Improved diagnostic value of combined time and frequency domain analysis of the signal-averaged electrocardiogram after myocardial infarction
Rafael Vázquez, MD*,
Edward B. Caref, PhD ,
Francisco Torres, MD*,
Margarita Reina, RN*,
Aurora Espina, MD* and
Nabil El-Sherif, MD
* Cardiology Unit of Valme University Hospital, Seville, Spain
Cardiology Divisions of the State University of New York Health Science and Veterans Administration Medical Centers, Brooklyn, New York, USA
View larger version (39K):
[in a new window]
|
Figure 1 Receiver operating characteristic curves with time domain (left panel) and spectral turbulence analysis (STA, right panel) parameters. The arrows point to the values selected as optimal cutoffs for each parameter. Note that the best balance between sensitivity and specificity is obtained with RMS40 (among time domain parameters) and spectral entropy (STA parameters). The opposite occurred with QRSD and ISCSD, which had a lower specificity than the other parameters over a wide range of sensitivity levels. Entropy = spectral entropy; ISCM = interslice correlation mean; ISCSD = interslice correlation standard deviation; LAS40 = duration of low amplitude signals <40 µV; LSCR = low slice correlation ratio; QRSD = total filtered QRS duration; RMS40 = root mean square voltage of the last 40 ms of the filtered QRS.
|
|
View larger version (36K):
[in a new window]
|
Figure 2 Receiver operating characteristic curve with the combined model of time domain and spectral turbulence analysis (TD+STA). The arrow points to the best sensitivity–specificity balance, obtained when the mathematical model reaches the value of 1.4. RBBB = right bundle branch block. Other abbreviations as in Figure 1.
|
|
View larger version (38K):
[in a new window]
|
Figure 3 Receiver operating characteristic (ROC) curves of the combined time domain and spectral turbulence analysis (TD+STA) model, compared with isolated time domain (TD, left panel) or spectral turbulence analysis (STA, right panel) techniques. In each panel the thick solid line represents the curve obtained with the combined TD+STA model. For each isolated technique, the thin solid line represents the curves obtained with its conventional cutoffs, and the dotted line represents the corresponding curve, using its "optimal" cutoffs (shown in Fig. 1). Both lines are quite close, but the optimized curves have a slightly higher specificity up to sensitivity levels of 90%; however, the differences between those ROC curves were not statistically significant. Note that in each panel, the best sensitivity–specificity balance is obtained with the combined TD+STA model, with a significantly greater area than any other ROC curve.
|
|
|
