Heart Model And The Inverse Problem Of Ecg Simulation

During the last decade, multiple advances in the radiofrequency catheter ablation (RFCA) has allowed this technique to blossom into one of the most powerful therapy available for non-drug treatment of arrhythmia. RFCA is considered beneficial in several types of arrhythmia. It is crucial to locate the proarrhythmia ectopic pacemaker before the ablation procedure. Research has been focused on locating the proarrhythmia point non-invasively, precisely and rapidly. Although endocardiac activation catheter mapping can give precise location of the ectopic pacemaker, it has several disadvantages. The time consuming intracardiac locating procedure increases the risk of severe complications and the dose of radiation exposure. New method are required to make improvement.Our research worked on the ECG forward and inverse problem, and proposed a possible way to improve clinical endocariac catheter mapping using noninvasive ECG signals and computer heart model. The whole research is composed of three related aspects. First, a high spatial resolution whole heart model for the ECG forward problem was constructed; Then, the relationship between the location of ectopic pacemaker and body surface potential maps (BSPM) was investigated. Finally, a feasible way was proposed to facilitate ectopic pacemaker localization based on current clinical trail.The research started from heart modeling using Visible Human Project (VHP) cross-section data. Image enhancement and tissue segmentation were performed on heart region and thorax surface. A realistic geometry heart model and thorax front surface were then built with a resolution of 0.5mm×0.5mm×0.5mm. Besides geometry reconstruction, our research also quantitated electrophysiology related conceptions and configured the electrophysiological parameters all-over the heart. A propagation algorithm was applied on the model to simulate electric activities of atrium and ventricle at a time accuracy of 1ms. The simulated electric activity was then projected to body surface potentials.The ECG inverse problem was explored on the base of forward problem. Adjustments to the normal heart model were made according to the characteristics of arrhythmia. Then, ventricular arrhythmic BSPM was simulated using the adjusted heart model. We set up different ectopic pacemakers on a certain region of endocardium, and studied the relation between paced location and paced BSPM. An artificial neural network (ANN) was designed and trained to localize paced point at an partition density of 1 cm×0.5mm, and its generalization capability was tested using simulated test set with and without noise. The overall accuracy came out to be 25/31 without noise and 23/31 with signal-to-noise ratio of 2dB, overcoming the ill-posed character in solving ECG inverse problem using mathematical theories solely.At last, a method, employing noninvasive BSPM, was proposed to facilitate endocardiac mapping and accelerate ectopic pacemaker localization according to clinical catheter paced mapping procedure. ANN was used to localize ectopic point to a small region, and then, the catheter could move towards the direction derived by analyzing the quantitative relation between abnormal pacemaker point location and BSPM.Our project has built up a high resolution heart model. And the model was used in the exploration on ECG inverse problem as well as the approach of a possible clinical application in locating ectopic pacemaker during RFCA. Our investigation provides useful information and lays a sound basis for promoting further research.