Heart-surface potentials estimated from 12-lead electrocardiograms

We investigated whether the ischemic region due to coronary-artery occlusion can be visualized on the heart-surface display by using as input data just ST measurements from the 12-lead ECG. A boundary-element torso model was used to estimate heart-surface potentials from body-surface potential maps (BSPMs) obtained from 12-lead ECG via transformation developed from a design set (n = 892) of 120-lead ECGs by a least-square solution to the linear regression problem. The test set consisted of 120-lead ECGs acquired for 45 patients during ischemia induced by balloon-inflation angioplasty; subgroups of equal size consisted of patients whose LAD, LCx, and RCA were occluded, respectively. BSPMs at J point of each patient were predicted from the 12-lead ECG and similarity of original and predicted maps was assessed by a similarity coefficient SC (0-100%). By inverse solution, heart-surface potentials were then estimated from original and predicted BSPMs and visualized on a bull's-eye display. Reconstitution of known BSPMs from the 12-lead ECG achieved an overall SC 92.46 ± 6.95% (mean ± SD); for subgroups, SC was: 89.79 ± 9.95% for LAD, 93.89 ± 2.86% for LCx, and 93.70 ± 5.77% for RCA group. Estimated heart-surface potential distributions featured an area of positive potentials corresponding, in general, to the underperfused territory caused by the occlusion. Encouraging results can be attributed to the strongly dipolar character of BSPMs caused by injury current. This approach shows promise for ischemia detection and quantitation.