Further results utilizing the MRE principle in inverse electrocardiography

The authors explore two linearized implementations of the application of a frequency-domain method, the minimum relative entropy (MRE) reconstruction principle, to the inverse problem in electrocardiography. These methods reconstruct the power spectrum of each epicardial signal, in contrast to most methods that have been applied to this problem, which operate in the time domain of the data using regularization techniques and minimizing squared error. The least-squares methods have important weaknesses for clinical applications. Simulations and comparisons are carried out using canine epicardial data and an accurate model of the human torso taken from CT (computerized tomography) scans