Obaidah Anees Abuhashem1, Berkin Bilgic1, Elfar Adalsteinsson1, 2
1EECS, Massachusetts Institute of Technology, Cambridge, MA, United States; 2Harvard-MIT Division of Health Sciences and Technology, Cambridge, MA, United States
Quantitative Susceptibility Mapping (QSM) is used to quantify tissue magnetic susceptibility, leading to applications such as tissue contrast enhancement, venous blood oxygenation, and iron quantification. Quantification of the susceptibility distribution χ involves removal of background effects on the MRI signal phase and the solution of an ill-posed inverse problem describing the mapping from the phase to the tissue susceptibility. In this work, background removal is achieved by using the effective dipole fitting algorithm and susceptibility inversion is performed via imposing ℓ1 norm regularization on the spatial gradients of χ. As both algorithms are computationally demanding, it is crucial to increase the computational throughput and make regularized QSM a feasible and real-time methodology. Herein, the computational power Graphics Processing Cards (GPUs) is utilized to greatly accelerate the processing times, and both MATLAB and GPU libraries of the regularized QSM method are made available online for reproducibility.