2D ECG differences in Frontal vs Preferential planes in patients referred for percutaneous transluminal coronary angioplasty

BONOMINI M.P:; CORIZZO S.; LAGUNA P.; ARINI P.D.

BIOMEDICAL SIGNAL PROCESSING AND CONTROL

ELSEVIER SCI LTD

Lugar: Amsterdam; Año: 2014 vol. 11 p. 97 - 106

1746-8094

This work proposes a comparative study of a pair of electrocardiographic 2D representations: the frontalplane (FP) and a preferential plane (PP) obtained from ECG data. During depolarization and repolarization,main electrical vectors were analyzed and compared between healthy subjects and patients referred forpercutaneous transluminal coronary angioplasty (PTCA). Recordings were obtained at rest. Many patientsfrom the latter group presented normal ECGs, thus, the hypothesis to prove was that electrical axis in anyof the studied planes would effectively discriminate silent ischemia records from healthy ones. The FP wasconstructed with I and aVF leads, while the PP used the two first eigenvectors of the spatial correlationmatrix of the ECG. Although the depolarization and repolarization vectors from both groups resultednormal, those from the silent ischemia group appeared strongly biased to the left, closer to the limit ofthe normality range. This slight change originated a significant separation between health and disease inthe FP. Here, most of the parameters resulted highly informative, even those related to the depolarizationphase. The cardiac vector, integrating both depolarization and repolarization information, presentedthe highest performance (AUC = 0.89). Parameters in the PP, however, did not produce an acceptablediscrimination, except for the amplitude of the T-wave (AUC = 0.79). Additionally, the repolarizationorientation in the FP was the only marker that simultaneously discriminated three different groups ofpatients according to their occlusion sites (p < 0.0001). In conclusion: the FP offered a 2D representationgeneral enough to enable the separation of silent ischemia versus health populations while the PP didnot, due mainly to its individually optimized nature, failing to provide a unique referencial frame for allthe subjects.