Separation of atrial and ventricular components of body surface potentials in atrial fibrillation using principal component analysis: A computer modelling study

We present a computer model which is used to assess Principal Component Analysis (PCA) as an algorithm for separating atrial and ventricular activity from body surface potentials (BSP). The model is based on separate dipoles for atrial and ventricular activity, where the amplitude and orientation of the dipoles has been extracted from real data taken from 12-lead ECGs. The model simulates BSPs at over 300 sites on a cylindrical torso. Principal components (PC) of the BSPs were calculated for models with i) atrial dipole only, ii) ventricular dipole only and iii) simultaneous atrial and ventricular dipoles. Maps of the individual contributions to the PCs from the BSPs were produced. The atrial and ventricular activities produced distinct maps for the individual dipoles. There were distinct regions on the torso corresponding to atrial and ventricular activities for simultaneous atrial and ventricular dipoles.