AVERAGE INTENSITY RECONSTRUCTION AND WIENER RECONSTRUCTION OF BIOELECTRIC CURRENT DISTRIBUTION BASED ON ITS ESTIMATED COVARIANCE-MATRIX

This paper proposes two methods for reconstructing current distributio ns from biomagnetic measurements. Both of these methods are based on e stimating the source-current covariance matrix from the measured-data covariance matrix. One method is the reconstruction of average current intensity distributions. This method first estimates the source-curre nt covariance matrix and, using its diagonal terms, it reconstructs cu rrent intensity distributions averaged over a certain time. Although t he method does not reconstruct the orientation of each current element at each time instant, it can retrieve information regarding the curre nt time-averaged intensity at each voxel location using extremely low SNR data. The second method is Wiener reconstruction using the estimat ed source-current covariance matrix. Unlike the first method, this Wie ner reconstruction can provide a current distribution with its orienta tion at each time instant. Computer simulation shows that the Wiener m ethod is less affected by the choice of the regularization parameter, resulting in a method that is more effective than the conventional min imum-norm method when the SNR of the measurement is low.