Search

Introduction: We previously developed an LGE-MRI-based virtual-heart arrhythmia ablation targeting (VAAT) methodology to non-invasively determine potential ablation targets for infarct-related VT. However, it is unknown whether VAAT’s predictions correspond with surface ECG predictions. Hypothesis: We hypothesized that the VAAT predicted VT circuits and potential ablation lesions would co-localize with ECG-based VT-exit predictions from a previously validated population-derived automated VT exit localization (PAVEL) system. Methods: We retrospectively enrolled 5 post-infarct patients who underwent LV endocardial VT ablation and had pre-procedural 2D LGE-MRIs. The PAVEL system based on a population-derived statistical method was used to localize VT-exit sites onto one of 238 triangles on the patient-specific virtual-heart LV endocardial surface using 8 independent ECG leads (I, II, V1-V6). The VAAT methodology incorporating patient-specific scar and infarct border zone distributions was used to identify potential VT circuits and find ablation lesions. Results: Eleven induced VTs were analyzed. Ten VT-exit sites were localized onto the patient-specific virtual-heart LV endocardial surface by the PAVEL system, and were used for the comparisons. One VT-exit site was too basal to be localized onto the virtual-heart geometry. The spatial resolution of the 10 predicted VT-exit sites was 13.8 ± 1.8 mm. VAAT-predicted VT circuits and ablation lesions correlated well with all 10 predicted VT-exit sites. Lastly, VAAT ablation lesions fell within the regions ablated clinically. Conclusions: The VAAT-predicted VT circuits and ablation lesions matched VT-exit sites predicted by the surface ECG-based PAVEL system. Combining these two complementary technologies may improve accuracy for non-invasively identifying optimal ablation targets to increase ablation efficacy.

Deep learning networks have shown state-of-the-art performance in many image reconstruction problems. However, it is not well understood what properties of representation and learning may improve the generalization ability of the network. In this paper, we propose that the generalization ability of an encoder-decoder network for inverse reconstruction can be improved in two means. First, drawing from analytical learning theory, we theoretically show that a stochastic latent space will improve the ability of a network to generalize to test data outside the training distribution. Second, following the information bottleneck principle, we show that a latent representation minimally informative of the input data will help a network generalize to unseen input variations that are irrelevant to the output reconstruction. Therefore, we present a sequence image reconstruction network optimized by a variational approximation of the information bottleneck principle with stochastic latent space. In the application setting of reconstructing the sequence of cardiac transmembrane potential from body-surface potential, we assess the two types of generalization abilities of the presented network against its deterministic counterpart. The results demonstrate that the generalization ability of an inverse reconstruction network can be improved by stochasticity as well as the information bottleneck.

Noninvasive reconstruction of cardiac electrical activity from surface electrocardiograms (ECG) involves solving an ill-posed inverse problem. Cardiac electrophysiological (EP) models have been used as important a priori knowledge to constrain this inverse problem. However, the reconstruction suffer from inaccuracy and uncertainty of the prior model itself which could be mitigated by estimating a priori model error. Unfortunately, due to the need to handle an additional large number of unknowns in a problem that already suffers from ill-posedness, model error estimation remains an unresolved challenge. In this paper, we address this issue by modeling and estimating the a priori model error in a low dimensional space using a novel sparse prior based on the variational approximation of L0 norm. This prior is used in a posterior regularized Bayesian formulation to quantify the error in a priori EP model during the reconstruction of transmural action potential from ECG data. Through synthetic and real-data experiments, we demonstrate the ability of the presented method to timely capture a priori model error and thus to improve reconstruction accuracy compared to approaches without model error correction.

Discriminant analysis was performed on ECG data simultaneously recorded using 120 leads for 250 normal subjects and 214 patients with left ventricular hypertrophy (LVH). Instantaneous voltages on time-normalized P, PR, QRS, and ST-T waveforms as well as the durations of these waveforms were used as features. A total of six features from five torso sites accounted for a specificity of 97% and a sensitivity of 94%. The single most potent discriminator was the duration of the P wave; voltages were measured in mid and late P, in mid QRS, and slightly before the peak of the T wave. The optimal sites for LVH diagnosis were in general outside the conventional ECG lead locations. In comparison, multivariate analysis on the standard 12 leads correctly classified 86% of pure LVH patients and 83% of complicated LVH cases at specificity rates of 94% and 93%, respectively. >

There is a continuing need of methods to identify subgroups of patients at high risk of ventricular arrhythmias, in particular after myocardial infarction (MI).We performed a singular value decomposition of repolarization potentials in individual recordings in 134 healthy males, in 203 males with old MI and without documented sustained ventricular tachycardia (VT) and in 104 MI males with documented VT. We considered the absolute correlation coefficient between the first orthogonal component, constructed by matrix multiplication of the first left and right singular vectors and the QRS integral (RT1) and a similar index for the second component (RT2).Abnormally high (more than two standard deviations above the mean) value of the RT1 had a 89% specificity for VT in MI patients. Abnormally low RT2 had specificity of 87%. Both indices combined had a 97% specificity. However, sensitivity of the combined indices was only 13%.Abnormalities in the correlation of orthogonal components of repolarization with depolarization are highly specific for a small group of patients with old myocardial infarction at high risk of ventricular tachycardia.

In view of the increased risk of cardiovascular mortality associated with left ventricular (LV) hypertrophy, early recognition and quantitation of LV hypertrophy are important clinical goals. The standard 12-lead electrocardiogram is the easiest and most widely used noninvasive method for the diagnosis of LV hypertrophy; unfortunately, the diagnostic accuracy of commonly used electrocardiographic criteria remains unsatisfactory. Body surface potential maps contain diagnostic information not present in conventional lead systems. The present investigation combines the increased information content of surface maps with the power of multivariate statistical techniques in order to identify practical subsets of electrocardiographic leads that would allow improved diagnosis of LV hypertrophy. Discriminant analysis was performed on 120-lead data simultaneously recorded in 250 normal subjects and 214 patients with LV hypertrophy using as features instantaneous voltages on time-normalized P, PR, QRS and ST-T waveforms as well as the duration of these waveforms. Leads and features for optimal separation of 173 normal subjects aged greater than or equal to 30 years from 122 patients with pure LV hypertrophy were selected. A total of 6 features from 5 torso sites accounted for a specificity of 97% and a sensitivity of 94%. The single most potent discriminator was the duration of the P wave; voltages were measured in mid and late P on leads located in the lower left parasternal area, the left precordial region and the upper right back, in mid-QRS on a lead positioned 10 cm below V1 and slightly before the peak of the T wave on a lead in the lower left flank.(ABSTRACT TRUNCATED AT 250 WORDS)

Most studies on diagnostic classification of the electrocardiogram (ECG) deal with only two diagnostic categories at once, for example normals versus anterior myocardial infarction, normals versus inferior myocardial infarction, or normal versus left ventricular hypertrophy. ~-5 Such procedures can be helpful for selecting optimal measurements and providing better insight in diagnostic criteria, and in some important applications, such as monitoring patients with acute myocardial infarction, a normal versus myocardial infarction setting may suffice. Bigroup comparisons, however, are not realistic in clinical practice, where often more than two diagnostic entities must be considered. 6 The multigroup approach was first developed by Pipberger et al. 7 Other investigators have since then applied multivariate statistical techniques for classification of both the vectorcardiogram (VCG) and the ECG in a multigroup setting. 8-1~ The set of measurements to be entered into the multigroup classification scheme generally results from pooling the most disciminating variables selected from each pairwise comparison. The total accuracy--percentage of correctly classified subjects--has varied from 63% to 87%, depending on the lead system, the number of groups, and whether prior probabilities were used. Most reports have placed greater emphasis on the

There is little doubt today that body surface potential maps (BSPM) contain diagnostic information not present in conventional lead systems (1, 2, 3). However, the need of specialized hardware and the large amounts of data to process and to analyze have hampered their widespread use.

In view of the increasing interest in quantifying and modifying the size of myocardial infarction (MI), it is important to look for clinically practical subsets of electrocardiographic leads that allow the earliest and most accurate diagnosis of the presence and electrocardiographic type of MI. A practical approach is described, taking advantage of the increased information content of body surface potential maps over standard electrocardiographic techniques for facilitating clinical use of body surface potential maps for such a purpose. Multivariate analysis was performed on 120-lead electrocardiographic data, simultaneously recorded in 236 normal subjects, 114 patients with anterior MI and 144 patients with inferior MI, using as features instantaneous voltages on time-normalized QRS and ST-T waveforms. Leads and features for optimal separation of normal subjects from, respectively, anterior MI and inferior MI patients were selected. Features measured on leads originating from the upper left precordial area, lower midthoracic region and the back correctly identified 97% of anterior MI patients, with a specificity of 95%; in patients with inferior MI, features obtained from leads located in the lower left back, left leg, right subclavicular area, upper dorsal region and lower right chest correctly classified 94% of the group, with specificity kept at 95%. Most features were measured in early and mid-QRS, although very potent discriminators were found in the late portion of the T wave. Repeatability of the results was investigated by separating the study population in training and testing sets; no deterioration was observed when the discriminant functions computed on the training sets were run on the testing sets. In comparison, at the same level of specificity and with the same number of features (n = 6), the standard 12-lead electrocardiogram correctly diagnosed 89% of anterior MI and 85% of inferior MI patients. Thus, diagnosis of anterior and inferior MI can be substantially improved by appropriate selection of electrocardiographic leads and features.

Body surface potential maps were recorded from 120 electrode sites in 236 normal subjects and 258 patients with initial evidence of either anterior myocardial infarction (MI) or inferior MI to identify characteristic map patterns in both groups. After time normalization, averaged map distributions were displayed at 18 equal time intervals during both QRS and ST-T waveforms from the normal, anterior MI and inferior MI groups. At each time instant, the 120-point averaged normal map was subtracted in turn from the corresponding anterior and inferior MI maps; the resulting differences at each electrode site were divided by the pooled standard deviation and the obtained values (discriminant indexes), plotted as contour lines with 1 standard deviation increments, producing discriminant maps for each bi-group comparison. The most consistent discriminant patterns in 114 patients with anterior MI were observed in early QRS in the upper left anterior chest where abnormal negative voltages reflected loss of electric potentials while reciprocal changes were noticed in the lower back; by mid-QRS, both distributions had moved jointly and vertically, the former in the lower torso on the midsternal line, the latter in the upper back. In 144 patients with inferior MI, abnormal positive distributions were observed in early QRS in the upper back, followed later by excessive negative voltages in the inferior right anterior chest; at mid-QRS, both distributions had migrated horizontally, the former proceeding toward the upper anterior torso, the latter to the lower left dorsal area.(ABSTRACT TRUNCATED AT 250 WORDS)