Motivation: Ex vivo quantitative susceptibility mapping is motivated by artifacts in phantom or post-mortem reconstructions serving as gold standards for new methods. The nonlocal dipole deconvolution propagates artifacts and introduces error into these reconstructions, impeding the validation process. Goal(s): The presented pipeline is an open-source tool to mitigate artifacts in susceptibility reconstruction and subsequent processing such as source separation and myelin mapping. Approach: A numerical phantom and ex vivo brain tissues are reconstructed with the proposed method. Results: Artifact reduction is observed and high structural similarity is observed. Impact: The presented method mitigates ex vivo artifacts to increase confidence in validation. High structural similarity is observed from whole region of interest filtering and homogeneity regularization. This open-source pipeline improves source separation and potentially myelin mapping, and oxygen extraction fraction.
Roberts et al. (Tue,) studied this question.