| Determining cardiac electrical activity from the body surface potentials is known as the inverse problem in electrocardiography. Epicardial potentials have been shown to reflect accurately the underlying electrical activity. A non-invasive method of recovering epicardial potentials from body surface potentials would permit the detection and localization of abnormal electrical activity prior to surgery, may shorten the time necessary to map the heart potentials intraoperatively, and may eliminate the need for detailed invasive surgical mapping.; Inverse studies were performed with two models. The first was a tank, molded in a human shape, containing a beating dog's heart in normal sinus rhythm. The second model was a system of eccentric spheres, incorporating all of the important geometry and conductivity parameters of the human torso. Regularizors in the Tikhonov family were tested on both models to determine the ability of the inverse procedure to resolve details of the epicardial potential distribution.; The results were as follows: (1) the inverse procedure was able to identify and localize different areas of ventricular depolarization at different times during the QRS, with typical errors of +/{dollar}-1{dollar} cm in the position of the recovered features; (2) in the presence of a 1 cm error in estimating the heart position, the features of the potential distribution were slightly distorted and shifted, but otherwise recognizable; (3) in situations of small geometrical error (less than 1 cm), inversion accuracy could be restored by using the Twomey or a constrained inversion, incorporating limited but accurate a priori information about the epicardial potential solution; (4) a subset of electrodes, consisting of 120 leads, densely distributed on the percordial area, the left side, and left back, produced a fairly accurate inversion; (5) smoothing the body surface potential data improved the accuracy for closely spaced lead distributions, but not for sparsely distributed leads with this experiment; (6) use of a stylized torso, constructed from eight simple measurements, resulted in only slight increases in error of the inversion.; These studies indicate that the inverse procedure could be adapted successfully to an experimental or clinical setting. The model studies with the eccentric spheres and with the realistic geometry torso tank provide guidelines for inverse reconstruction in the in vivo situation. |
