Magnetocardiography Current Source Estimation using Multiple Spatial Filters

抄録

  In this study, cardiac current source localization was simulated using the spatial filter method. We estimated current sources using data from two measurement surfaces, i.e., the front and back of the subject, to reduce the large spatial extension in the depth direction. Three methods for creating spatial filters were developed using the exact low-resolution brain electromagnetic tomography method based on different creation procedures. The first method is a general method that has been conventionally used to create a spatial filter (64ch spatial filter) using a 64ch sensor on the front side. The second method is to create a spatial filter (128ch spatial filter) using two 64ch sensors on the front and back sides as a 128ch sensor. The third method is to create two spatial filters separately and use them as one spatial filter (2 × 64ch spatial filter). In this case, different Tikhonov regularization parameters can be determined on the front and back sides, and optimum noise immunity can be given to each. The accuracy of the estimated current sources using the three types of spatial filters was examined using simulation and MCG data. The results show that the 2 × 64 spatial filter has the highest estimation accuracy and suppresses the large spatial extension of the estimated current sources.