Meeting Banner
Abstract #2201

Fast and accurate reconstruction for susceptibility source separation in QSM

Seyoon Ko1, Jingu Lee2, Joong-Ho Won1, and Jongho Lee2

1Department of Statistics, Seoul National University, Seoul, Republic of Korea, 2Laboratory for Imaging Science and Technology, Department of Electrical and Computer Engineering, Seoul National University, Seoul, Republic of Korea

We investigate fast and accurate reconstruction methods for susceptibility source separation (S3) in quantitative susceptibility mapping (QSM). S3 separates positive and negative susceptibility sources within a voxel utilizing signal relaxation (R2') for dipole inversion. We propose new primal-dual (PD) methods for S3 and compare them with the alternating Gauss-Newton conjugate gradient (A-GNCG). A-GNCG alters the energy functional, and furthermore its convergence is not guaranteed. In contrast, the proposed PD methods are exact and have convergence guarantees. Validation on a simulated phantom and in-vivo data shows that the PD methods converge faster with better accuracies.

This abstract and the presentation materials are available to members only; a login is required.

Join Here