarct tissue heterogeneity on contrast-enhanced MRI and the occurrence of spontaneous ventricular arrhythmia (or sudden cardiac death) is unknown. Therefore, the ...

A model-based method for three-dimensional image segmentation was developed and its performance assessed in segmentation of volumetric cardiac magnetic resonance (MR) images and echocardiographic temporal image sequences. Comprehensive design of a three-dimensional (3-D) active appearance model (AAM) is reported for the first time as an involved extension of the AAM framework introduced by Cootes et al. The model's behavior is learned from manually traced segmentation examples during an automated training stage. Information about shape and image appearance of the cardiac structures is contained in a single model. This ensures a spatially and/or temporally consistent segmentation of three-dimensional cardiac images. The clinical potential of the 3-D AAM is demonstrated in short-axis cardiac MR images and four-chamber echocardiographic sequences. The method's performance was assessed by comparison with manually identified independent standards in 56 clinical MR and 64 clinical echo image sequences. The AAM method showed good agreement with the independent standard using quantitative indexes of border positioning errors, endo- and epicardial volumes, and left ventricular mass. In MR, the endocardial volumes, epicardial volumes, and left ventricular wall mass correlation coefficients between manual and AAM were R/sup 2/=0.94,0.97,0.82, respectively. For echocardiographic analysis, the area correlation was R/sup 2/=0.79. The AAM method shows high promise for successful application to MR and echocardiographic image analysis in a clinical setting.

Our results provide histologic validation for the use of postcontrast T1 mapping to noninvasively assess interstitial myocardial fibrosis in patients with clinically evident cardiomyopathy. PurposeTo determine the utility of cardiac magnetic resonance (MR) T1 mapping for quantification of diffuse myocardial fibrosis compared with the standard of endomyocardial biopsy.Materials and Met...

Substrate-based ablation of ventricular tachycardia (VT) relies on electroanatomical voltage mapping (EAVM). Integration of scar information from contrast-enhanced magnetic resonance imaging (CE-MRI) with EAVM may provide supplementary information. This study assessed the relation between electrogram voltages and CE-MRI scar characteristics using real-time integration and reversed registration. Fifteen patients without implantable cardiac defibrillator (14 males, 64 ± 9 years) referred for VT ablation after myocardial infarction underwent CE-MRI. Contours of the CE-MRI were used to create three-dimensional surface meshes of the left ventricle (LV), aortic root, and left main stem (LM). Real-time integration of CE-MRI-derived scar meshes with EAVM of the LV and aortic root was performed using the LM and the CARTO surface registration algorithm. Merging of CE-MRI meshes with EAVM was successful with a registration error of 3.8 ± 0.6 mm. After the procedure, voltage amplitudes of each mapping point were superimposed on the corresponding CE-MRI location using the reversed registration matrix. Infarcts on CE-MRI were categorized by transmurality and signal intensity. Local bipolar and unipolar voltages decreased with increasing scar transmurality and were influenced by scar heterogeneity. Ventricular tachycardia reentry circuit isthmus sites were correlated to CE-MRI scar location. In three patients, VT isthmus sites were located in scar areas not identified by EAVM. Integration of MRI-derived scar maps with EAVM during VT ablation is feasible and accurate. Contrast-enhanced magnetic resonance imaging identifies non-transmural scars and infarct grey zones not detected by EAVM according to the currently used voltage criteria and may provide important supplementary substrate information in selected patients.

A deep learning–based algorithm can achieve fully automated, accurate left ventricle segmentation from short-axis cine MR images when tested against a multivendor, multicenter data set. PurposeTo develop a deep learning–based method for fully automated quantification of left ventricular (LV) function from short-axis cine MR images and to evaluate its performance in a multivendor and multicenter setting.Materials and MethodsThis retrospective study included cine MRI data sets obtained from three major MRI vendors in four medical centers from 2008 to 2016. Three convolutional neural networks (CNNs) with the U-NET architecture were trained on data sets of increasing variability: (a) a single-vendor, single-center, homogeneous cohort of 100 patients (CNN1); (b) a single-vendor, multicenter, heterogeneous cohort of 200 patients (CNN2); and (c) a multivendor, multicenter, heterogeneous cohort of 400 patients (CNN3). All CNNs were tested on an independent multivendor, multicenter data set of 196 patients. CNN performance was evaluated with respect to the manual annotations from three experienced observers in terms of (a) LV detection accuracy, (b) LV segmentation accuracy, and (c) LV functional...