James W. Warren

Cardiac infarction injury score: an electrocardiographic coding scheme for ischemic heart disease.

A multivariate decision-theoretic electrocardiogram (ECG) classification scheme called Cardiac Infarction Injury Score (CIIS) was developed using ECGs of 387 patients with myocardial infarction (MI) and 320 subjects without infarction. The most accurate and stable classification was achieved by using a combination of eight binary (single threshold), three ternary (two thresholds), and four ECG features measured on a continuous scale. For practical visual coding of ECGs, the CIIS coding procedure uses a checklist containing 12 items measured from the conventional 12-lead ECG.The CIIS test results indicate that, in comparison with conventional ECG criteria for MI used in clinical trials, the diagnostic accuracy can be considerably improved by optimizing feature and threshold selection and by multivariate analysis. The CIIS detected MI with a sensitivity of 85% and a specificity of 95%. Using a higher severity level, a specificity of 99% was achieved, with a sensitivity of 71%. One of the primary uses of the CIIS is coding of significant worsening of the ECG with new coronary events from annually recorded ECGs in clinical trials and epidemiologic studies.

Identification of best electrocardiographic leads for diagnosing myocardial infarction by statistical analysis of body surface potential maps

This study describes a practical approach for the extraction of diagnostic information from body surface potential maps. Body surface potential map data from 361 subjects were used to identify optimal subsets of leads and features to distinguish 184 normal subjects from 177 patients with myocardial infarction (MI). Multivariate analysis was performed on 120-lead data, using as features instantaneous voltage measurements on time-normalized QRS and STT waveforms. Several areas on the map, most of which were located outside the precordial region, contained leads with important discriminant features; 2 of the 3 limb leads (aVR and aVF) also exhibited high diagnostic capability. A total of 6 features (mostly STT measurements) from 3 locations accounted for a specificity of 95% and a sensitivity of 95%; these were the right subclavicular area, the left posterior axillary region and the left leg. As a comparison, the same number of features from the standard 12-lead electrocardiogram yielded a sensitivity of 88% for a specificity of 95%. To investigate the repeatability of the results, the entire population was separated into a training set (100 normal subjects and 100 patients with MI) and a testing set (84 normal subjects and 77 patients with MI); computing a discriminant function on the training set and applying it to the testing set only moderately deteriorated the diagnostic classification. It is concluded that this approach achieves efficient information extraction from body surface potential maps for improved diagnostic classification.

Spatial Features in Body-Surface Potential Maps Can Identify Patients With a History of Sustained Ventricular Tachycardia

BACKGROUND Regional disparities of ventricular primary-repolarization properties contribute to an electrophysiological substrate for arrhythmias. Such disparities can be assessed from body-surface distributions of ECG QRST areas. Our objective was to isolate and test those features of QRST-area distributions that would be suitable for identifying patients at risk for life-threatening ventricular arrhythmias. METHODS AND RESULTS We recorded ECGs simultaneously from 120 leads during sinus rhythm for 204 patients taking no antiarrhythmic drugs: half had had sustained ventricular tachycardia (VT); the other half, a myocardial infarction but no history of VT. For each patient, we calculated the QRST area in each lead and, using Karhunen-Loeve (K-L) expansion, reduced these data to 16 coefficients (each relating to one spatial feature, an eigenvector, derived from the total set of 204 QRST-area maps). Using stepwise discriminant analysis, we selected feature subsets that best discriminated between the two groups, and we estimated by a bootstrap procedure using 1000 trials how these subsets would perform on a prospective patient population. The mean diagnostic performance of the classifier for 1000 randomly selected training sets (n = 102 in each, with both groups equally represented) increased monotonically with the number of features used for classification. The initial trend for the corresponding test sets (n = 102 in each) was the same but reversed when the number of features exceeded eight. For an optimal set of eight spatial features, the sensitivity and specificity of the classifier for detecting patients with VT in 1000 test sets were (mean +/- SD) 90.3 +/- 4.3% and 78.0 +/- 6.1%, and its positive and negative predictive accuracies were 80.7 +/- 4.2% and 89.2 +/- 4.2%, respectively. Use of QRS duration as a supplementary feature to eight K-L coefficients can, in the test sets, increase specificity to 80.9 +/- 5.4% and positive predictive accuracy to 82.8 +/- 3.9% compared with the results for the optimal number of eight K-L features alone. CONCLUSIONS Multiple body-surface ECGs contain valuable spatial features that can identify the presence of an arrhythmogenic substrate in the myocardium of patients at risk for ventricular arrhythmias. Our results compare very favorably with those achieved by any other known test, invasive or noninvasive, for arrhythmogenicity.

Identification of best electrocardiographic leads for diagnosing anterior and inferior myocardial infarction by statistical analysis of body surface potential maps

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.(ABSTRACT TRUNCATED AT 250 WORDS)

Prediction of left ventricular mass from the electrocardiogram

Multiple stepwise regression methods were used to derive electrocardiographic (ECG) models for prediction of the echocardiographic left ventricular (LV) mass index from standard 12-lead ECG measurements using data files of 203 men and 252 women. The correlation between echocardiographic and ECG estimates of LV mass index was R2 = 0.58 for men and R2 = 0.42 for women. A separate logistic regression model was derived for classification of LV hypertrophy as a dichotomized dependent variable. This classifier chose R (aVL), T (V6), and S (V1) for men and R (aVL), T (V6), and S (I) for women and produced a moderate sensitivity (53.7% for men and 63.4% for women) and specificity (94.9% for men and 92.9% for women). We conclude that the initial performance of these and other recently developed multivariate estimators of LV mass and LV hypertrophy classifiers is promising enough to subject them to further studies to evaluate their utility as risk predictors.

Electrocardiographic Predictors of Coronary Heart Disease and Sudden Cardiac Deaths in Men and Women Free From Cardiovascular Disease in the Atherosclerosis Risk in Communities Study

Background We evaluated predictors of coronary heart disease (CHD) death and sudden cardiac death (SCD) in the Atherosclerosis Risk in Communities (ARIC) study. Methods and Results The study population included 13 621 men and women 45 to 65 years of age free from manifest cardiovascular disease at entry. Hazard ratios from Cox regression with 95% confidence intervals were computed for 18 dichotomized repolarization‐related ECG variables. The average follow‐up was 14 years. Independent predictors of CHD death in men were TaVR‐ and rate‐adjusted QTend (QTea), with a 2‐fold increased risk for both, and spatial angles between mean QRS and T vectors and between Tpeak (Tp) and normal R reference vectors [θ(Rm|Tm) and θ(Tp|Tref), respectively], with a >1.5‐fold increased risk for both. In women, independent predictors of the risk of CHD death were θ(Rm|Tm), with a 2‐fold increased risk for θ(Rm|Tm), and θ(Tp|Tref), with a 1.7‐fold increased risk. Independent predictors of SCD in men were θ(Tp|Tref) and QTea, with a 2‐fold increased risk, and θ(Tinit|Tterm), with a 1.6‐fold increased risk. In women, θ(Tinit|Tterm) was an independent predictor of SCD, with a >3‐fold increased risk, and θ(Rm|Tm) and TV1 were >2‐fold for both. Conclusions θ(Rm|Tm) and θ(Tp|Tref), reflecting different aspects of ventricular repolarization, were independent predictors of CHD death and SCD, and TaVR and TV1 were also independent predictors. The risk levels for independent predictors for both CHD death and SCD were stronger in women than in men, and QTea was a significant predictor in men but not in women.

Effective extraction of diagnostic ECG waveform information using orthonormal basis functions derived from body surface potential maps

A common basis of orthogonal waveform functions was derived from 128 lead body surface potential maps of 405 subjects. Twelve such orthogonal functions or frames were adequate for reconstruction of original ECGs from the beginning of QRS to the end of T. A larger number of frames (18) was required when basis functions were derived separately for QRS (10) and ST-T segments (8). Diagnostic information content of the coefficients of the orthogonal basis functions was evaluated in comparison with Minnesota Code criteria for myocardial infarction and with a more advanced multivariate ECG analysis program (Pipberger Program). This was done by deriving a linear discriminant function for separating normals from ECGs of patients with myocardial infarction and testing the discriminant in a different test population of infarcts and normals. The diagnostic accuracy of orthogonal basis functions was as good as that of Pipbergers program and considerably better than that of the Minnesota Code. The classification method described is insensitive to noise and errors in detecting QRS and T wave onsets and offsets or in selecting proper baseline for amplitude measurements. The robustness and enhanced classification stability with respect to noise and minor wave detection errors is a potential advantage particularly in serial ECG comparison.

Waveform Vector Analysis of Orthogonal Electrocardiograms: Quantification and Data Reduction*

Summary Quantification of vectorcardiogram (VCG) analysis by waveform vector analysis (WVA) provides a more stable data base for computer analysis of P, ST and T vectors than does polar-cardiography. Furthermore, compared with data bases composed of Cartesian or polar coordinates of ‘instantaneous’ vectors, it reduces the data and also reduces noise by smoothing. This method is applicable to electrocardiogram (ECG) analysis also. Modified Chebyshev polynomials up to the quaternary term are used as waveform functions for signal representation of scalar ECG, and corresponding waveform coefficients quantify the mean value, gradient, degree of concavity or convexity, biphasic or sigmoid shape, and triphasic or quarternary features. Applied to a vector function of time, the waveform coefficients comprise components of the five vectors—mean (MV), gradient (GV), convex (CV), sigmoid (SV), and quaternary vectors (QV). A waveform power index is derived to express the contribution of each component to overall variation. P, ST, and T vectors of 1,984 rest and exercise VCG of 992 male subjects were analyzed. Representation of the vector function was considered adequate when the approximation error was μ V (root mean square; RMS) for each scalar component. In 95% of the records, two waveform vectors adequately reproduced the ST vectors, four were needed for P vectors, and five for T vectors.

Comparison of epicardial potential maps derived from the 12-lead electrocardiograms with scintigraphic images during controlled myocardial ischemia ☆ ☆☆

Our aim was to cross-validate electrocardiographic (ECG) and scintigraphic imaging of acute myocardial ischemia. The former method was based on inverse calculation of heart-surface potentials from the body-surface ECGs, and the latter, on a single photon emission computed tomography (SPECT). A boundary-element torso model with 352 body-surface and 202 heart-surface nodes was used to perform the ECG inverse solution. Potentials at 352 body-surface nodes were calculated from those acquired at 12-lead ECG measurement sites using regression coefficients developed from a design set (n = 892) of body-surface potential mapping (BSPM) data. The test set (n = 18) consisted of BSPM data from patients who underwent a balloon-inflation angioplasty of either the left anterior descending coronary artery (LAD) (n = 7), left circumflex coronary artery (LCx) (n = 2), or the right coronary artery (RCA) (n = 9). Body-surface potential mapping distributions at J point for 352 nodes were estimated from the 12-lead ECG, and an agreement with those estimated from 120 leads was assessed by a correlation coefficient (CC) (in percent). These estimates yielded very similar BSPM distributions, with a CC of 91.0% ± 8.1% (mean ± SD) for the entire test set and 94.1% ± 1.4%, 96.7% ± 0.8%, and 87.4% ± 10.3% for LAD, LCx, and RCA subgroups, respectively. Corresponding heart-surface potential distributions obtained by inverse solution correlated with a lower CC of 69.3% ± 18.0% overall and 73.7% ± 10.8%, 84.7% ± 1.1%, and 62.6% ± 21.8%, respectively, for subgroups. Bulls-eye displays of heart-surface potentials calculated from estimated BSPM distributions had an area of positive potentials that qualitatively corresponded, in general, with the underperfused territory suggested by SPECT images. For the LAD and LCx groups, all 9 ECG-derived bulls-eye images indicated the expected territory; for the RCA group, 6 of 9 ECG-derived images were as expected; 2 of 3 misclassified cases had very small ECG changes in response to coronary-artery occlusion, and their SPECT images showed indiscernible patterns. In conclusion, our findings demonstrate that noninvasive ECG imaging based on just the 12-lead ECG might provide useful estimates of the regions of myocardial ischemia that agree with those provided by scintigraphic techniques.

Discrimination of ST deviation caused by acute coronary occlusion from normal variants and other abnormal conditions, using computed electrocardiographic imaging based on 12-lead ECG.

BACKGROUND Many graphical methods for displaying ST-segment deviation in the ECG have been tried for enhancing decision-making in patients with suspected acute coronary syndromes. Computed electrocardiographic imaging (CEI), based on a mathematical inverse solution, has been recently applied to transform ST-J point measurements made in conventional 12-lead ECG into a display of epicardial potentials in bulls-eye format. The purpose of this study is to assess utility of CEI in the clinical setting. METHODS In 99 patients with stable coronary disease, 12-lead ECGs were recorded during elective percutaneous coronary intervention (PCI), first before balloon-catheter insertion and then when an intracoronary balloon blocked blood supply to a region of myocardium for more than 4minutes (typically 5minutes). Four groups of patients were additionally studied, namely those with preexcitation, pericarditis, early repolarization syndrome (ERS), and left ventricular hypertrophy (LVH) with strain. Comparisons between performances of published criteria for ST-elevation myocardial infarction (STEMI) and quantitative as well as visual assessment of CEI images were based on sensitivities and specificities. RESULTS Visual assessment of CEI outperformed STEMI criteria. This was especially evident for the capability of detecting LCx occlusion with sensitivities for STEMI criteria=35% and for visual assessment of CEI by 2 physicians=71%, i. e. twice as many patients were correctly identified by CEI. False positive rates for CEI were low in patients with LVH with strain as well as with preexcitation for both methods. For pericarditis and ERS, visual as well as quantitative assessment of CEI performed better than STEMI criteria. CONCLUSION Visual assessment of CEI is a promising method for increasing the accuracy of ECG-based triage to PCI or conservative care.