Visual transform applications for estimating the spatial QRS-T angle from the conventional 12-lead ECG: Kors is still most Frank.

BACKGROUND: The 12-lead ECG-derived spatial QRS-T angle has prognostic and diagnostic utility, but most ECG machines currently fail to report it. The primary goal was to determine if reasonably accurate methods exist for rapid visual estimations of the spatial peaks QRS-T angle from conventional 12-lead ECG tracings. METHODS AND
RESULTS: Simultaneous 12-lead and Frank XYZ-lead recordings were obtained from a publicly available database for 100 post-myocardial infarction patients and 50 controls. ANOVA, Pearson's correlation coefficients and concordance plots were used to evaluate agreement for spatial peaks QRS-T angle results from the true Frank leads versus from several visually applied 12-to-Frank XYZ-lead transforms. The latter included Kors et al.'s regression and quasi-orthogonal, Bjerle and Arvedson's quasi-orthogonal, Dower's inverse, and Hyttinen et al.'s, Dawson et al.'s and Guillem et al.'s transforms. Spatial peaks QRS-T angles derived from the true Frank leads were not statistically significantly different from those derived from any visually applied transform. Of the visually applied transforms, the Kors' regression and Kors' quasi-orthogonal yielded the highest Pearson correlation coefficients against the gold-standard true Frank lead results [0.84 and 0.77, respectively, when individuals with bundle branch blocks were included (N=150), and 0.88 and 0.80, respectively, when individuals with bundle branch blocks were excluded (N=137)]. Bland-Altman 95% confidence intervals showed similar results, with the two Kors'-related methods also having the narrowest confidence intervals.
CONCLUSIONS: When visually applied, the Kors' regression-related and quasi-orthogonal transforms allow for reasonably precise spatial peaks QRS-T estimates and thus a potentially practical way to visually estimate spatial peaks QRS-T angles from conventional 12-lead ECGs.