Abstract
Recently, magnetic particle imaging (MPI) has gained attention as a new medical imaging diagnosis technology. In MPI, an image is reconstructed by detecting the signals from magnetic nanoparticles (MNPs) injected into the body. Since MNPs have the property of accumulating in cancer cells, MPI is expected to be applicable to early diagnosis of cancer. The fundamental method of MPI involves reconstructing the MNP distribution by detecting the odd harmonics generated from MNPs. However, this method has a problem in that image blurring and artifacts occur due to signals from MNPs outside the signal detection region. In order to resolve this problem, a reconstruction method based on solution of the inverse problem has been proposed. This method suppresses image blurring and artifacts by considering the signals from MNPs outside the target measurement region. However, this method requires huge matrix operation and thereby increasing reconstruction time. In this paper, we propose a new image reconstruction method with orthonormal basis calculated using singular value decomposition. The system function and the observed signals from the distribution of unknown MNPs, which are used for image reconstruction, are expanded using the orthonormal basis. Since the proposed method has a reduced matrix size compared with the conventional solution of the inverse problem, image reconstruction time can be reduced. By numerical simulation, we confirmed that image reconstruction with an image quality equivalent to that of the conventional solution of the inverse problem was obtained in 1/39 calculation time for a 11×11 matrix size image.
