Computerized method for finding the ideal patient-specific location to place an equivalent electric dipole to derive an estimation of the electrical activity of the heart.

摘要

Electrocardiology has been a powerful tool for doctors for more than 100 years in detecting heart-related problems. More importantly, cardiologists have relied on noninvasive monitoring tools for recording the electrical activity of the heart at the surface of the torso. Some advanced techniques, like the popular EASI method in which the standard 12-lead ECG is derived from 4 easy-to-place electrodes, have been developed to obtain a representation of the electrical activity at the surface of the heart from the signals measured at the surface a the torso. These techniques, however, rely on standard models that use generic transformation parameters, which correspond to an average person. The body for each patient has a different amount of fat and muscles, quality of tissue, shape of the body, etc. Moreover, the heart's position and size vary from patient to patient, producing changes in the ECG. All these factors together yield greater error when using general-population methods or parameters to obtain derive the ECG, the same way they may affect diagnosis when analyzing the ECG.; Therefore, a patient-specific position for placement of an electrical dipole to represent the activity of the heart is proposed. This technique is based on the theory that a fixed-single electric dipole can accurately represent the electrical activity of the human heart. The ideal location, or the one that best reproduces the ECG, is found from among more than 20,000 possible locations in a cube-shaped volume reference placed in the normal location of the heart in a human torso. This process is done by obtaining the characteristic parameters for each desired electrode location on the torso for each possible location of the electric dipole within the specified cube-shaped volume. Then, the signals are derived for each desired ECG electrode and compared to the measured signal. The location of the dipole that best derives all the signals as calculated by the Normalized Root-Mean-Squared error (NRMSe) per ECG signal is selected. The results show a mean NRMSe of 4.87% per signal with a standard deviation of 3.79%, considering the standard 6 precordial signals of a subject's ECG. Moreover, the mean correlation coefficient per signal was 0.9634 with a standard deviation of 0.0868. The results show that the reproducibility of the signals is improved by allowing the adaptation of patient specific parameters and that this method could be used to improve techniques that make use of forward-problem or inverse-problem techniques.