It is widely accepted that the heart current source can be reduced into a current multipole. By adopting three linear inverse methods, the cardiac magnetic imaging is achieved in this article based on the current multipole model expanded to the first order terms. This magnetic imaging is realized in a reconstruction plane in the centre of human heart, where the current dipole array is employed to represent realistic cardiac current distribution. The current multipole as testing source generates magnetic fields in the measuring plane, serving as inputs of cardiac magnetic inverse problem. In the heart-torso model constructed by boundary element method, the current multipole magnetic field distribution is compared with that in the homogeneous infinite space, and also with the single current dipole magnetic field distribution. Then the minimum-norm least-squares （MNLS） method, the optimal weighted pseudoinverse method （OWPIM）, and the optimal constrained linear inverse method （OCLIM） are selected as the algorithms for inverse computation based on current multipole model innovatively, and the imaging effects of these three inverse methods are compared. Besides, two reconstructing parameters, residual and mean residual, are also discussed, and their trends under MNLS, OWPIM and OCLIM each as a function of SNR are obtained and compared.
This paper discusses the forward and inverse problem for cardiac magnetic fields and electric potentials. A torso-heart model established by boundary element method （BEM） is used for studying the distributions of cardiac magnetic fields and electric potentials. Because node-to-node and triangle-to-triangle BEM can lead to discrepant field distributions, their properties and influences are compared. Then based on constructed torso-heart model and supposed current source functional model-current dipole array, the magnetic and electric imaging by optimal constrained linear inverse method are applied at the same time. Through figure and reconstructing parameter comparison, though the magnetic current dipole array imaging possesses better reconstructing effect, however node-to-node BEM and triangleto-triangle BEM make little difference to magnetic and electric imaging.
Two cardiac functional models are constructed in this paper. One is a single current model and the other is a current multipole model. Parameters denoting the properties of these two models are calculated by a least-square fit to the measurements using a simulated annealing algorithm. The measured signals are detected at 36 observation nodes by a superconducting quantum interference device （SQUID）. By studying the trends of position, orientation and magnitude of the single current dipole model and the current multipole model in the QRS complex during one time span and comparing the reconstructed magnetocardiography （MCG） of these two cardiac models, we find that the current multipole model is a more appropriate model to represent cardiac electrophysiological activity.
