Volume of interest-based fourier transform method for calculation of static magnetic field maps from susceptibility distributions

Purpose Calculation of static magnetic field (B₀) inhomogeneity maps of high-resolution susceptibility models by means of convolution with dipole kernels often encounters limitations in computer memory (RAM) for large input data matrices. In many applications, only a small portion of the full volume of the computer model is a volume of interest (VOI) or a susceptibility perturbation source. This work presents a VOI-based method to significantly reduce the computer memory usage for such applications. Theory and Methods The VOI-based method is presented and compared with the conventional method for calculation of the B₀ field in the brain and heart of a human body model in terms of calculation speed, memory requirement, and calculation error relative to the full model results. Results Use of the VOI-based method significantly reduced memory usage in the human body model calculations over the conventional method without loss of accuracy and with comparable calculation speed. Conclusion The proposed method can be valuable for rapid calculation of B₀ distributions on standard computer hardware for applications such as subject-specific B₀ field calculations derived from anatomic scans.