Image reconstruction method based on orthonormal basis of observation signal by singular value decomposition for magnetic particle imaging

The inverse-matrix solution based on a least-squares method is used as a general image reconstruction method in magnetic particle imaging. However, it is necessary to improve the image quality further because this method tends to suffer from the effects of noise. Therefore, we have proposed a different reconstruction method based on the correlation information between the system function and the observed signal. In this method, however, image blurring appears theoretically because the differences between the system functions corresponding to magnetic nanoparticles (MNPs) arranged at various positions are small when the applied gradient field strength is comparatively weak. To overcome these problems, we propose a new reconstruction method using the orthonormal basis of the observed signal itself obtained by singular value decomposition (SVD). The particle distribution is estimated on the basis of the differences between the singular value matrices of the observed signal and the system function, which are calculated with two singular vectors (orthonormal basis) of the observed signal. Evaluating the correlation with these singular value matrices is expected to reduce the effect of noise and improve the image resolution.