Visual transform applications for estimating the spatial QRS–T angle from the conventional 12-lead ECG: Kors is still most Frank
Introduction
The spatial QRS–T angle is a vectorcardiographic (VCG) parameter with notable diagnostic [1], [2], [3], [4], [5] and prognostic [2], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15] utility. For example in a recent study by Borleffs et al. [6] of 412 patients who had coronary ischemia as well as implanted cardiac defibrillators (ICD's) and ejection fractions ≤ 40%, a spatial angle greater than 100 degrees was associated with a hazard ratio for appropriate ICD discharge of 7.3. In addition to assisting with risk stratification for cardiac events, the angle is also useful for evaluation of incident coronary heart disease [2], [4], [12], heart failure [4], [11], [13], and the efficacy of therapy for adult hypertension [1] and diabetes mellitus [5]. Most recently the spatial QRS–T angle has also been shown to be more useful than any parameter from the conventional scalar 12-lead ECG for identifying both adult and pediatric hypertrophic cardiomyopathy [3], [16]. As noted in previous publications, of the various methods used for deriving the Frank XYZ leads from conventional 12-lead ECG recordings, and then in turn for estimating secondary parameters such as the spatial QRS–T angle, the regression method described by Kors et al. [17] has thus far tended to have the best overall performance [18], [19].
At the present time, arguably the main practical impediment to day-to-day clinical use of the 12-lead ECG-derived spatial QRS–T angle is the fact that its expeditious derivation requires dedicated software code that most ECG manufacturers have not yet implemented. However given the important diagnostic and prognostic utility of the angle, and in the interim until more programs containing such code become available, it would be helpful to provide physicians the means to estimate, in a purely visual but still reasonably precise fashion, the value of the angle from any scalar 12-lead ECG. Also, if the value of the angle were calculable with reasonable precision from any 12-lead ECG, including from previous 12-lead ECG printouts that are no longer electronically available, then a wealth of additional information might also become available for further scrutiny of the utility of the angle from retrospective paper-based ECG studies. While derivation of the so-called spatial “mean” QRS–T angle requires the ability to fit the entire spatial QRS and T-loops into a complex function (see Appendix), thus making visual estimates of that angle very difficult to perform from strictly scalar 12-lead ECGs, the so-called spatial “peaks” QRS–T angle [18] on the other hand requires only the absolute (“peak”) voltage values from the QRS and T waves (“loops”) as inputs. These voltages can in turn be visually estimated from the scalar 12-lead ECG and their values entered into pre-existing derivation formulae (Appendix) to calculate the “peaks” angle. Once the maximum voltage values are visually estimated from the scalar 12-lead, they can also be easily entered into simple preformatted calculators. For example, online calculators or a spreadsheet on a clinician's personal cell phone or tablet, might be used to in turn derive the “peaks” angle.
The primary goal of this study was to explore whether a reasonably precise method could be derived for rapid visual estimations of the spatial peaks QRS–T angle from strictly conventional 12-lead ECG tracings. A secondary goal was to determine whether the precisions of the best method(s) for visually estimating the peaks angle from scalar 12-lead ECGs approach those that occur when the same angle is quantified purely mathematically (automatically and non-visually) from the underlying digital data using advanced ECG software.
Section snippets
Data collection
All data were obtained from a publicly available source, the Physikalisch-Technische Bundesanstalt (PTB) Diagnostic ECG Database available at: http://www.physionet.org/physiobank/database/ptbdb/ [20]. This was the same database used for the Physionet/Computers in Cardiology Challenge 2006 [21]. The PTB ECG data were collected in the 1990s by Dr Michael Oeff et al. at the Department of Cardiology of University Clinic Benjamin Franklin in Berlin, Germany. These investigators used a non-commercial
Results
Pearson correlation coefficients for the variability of the visual estimates of the spatial peaks QRS–T angle for the Kors' regression-related and Kors' quasi-orthogonal methods were 0.98 and 0.97 (intra-observer variability), and 0.95 and 0.98 (inter-observer variability), respectively. Associated Bland–Altman 95% confidence intervals were relatively narrow with lower to upper limits of − 15.6 to + 13.7 (intra-observer) and − 20.9 to + 24.1 (inter-observer), respectively. Table 1 shows the mean ± SD
Discussion
The most important finding from this study is that spatial peak QRS–T angle results derived from a purely visual application of Kors' regression-related transform can, with reasonable accuracy and precision, estimate simultaneous, automatically derived results from the true Frank leads. Correlation coefficients are typically the 0.78–0.90 range, but with exact values depending on the specific clinical group being studied. Not unexpectedly, correlation coefficients against the gold standard are
Conclusion
It is possible to visually estimate, with reasonable precision and accuracy, the spatial peaks QRS–T angle from any standard scalar 12-lead ECG tracing. While results from a visual application of Kors' full eight-channel regression transform most closely estimate simultaneous results from the true Frank leads (and even more closely estimate simultaneous results from a fully automated implementation of Kors' regression transform), Kors' et al.'s other transform—i.e., the quasi-orthogonal
Acknowledgments
The authors thank Physionet and Erasmus MC for provision, respectively, of the online database and the SCP ECG viewer utilized in this study.
References (28)
- et al.
Spatial QRS–T angle as a risk indicator of cardiac death in an elderly population
J Electrocardiol
(2003) - et al.
Comparison of mortality risk for electrocardiographic abnormalities in men and women with and without coronary heart disease (from the Cardiovascular Health Study)
Am J Cardiol
(2006) - et al.
Electrocardiographic predictors of new-onset heart failure in men and in women free of coronary heart disease (from the Atherosclerosis in Communities [ARIC] Study)
Am J Cardiol
(2007) - et al.
Electrocardiographic predictors of cardiovascular outcome in women: the National Heart, Lung, and Blood Institute-sponsored Women's Ischemia Syndrome Evaluation (WISE) study
J Am Coll Cardiol
(2005) - et al.
Spatial QRS–T angle predicts cardiac death in a clinical population
Heart Rhythm
(2005) - et al.
When deriving the spatial QRS–T angle from the 12-lead electrocardiogram, which transform is more Frank: regression or inverse Dower?
J Electrocardiol
(2010) - et al.
The spatial QRS–T angle in the Frank vectorcardiogram: accuracy of estimates derived from the 12-lead electrocardiogram
J Electrocardiol
(2010) - et al.
Vectorcardiogram synthesized from a 12-lead ECG: superiority of the inverse Dower matrix
J Electrocardiol
(1988) - et al.
Linear affine transformations between 3-lead (Frank XYZ leads) vectorcardiogram and 12-lead electrocardiogram signals
J Electrocardiol
(2009) - et al.
Vectorcardiographic and electrocardiographic criteria to distinguish new and old left bundle branch block
Heart Rhythm
(2010)